diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,53526 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 8916,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7313432835820895e-07,
+      "loss": 2.1601,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.462686567164179e-07,
+      "loss": 2.0771,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.119402985074627e-06,
+      "loss": 2.106,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4925373134328358e-06,
+      "loss": 2.2354,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8656716417910446e-06,
+      "loss": 2.1278,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.238805970149254e-06,
+      "loss": 2.3031,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6119402985074627e-06,
+      "loss": 2.2564,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9850746268656716e-06,
+      "loss": 2.0494,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.358208955223881e-06,
+      "loss": 1.9589,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7313432835820893e-06,
+      "loss": 1.8559,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1044776119402985e-06,
+      "loss": 1.7524,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.477611940298508e-06,
+      "loss": 1.8363,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.850746268656717e-06,
+      "loss": 1.8006,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2238805970149255e-06,
+      "loss": 1.9885,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.597014925373135e-06,
+      "loss": 1.6376,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.970149253731343e-06,
+      "loss": 1.6889,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.343283582089552e-06,
+      "loss": 1.6244,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.716417910447762e-06,
+      "loss": 1.5703,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.08955223880597e-06,
+      "loss": 1.7634,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4626865671641785e-06,
+      "loss": 1.6886,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.835820895522389e-06,
+      "loss": 1.6533,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.208955223880597e-06,
+      "loss": 1.6498,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.582089552238805e-06,
+      "loss": 1.5866,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.955223880597016e-06,
+      "loss": 1.4679,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.328358208955226e-06,
+      "loss": 1.5161,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.701492537313434e-06,
+      "loss": 1.2735,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0074626865671643e-05,
+      "loss": 1.4608,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0447761194029851e-05,
+      "loss": 1.5189,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.082089552238806e-05,
+      "loss": 1.3719,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.119402985074627e-05,
+      "loss": 1.4454,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1567164179104478e-05,
+      "loss": 1.4664,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1940298507462686e-05,
+      "loss": 1.552,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2313432835820896e-05,
+      "loss": 1.3055,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2686567164179105e-05,
+      "loss": 1.3774,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3059701492537313e-05,
+      "loss": 1.427,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3432835820895523e-05,
+      "loss": 1.3549,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3805970149253733e-05,
+      "loss": 1.4365,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.417910447761194e-05,
+      "loss": 1.4266,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.455223880597015e-05,
+      "loss": 1.3831,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4925373134328357e-05,
+      "loss": 1.3579,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.529850746268657e-05,
+      "loss": 1.4464,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5671641791044777e-05,
+      "loss": 1.2574,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6044776119402986e-05,
+      "loss": 1.3159,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6417910447761194e-05,
+      "loss": 1.2727,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6791044776119406e-05,
+      "loss": 1.2907,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.716417910447761e-05,
+      "loss": 1.2869,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7537313432835823e-05,
+      "loss": 1.4262,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.791044776119403e-05,
+      "loss": 1.3589,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.828358208955224e-05,
+      "loss": 1.3101,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.865671641791045e-05,
+      "loss": 1.2011,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9029850746268656e-05,
+      "loss": 1.2266,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9402985074626868e-05,
+      "loss": 1.2879,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9776119402985073e-05,
+      "loss": 1.2503,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0149253731343285e-05,
+      "loss": 1.2282,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0522388059701493e-05,
+      "loss": 1.2232,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0895522388059702e-05,
+      "loss": 1.1915,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.126865671641791e-05,
+      "loss": 1.2279,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.164179104477612e-05,
+      "loss": 1.2148,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.201492537313433e-05,
+      "loss": 1.2356,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.238805970149254e-05,
+      "loss": 1.2555,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2761194029850747e-05,
+      "loss": 1.2252,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3134328358208956e-05,
+      "loss": 1.1469,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3507462686567168e-05,
+      "loss": 1.176,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3880597014925373e-05,
+      "loss": 1.2508,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4253731343283584e-05,
+      "loss": 1.1577,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4626865671641793e-05,
+      "loss": 1.1356,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 1.1143,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.537313432835821e-05,
+      "loss": 1.1832,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.574626865671642e-05,
+      "loss": 1.1436,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6119402985074626e-05,
+      "loss": 1.1247,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6492537313432835e-05,
+      "loss": 1.206,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6865671641791047e-05,
+      "loss": 1.1339,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7238805970149255e-05,
+      "loss": 1.1284,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7611940298507467e-05,
+      "loss": 1.1494,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7985074626865672e-05,
+      "loss": 1.2107,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.835820895522388e-05,
+      "loss": 1.1383,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8731343283582092e-05,
+      "loss": 1.1491,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.91044776119403e-05,
+      "loss": 1.1793,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9477611940298512e-05,
+      "loss": 1.169,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9850746268656714e-05,
+      "loss": 1.1381,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0223880597014926e-05,
+      "loss": 1.1874,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.059701492537314e-05,
+      "loss": 1.1787,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0970149253731346e-05,
+      "loss": 1.0741,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1343283582089554e-05,
+      "loss": 1.169,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.171641791044776e-05,
+      "loss": 1.1647,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.208955223880597e-05,
+      "loss": 1.1443,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.246268656716418e-05,
+      "loss": 1.1101,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.283582089552239e-05,
+      "loss": 1.0696,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.32089552238806e-05,
+      "loss": 1.1006,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.358208955223881e-05,
+      "loss": 1.1298,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.395522388059701e-05,
+      "loss": 1.1841,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.432835820895522e-05,
+      "loss": 1.1044,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.470149253731344e-05,
+      "loss": 1.123,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5074626865671645e-05,
+      "loss": 1.0933,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5447761194029854e-05,
+      "loss": 1.0573,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.582089552238806e-05,
+      "loss": 1.0801,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.619402985074627e-05,
+      "loss": 1.1203,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.656716417910448e-05,
+      "loss": 1.1843,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.694029850746269e-05,
+      "loss": 1.1404,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.73134328358209e-05,
+      "loss": 1.0816,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7686567164179104e-05,
+      "loss": 1.0124,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.805970149253731e-05,
+      "loss": 1.0797,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.843283582089552e-05,
+      "loss": 1.0963,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8805970149253736e-05,
+      "loss": 1.1508,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9179104477611945e-05,
+      "loss": 1.0624,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9552238805970146e-05,
+      "loss": 1.0625,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.992537313432836e-05,
+      "loss": 1.1421,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.029850746268657e-05,
+      "loss": 1.0658,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.067164179104478e-05,
+      "loss": 1.092,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.104477611940299e-05,
+      "loss": 1.1327,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1417910447761195e-05,
+      "loss": 1.0323,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1791044776119404e-05,
+      "loss": 1.1511,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.216417910447761e-05,
+      "loss": 1.0655,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.253731343283582e-05,
+      "loss": 1.0758,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2910447761194036e-05,
+      "loss": 1.0854,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.328358208955224e-05,
+      "loss": 0.9939,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3656716417910446e-05,
+      "loss": 0.9554,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.402985074626866e-05,
+      "loss": 1.0758,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.440298507462687e-05,
+      "loss": 1.0765,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.477611940298508e-05,
+      "loss": 1.1331,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5149253731343286e-05,
+      "loss": 1.0285,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5522388059701495e-05,
+      "loss": 1.0186,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.58955223880597e-05,
+      "loss": 1.0832,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.626865671641791e-05,
+      "loss": 1.0358,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.664179104477612e-05,
+      "loss": 0.9947,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7014925373134335e-05,
+      "loss": 1.0947,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.738805970149254e-05,
+      "loss": 1.0447,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7761194029850745e-05,
+      "loss": 1.0068,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.813432835820896e-05,
+      "loss": 1.0772,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.850746268656717e-05,
+      "loss": 1.012,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.888059701492538e-05,
+      "loss": 0.9648,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9253731343283586e-05,
+      "loss": 0.9717,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9626865671641794e-05,
+      "loss": 1.0376,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 0.9567,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0373134328358204e-05,
+      "loss": 1.0865,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.074626865671642e-05,
+      "loss": 1.0658,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.111940298507463e-05,
+      "loss": 1.1227,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.149253731343284e-05,
+      "loss": 1.0454,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1865671641791044e-05,
+      "loss": 1.0894,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.223880597014925e-05,
+      "loss": 1.0559,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.261194029850747e-05,
+      "loss": 1.0767,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.298507462686567e-05,
+      "loss": 1.012,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3358208955223885e-05,
+      "loss": 1.0323,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.373134328358209e-05,
+      "loss": 1.004,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4104477611940295e-05,
+      "loss": 0.9923,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.447761194029851e-05,
+      "loss": 1.0444,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.485074626865672e-05,
+      "loss": 0.9763,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5223880597014934e-05,
+      "loss": 0.9041,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5597014925373135e-05,
+      "loss": 1.0974,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5970149253731344e-05,
+      "loss": 1.0375,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.634328358208956e-05,
+      "loss": 0.9965,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.671641791044776e-05,
+      "loss": 1.0748,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7089552238805976e-05,
+      "loss": 0.9967,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7462686567164184e-05,
+      "loss": 0.9869,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7835820895522386e-05,
+      "loss": 1.0439,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.82089552238806e-05,
+      "loss": 1.0156,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.85820895522388e-05,
+      "loss": 0.9881,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8955223880597025e-05,
+      "loss": 0.8893,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9328358208955226e-05,
+      "loss": 0.9654,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.970149253731343e-05,
+      "loss": 1.0971,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.007462686567164e-05,
+      "loss": 1.0233,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.044776119402985e-05,
+      "loss": 0.9885,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.082089552238807e-05,
+      "loss": 0.963,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.119402985074628e-05,
+      "loss": 1.0074,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.156716417910448e-05,
+      "loss": 1.1093,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.194029850746269e-05,
+      "loss": 0.9376,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.23134328358209e-05,
+      "loss": 1.0311,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.268656716417911e-05,
+      "loss": 0.9679,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.305970149253731e-05,
+      "loss": 0.8725,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.343283582089553e-05,
+      "loss": 1.0965,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.380597014925374e-05,
+      "loss": 0.9385,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.417910447761194e-05,
+      "loss": 1.0613,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.455223880597016e-05,
+      "loss": 1.0488,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.492537313432836e-05,
+      "loss": 0.9979,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.529850746268657e-05,
+      "loss": 1.0338,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.567164179104478e-05,
+      "loss": 1.0164,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.604477611940298e-05,
+      "loss": 0.9104,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.64179104477612e-05,
+      "loss": 0.9717,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.679104477611941e-05,
+      "loss": 0.9828,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.716417910447762e-05,
+      "loss": 0.9571,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.753731343283583e-05,
+      "loss": 1.006,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.791044776119403e-05,
+      "loss": 0.9977,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.828358208955224e-05,
+      "loss": 1.0698,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.865671641791044e-05,
+      "loss": 1.0106,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.902985074626866e-05,
+      "loss": 1.0244,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.940298507462687e-05,
+      "loss": 1.0109,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.977611940298508e-05,
+      "loss": 0.9693,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.014925373134329e-05,
+      "loss": 0.9535,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.052238805970149e-05,
+      "loss": 1.0559,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.089552238805971e-05,
+      "loss": 0.9774,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.126865671641791e-05,
+      "loss": 1.0448,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.164179104477612e-05,
+      "loss": 0.973,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.201492537313434e-05,
+      "loss": 1.0366,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.238805970149254e-05,
+      "loss": 0.9796,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.276119402985076e-05,
+      "loss": 0.9602,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.313432835820896e-05,
+      "loss": 0.8761,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.350746268656716e-05,
+      "loss": 1.0083,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.388059701492537e-05,
+      "loss": 0.9287,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.425373134328359e-05,
+      "loss": 0.9996,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.46268656716418e-05,
+      "loss": 0.9141,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.986,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.537313432835821e-05,
+      "loss": 0.9711,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.574626865671642e-05,
+      "loss": 0.8969,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.611940298507463e-05,
+      "loss": 1.0015,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.649253731343284e-05,
+      "loss": 0.993,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.686567164179104e-05,
+      "loss": 0.9959,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.723880597014926e-05,
+      "loss": 0.9926,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.761194029850747e-05,
+      "loss": 1.0196,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.798507462686567e-05,
+      "loss": 0.9589,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.835820895522389e-05,
+      "loss": 0.9204,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.873134328358209e-05,
+      "loss": 0.967,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.910447761194029e-05,
+      "loss": 0.9276,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.947761194029851e-05,
+      "loss": 0.9867,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.985074626865672e-05,
+      "loss": 1.1217,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.022388059701494e-05,
+      "loss": 1.0306,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.059701492537314e-05,
+      "loss": 0.9571,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.097014925373134e-05,
+      "loss": 0.97,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.134328358208956e-05,
+      "loss": 1.0613,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.171641791044776e-05,
+      "loss": 0.9966,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.208955223880597e-05,
+      "loss": 1.041,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.246268656716419e-05,
+      "loss": 0.9468,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.283582089552239e-05,
+      "loss": 1.0408,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.32089552238806e-05,
+      "loss": 0.9599,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.358208955223881e-05,
+      "loss": 0.992,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.395522388059702e-05,
+      "loss": 0.9164,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.432835820895522e-05,
+      "loss": 0.8068,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.470149253731343e-05,
+      "loss": 0.9893,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.507462686567164e-05,
+      "loss": 0.9557,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.544776119402986e-05,
+      "loss": 0.977,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.582089552238807e-05,
+      "loss": 0.9648,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.619402985074627e-05,
+      "loss": 1.0114,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.656716417910447e-05,
+      "loss": 0.9965,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.694029850746269e-05,
+      "loss": 1.0045,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.731343283582089e-05,
+      "loss": 0.9898,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.76865671641791e-05,
+      "loss": 0.9328,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.805970149253732e-05,
+      "loss": 0.9275,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.843283582089554e-05,
+      "loss": 1.0031,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.880597014925374e-05,
+      "loss": 0.9471,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.917910447761194e-05,
+      "loss": 1.1068,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.955223880597016e-05,
+      "loss": 0.9465,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.992537313432836e-05,
+      "loss": 0.9218,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.029850746268657e-05,
+      "loss": 1.0322,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.067164179104479e-05,
+      "loss": 0.8794,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.104477611940299e-05,
+      "loss": 0.9877,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.14179104477612e-05,
+      "loss": 0.9134,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.17910447761194e-05,
+      "loss": 0.9723,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.216417910447762e-05,
+      "loss": 0.9999,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.253731343283582e-05,
+      "loss": 0.9128,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.291044776119402e-05,
+      "loss": 0.9475,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.328358208955224e-05,
+      "loss": 0.9804,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.365671641791045e-05,
+      "loss": 0.9926,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.402985074626867e-05,
+      "loss": 1.0154,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.440298507462687e-05,
+      "loss": 0.953,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.477611940298507e-05,
+      "loss": 0.9796,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.514925373134329e-05,
+      "loss": 0.8981,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.552238805970149e-05,
+      "loss": 1.0355,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.58955223880597e-05,
+      "loss": 0.9785,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.626865671641792e-05,
+      "loss": 0.9592,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.664179104477612e-05,
+      "loss": 1.0122,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.701492537313434e-05,
+      "loss": 0.9343,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.738805970149254e-05,
+      "loss": 0.8752,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.776119402985075e-05,
+      "loss": 0.9726,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.813432835820896e-05,
+      "loss": 0.9827,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.850746268656717e-05,
+      "loss": 1.0537,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.888059701492539e-05,
+      "loss": 0.9631,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.925373134328359e-05,
+      "loss": 0.9791,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.96268656716418e-05,
+      "loss": 1.0411,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 0.881,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999670080196e-05,
+      "loss": 0.8681,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999998680320823e-05,
+      "loss": 0.8743,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999997030722014e-05,
+      "loss": 0.8925,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999994721283985e-05,
+      "loss": 0.8973,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999991752007043e-05,
+      "loss": 0.9266,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99998812289158e-05,
+      "loss": 0.8563,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999983833938072e-05,
+      "loss": 0.9329,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999978885147086e-05,
+      "loss": 0.9503,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999973276519276e-05,
+      "loss": 0.9603,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999967008055384e-05,
+      "loss": 1.0425,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999960079756234e-05,
+      "loss": 0.9278,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99995249162274e-05,
+      "loss": 0.966,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999944243655907e-05,
+      "loss": 1.0069,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99993533585682e-05,
+      "loss": 0.9837,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999925768226657e-05,
+      "loss": 0.9463,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999915540766677e-05,
+      "loss": 0.7989,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999904653478235e-05,
+      "loss": 0.9733,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999893106362763e-05,
+      "loss": 0.9386,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999880899421789e-05,
+      "loss": 0.9585,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999868032656921e-05,
+      "loss": 1.0342,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999854506069856e-05,
+      "loss": 0.9979,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999840319662381e-05,
+      "loss": 0.9476,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999825473436369e-05,
+      "loss": 1.0265,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999809967393778e-05,
+      "loss": 0.9598,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999793801536654e-05,
+      "loss": 1.0362,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999776975867133e-05,
+      "loss": 1.0032,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999759490387432e-05,
+      "loss": 0.9347,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999741345099859e-05,
+      "loss": 1.0196,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99972254000681e-05,
+      "loss": 0.9708,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999703075110766e-05,
+      "loss": 0.9359,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999682950414296e-05,
+      "loss": 0.9004,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999662165920055e-05,
+      "loss": 0.9601,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999640721630787e-05,
+      "loss": 0.8873,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999618617549322e-05,
+      "loss": 1.0481,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999595853678576e-05,
+      "loss": 1.0002,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999572430021553e-05,
+      "loss": 0.9405,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999548346581345e-05,
+      "loss": 0.9064,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999523603361131e-05,
+      "loss": 1.0579,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999498200364175e-05,
+      "loss": 0.9553,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999472137593829e-05,
+      "loss": 0.9659,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999445415053534e-05,
+      "loss": 0.9914,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999418032746815e-05,
+      "loss": 0.9894,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999389990677287e-05,
+      "loss": 0.9304,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999361288848649e-05,
+      "loss": 0.9393,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99933192726469e-05,
+      "loss": 0.9547,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999301905929286e-05,
+      "loss": 0.9492,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999271224846396e-05,
+      "loss": 0.9788,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99923988402007e-05,
+      "loss": 0.9891,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999207883454444e-05,
+      "loss": 0.9094,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999175223153742e-05,
+      "loss": 0.9735,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999141903122274e-05,
+      "loss": 0.9378,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999107923364434e-05,
+      "loss": 1.0118,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99907328388471e-05,
+      "loss": 0.8716,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999037984687674e-05,
+      "loss": 0.8774,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99900202577798e-05,
+      "loss": 0.9956,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998965407160377e-05,
+      "loss": 0.9323,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998928128839695e-05,
+      "loss": 1.0184,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998890190820855e-05,
+      "loss": 0.9367,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998851593108866e-05,
+      "loss": 0.8194,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998812335708817e-05,
+      "loss": 0.9069,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998772418625891e-05,
+      "loss": 0.9402,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998731841865356e-05,
+      "loss": 0.9558,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998690605432567e-05,
+      "loss": 0.9458,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998648709332965e-05,
+      "loss": 0.9816,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998606153572079e-05,
+      "loss": 0.9071,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998562938155524e-05,
+      "loss": 0.9304,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998519063089007e-05,
+      "loss": 0.9498,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998474528378314e-05,
+      "loss": 1.0065,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998429334029323e-05,
+      "loss": 0.9626,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998383480048e-05,
+      "loss": 0.8764,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998336966440395e-05,
+      "loss": 0.8902,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998289793212645e-05,
+      "loss": 0.9679,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998241960370978e-05,
+      "loss": 0.8878,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998193467921704e-05,
+      "loss": 0.9469,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998144315871225e-05,
+      "loss": 0.9762,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998094504226026e-05,
+      "loss": 0.9016,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998044032992679e-05,
+      "loss": 0.9543,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997992902177848e-05,
+      "loss": 0.9048,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997941111788276e-05,
+      "loss": 0.9473,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997888661830804e-05,
+      "loss": 0.9014,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997835552312348e-05,
+      "loss": 0.8893,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99778178323992e-05,
+      "loss": 0.8867,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997727354620614e-05,
+      "loss": 1.0173,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997672266461613e-05,
+      "loss": 0.9309,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997616518770189e-05,
+      "loss": 1.0111,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997560111553696e-05,
+      "loss": 0.9166,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997503044819581e-05,
+      "loss": 0.9923,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997445318575371e-05,
+      "loss": 1.0056,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997386932828687e-05,
+      "loss": 0.9258,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997327887587233e-05,
+      "loss": 0.9834,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997268182858804e-05,
+      "loss": 1.0005,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997207818651274e-05,
+      "loss": 0.9019,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997146794972611e-05,
+      "loss": 0.9483,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99708511183087e-05,
+      "loss": 0.9627,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99702276923419e-05,
+      "loss": 0.9645,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9969597671908e-05,
+      "loss": 1.0164,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99689610570901e-05,
+      "loss": 0.8573,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996831784797224e-05,
+      "loss": 0.9172,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99676680446393e-05,
+      "loss": 0.8946,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996701164717704e-05,
+      "loss": 1.0106,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996634865567207e-05,
+      "loss": 0.8932,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996567907021191e-05,
+      "loss": 0.8705,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996500289088489e-05,
+      "loss": 0.9011,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996432011778026e-05,
+      "loss": 0.9022,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99636307509881e-05,
+      "loss": 0.8581,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996293479059944e-05,
+      "loss": 0.8612,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996223223670609e-05,
+      "loss": 0.9013,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996152308940075e-05,
+      "loss": 0.9193,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996080734877702e-05,
+      "loss": 0.9564,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996008501492937e-05,
+      "loss": 0.9192,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99593560879531e-05,
+      "loss": 0.9527,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995862056794441e-05,
+      "loss": 0.9544,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995787845500037e-05,
+      "loss": 0.9273,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995712974921894e-05,
+      "loss": 0.8625,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995637445069887e-05,
+      "loss": 0.9512,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995561255953989e-05,
+      "loss": 0.9029,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995484407584252e-05,
+      "loss": 0.9718,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995406899970817e-05,
+      "loss": 1.0289,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995328733123912e-05,
+      "loss": 1.024,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995249907053854e-05,
+      "loss": 0.9518,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995170421771045e-05,
+      "loss": 0.8709,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995090277285977e-05,
+      "loss": 0.8975,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995009473609221e-05,
+      "loss": 0.9328,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994928010751446e-05,
+      "loss": 0.9571,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994845888723398e-05,
+      "loss": 1.057,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994763107535918e-05,
+      "loss": 0.8481,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99467966719993e-05,
+      "loss": 0.9473,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994595567726444e-05,
+      "loss": 0.9743,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994510809126559e-05,
+      "loss": 0.949,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994425391411459e-05,
+      "loss": 0.9689,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994339314592418e-05,
+      "loss": 0.8991,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994252578680796e-05,
+      "loss": 0.8837,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99416518368804e-05,
+      "loss": 0.9137,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994077129625678e-05,
+      "loss": 0.9446,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993988416505338e-05,
+      "loss": 1.0306,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993899044338721e-05,
+      "loss": 0.8862,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993809013137624e-05,
+      "loss": 0.8696,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99371832291393e-05,
+      "loss": 0.8727,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993626973679602e-05,
+      "loss": 0.8251,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9935349654467e-05,
+      "loss": 0.9874,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993442298227365e-05,
+      "loss": 1.0164,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993348972033825e-05,
+      "loss": 0.9286,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993254986878396e-05,
+      "loss": 0.903,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993160342773483e-05,
+      "loss": 0.8308,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993065039731573e-05,
+      "loss": 0.9463,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992969077765246e-05,
+      "loss": 0.9925,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992872456887164e-05,
+      "loss": 0.7892,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992775177110078e-05,
+      "loss": 0.9205,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992677238446827e-05,
+      "loss": 0.9259,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992578640910333e-05,
+      "loss": 0.8651,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992479384513611e-05,
+      "loss": 0.9746,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992379469269758e-05,
+      "loss": 0.9663,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99227889519196e-05,
+      "loss": 0.9203,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992177662293489e-05,
+      "loss": 0.9743,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992075770587705e-05,
+      "loss": 1.0148,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991973220088055e-05,
+      "loss": 0.9283,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991870010808071e-05,
+      "loss": 0.9654,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991766142761376e-05,
+      "loss": 0.8871,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991661615961673e-05,
+      "loss": 0.9789,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99155643042276e-05,
+      "loss": 0.961,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991450586158516e-05,
+      "loss": 0.9015,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991344083182908e-05,
+      "loss": 0.9216,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991236921509994e-05,
+      "loss": 0.945,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991129101153915e-05,
+      "loss": 0.9388,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9910206221289e-05,
+      "loss": 0.9711,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990911484449261e-05,
+      "loss": 0.8926,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990801688129407e-05,
+      "loss": 0.9359,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990691233183823e-05,
+      "loss": 0.9431,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990580119627087e-05,
+      "loss": 0.929,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990468347473862e-05,
+      "loss": 0.9337,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990355916738901e-05,
+      "loss": 0.9561,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990242827437035e-05,
+      "loss": 0.9353,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990129079583194e-05,
+      "loss": 0.917,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990014673192387e-05,
+      "loss": 0.9399,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989899608279713e-05,
+      "loss": 0.9324,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989783884860356e-05,
+      "loss": 0.8931,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989667502949586e-05,
+      "loss": 0.9361,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989550462562764e-05,
+      "loss": 0.9428,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989432763715336e-05,
+      "loss": 0.8865,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989314406422833e-05,
+      "loss": 0.9514,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989195390700874e-05,
+      "loss": 0.8722,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989075716565168e-05,
+      "loss": 0.9168,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988955384031505e-05,
+      "loss": 0.8784,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988834393115767e-05,
+      "loss": 0.9164,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988712743833919e-05,
+      "loss": 1.0222,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988590436202016e-05,
+      "loss": 0.9329,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988467470236201e-05,
+      "loss": 0.8516,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988343845952697e-05,
+      "loss": 1.0152,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98821956336782e-05,
+      "loss": 0.9838,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988094622497974e-05,
+      "loss": 0.9511,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987969023359644e-05,
+      "loss": 0.8873,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987842765969407e-05,
+      "loss": 1.0133,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987715850343923e-05,
+      "loss": 0.9606,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987588276499943e-05,
+      "loss": 0.8856,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987460044454302e-05,
+      "loss": 0.9226,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987331154223923e-05,
+      "loss": 0.8512,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987201605825814e-05,
+      "loss": 0.943,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98707139927707e-05,
+      "loss": 1.0244,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986940534594876e-05,
+      "loss": 0.9135,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986809011796502e-05,
+      "loss": 0.9614,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986676830899307e-05,
+      "loss": 0.9827,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98654399192073e-05,
+      "loss": 0.9156,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986410494878304e-05,
+      "loss": 0.898,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986276339789647e-05,
+      "loss": 0.8997,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98614152667246e-05,
+      "loss": 0.8869,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986006055544538e-05,
+      "loss": 0.9128,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985869926423757e-05,
+      "loss": 0.9597,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985733139328081e-05,
+      "loss": 0.8608,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985595694275563e-05,
+      "loss": 0.8988,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98545759128434e-05,
+      "loss": 0.7912,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985318830372637e-05,
+      "loss": 0.8877,
+      "step": 479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985179411558768e-05,
+      "loss": 0.932,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985039334861129e-05,
+      "loss": 1.0077,
+      "step": 481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984898600298209e-05,
+      "loss": 0.8256,
+      "step": 482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984757207888578e-05,
+      "loss": 0.8497,
+      "step": 483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984615157650895e-05,
+      "loss": 0.955,
+      "step": 484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984472449603908e-05,
+      "loss": 0.847,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984329083766448e-05,
+      "loss": 0.9091,
+      "step": 486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984185060157436e-05,
+      "loss": 0.9836,
+      "step": 487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984040378795878e-05,
+      "loss": 0.9902,
+      "step": 488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983895039700868e-05,
+      "loss": 0.9296,
+      "step": 489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983749042891583e-05,
+      "loss": 0.9319,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983602388387294e-05,
+      "loss": 0.8549,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983455076207353e-05,
+      "loss": 0.9396,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983307106371201e-05,
+      "loss": 0.8968,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983158478898365e-05,
+      "loss": 0.9478,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983009193808458e-05,
+      "loss": 0.9727,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982859251121181e-05,
+      "loss": 0.9611,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982708650856323e-05,
+      "loss": 0.9892,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982557393033758e-05,
+      "loss": 0.888,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982405477673447e-05,
+      "loss": 0.8928,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982252904795438e-05,
+      "loss": 0.8919,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982099674419864e-05,
+      "loss": 1.039,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981945786566948e-05,
+      "loss": 0.9473,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981791241257e-05,
+      "loss": 0.9646,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981636038510413e-05,
+      "loss": 0.9061,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981480178347669e-05,
+      "loss": 0.8962,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981323660789337e-05,
+      "loss": 0.9479,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98116648585607e-05,
+      "loss": 0.9951,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981008653568614e-05,
+      "loss": 0.856,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980850163947795e-05,
+      "loss": 0.9502,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980691017014529e-05,
+      "loss": 0.8089,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980531212789819e-05,
+      "loss": 0.9277,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980370751294754e-05,
+      "loss": 0.9041,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98020963255051e-05,
+      "loss": 0.8969,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980047856578348e-05,
+      "loss": 0.987,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979885423399617e-05,
+      "loss": 0.8651,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979722333035756e-05,
+      "loss": 0.9595,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979558585508286e-05,
+      "loss": 0.9427,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979394180838815e-05,
+      "loss": 0.8663,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97922911904904e-05,
+      "loss": 0.9126,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979063400160746e-05,
+      "loss": 0.861,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978897024195801e-05,
+      "loss": 0.9075,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97872999117616e-05,
+      "loss": 0.9501,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978562301123868e-05,
+      "loss": 0.9063,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978393954061052e-05,
+      "loss": 1.013,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97822495000993e-05,
+      "loss": 0.9414,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978055288992809e-05,
+      "loss": 0.9473,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977884971032072e-05,
+      "loss": 0.9704,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9777139961502e-05,
+      "loss": 0.877,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977542364369756e-05,
+      "loss": 0.908,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977370075713386e-05,
+      "loss": 0.8247,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97719713020383e-05,
+      "loss": 1.0143,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977023527863913e-05,
+      "loss": 0.7913,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976849268716541e-05,
+      "loss": 0.9212,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976674352784711e-05,
+      "loss": 0.8777,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976498780091508e-05,
+      "loss": 0.8413,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976322550660102e-05,
+      "loss": 1.0309,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97614566451375e-05,
+      "loss": 0.9339,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975968121675793e-05,
+      "loss": 0.9038,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975789922169663e-05,
+      "loss": 0.837,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975611066018875e-05,
+      "loss": 0.8188,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975431553247034e-05,
+      "loss": 0.8209,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975251383877829e-05,
+      "loss": 0.848,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975070557935038e-05,
+      "loss": 0.8673,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974889075442521e-05,
+      "loss": 0.8847,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97470693642423e-05,
+      "loss": 0.8693,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974524140904202e-05,
+      "loss": 0.924,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97434068890656e-05,
+      "loss": 0.8676,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974156580455511e-05,
+      "loss": 0.845,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973971815575356e-05,
+      "loss": 0.8894,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973786394290474e-05,
+      "loss": 0.9389,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973600316625337e-05,
+      "loss": 0.8357,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973413582604502e-05,
+      "loss": 0.9436,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97322619225261e-05,
+      "loss": 0.8624,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97303814559439e-05,
+      "loss": 0.865,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97284944265466e-05,
+      "loss": 0.869,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97266008345832e-05,
+      "loss": 0.9322,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972470068030362e-05,
+      "loss": 0.853,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972279396395863e-05,
+      "loss": 0.9141,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972088068579983e-05,
+      "loss": 0.9336,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97189608460797e-05,
+      "loss": 0.9136,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971703444505164e-05,
+      "loss": 0.878,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971510148296983e-05,
+      "loss": 0.8441,
+      "step": 562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97131619600894e-05,
+      "loss": 0.8433,
+      "step": 563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971121587666627e-05,
+      "loss": 0.8167,
+      "step": 564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970926323295726e-05,
+      "loss": 0.9407,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970730402922008e-05,
+      "loss": 0.9071,
+      "step": 566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970533826571329e-05,
+      "loss": 0.9272,
+      "step": 567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970336594269627e-05,
+      "loss": 0.894,
+      "step": 568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970138706042933e-05,
+      "loss": 0.8044,
+      "step": 569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969940161917361e-05,
+      "loss": 0.8799,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969740961919112e-05,
+      "loss": 0.8599,
+      "step": 571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969541106074476e-05,
+      "loss": 0.846,
+      "step": 572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969340594409825e-05,
+      "loss": 0.8847,
+      "step": 573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969139426951621e-05,
+      "loss": 0.9158,
+      "step": 574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968937603726414e-05,
+      "loss": 0.9332,
+      "step": 575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968735124760835e-05,
+      "loss": 0.9473,
+      "step": 576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968531990081605e-05,
+      "loss": 0.8471,
+      "step": 577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968328199715534e-05,
+      "loss": 0.9656,
+      "step": 578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968123753689511e-05,
+      "loss": 0.9423,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967918652030522e-05,
+      "loss": 0.8686,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96771289476563e-05,
+      "loss": 0.8372,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96750648192199e-05,
+      "loss": 0.9878,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96729941352684e-05,
+      "loss": 0.979,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96709168960751e-05,
+      "loss": 0.8499,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966883310191408e-05,
+      "loss": 0.9029,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966674275306038e-05,
+      "loss": 1.0013,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966464584978983e-05,
+      "loss": 0.8634,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966254239237917e-05,
+      "loss": 0.8311,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966043238110598e-05,
+      "loss": 0.8993,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965831581624871e-05,
+      "loss": 0.869,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965619269808668e-05,
+      "loss": 0.889,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96540630269001e-05,
+      "loss": 0.9927,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965192680296998e-05,
+      "loss": 0.9875,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964978402657825e-05,
+      "loss": 0.8835,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964763469800768e-05,
+      "loss": 0.9392,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964547881754194e-05,
+      "loss": 0.8817,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96433163854655e-05,
+      "loss": 0.9294,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964114740206374e-05,
+      "loss": 0.8854,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963897186762292e-05,
+      "loss": 0.8209,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963678978243014e-05,
+      "loss": 0.9049,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963460114677333e-05,
+      "loss": 0.879,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963240596094133e-05,
+      "loss": 0.9558,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963020422522387e-05,
+      "loss": 0.8414,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962799593991146e-05,
+      "loss": 0.9444,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962578110529555e-05,
+      "loss": 0.9732,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962355972166843e-05,
+      "loss": 0.8488,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962133178932324e-05,
+      "loss": 0.7783,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9619097308554e-05,
+      "loss": 0.9292,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96168562796556e-05,
+      "loss": 0.8241,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961460870292376e-05,
+      "loss": 0.8685,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961235457865511e-05,
+      "loss": 0.8214,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96100939071471e-05,
+      "loss": 0.9319,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96078266886981e-05,
+      "loss": 0.8119,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960555292360727e-05,
+      "loss": 0.8875,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96032726121747e-05,
+      "loss": 0.8999,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960098575470132e-05,
+      "loss": 0.9926,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959869235148892e-05,
+      "loss": 0.9172,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959639240284012e-05,
+      "loss": 0.8366,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959408590905848e-05,
+      "loss": 0.9131,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959177287044839e-05,
+      "loss": 0.7713,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958945328731506e-05,
+      "loss": 0.8903,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958712715996463e-05,
+      "loss": 0.776,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958479448870406e-05,
+      "loss": 0.9426,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958245527384118e-05,
+      "loss": 0.8486,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958010951568472e-05,
+      "loss": 0.8034,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957775721454421e-05,
+      "loss": 0.9414,
+      "step": 626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957539837073012e-05,
+      "loss": 0.9241,
+      "step": 627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957303298455369e-05,
+      "loss": 0.8199,
+      "step": 628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957066105632712e-05,
+      "loss": 0.9384,
+      "step": 629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956828258636341e-05,
+      "loss": 0.836,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956589757497643e-05,
+      "loss": 0.9027,
+      "step": 631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956350602248095e-05,
+      "loss": 0.9503,
+      "step": 632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956110792919257e-05,
+      "loss": 0.9797,
+      "step": 633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955870329542775e-05,
+      "loss": 0.8015,
+      "step": 634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955629212150383e-05,
+      "loss": 0.8415,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9553874407739e-05,
+      "loss": 0.8251,
+      "step": 636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955145015445235e-05,
+      "loss": 0.9221,
+      "step": 637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954901936196377e-05,
+      "loss": 0.9155,
+      "step": 638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954658203059408e-05,
+      "loss": 0.9749,
+      "step": 639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95441381606649e-05,
+      "loss": 0.9517,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954168775249874e-05,
+      "loss": 0.9255,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953923080641899e-05,
+      "loss": 0.828,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95367673227499e-05,
+      "loss": 0.8871,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953429730181653e-05,
+      "loss": 0.9087,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953182074394489e-05,
+      "loss": 0.9543,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952933764946179e-05,
+      "loss": 0.9003,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952684801869492e-05,
+      "loss": 0.8824,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952435185197281e-05,
+      "loss": 0.9383,
+      "step": 648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952184914962491e-05,
+      "loss": 0.8505,
+      "step": 649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951933991198146e-05,
+      "loss": 0.8356,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951682413937361e-05,
+      "loss": 0.889,
+      "step": 651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951430183213337e-05,
+      "loss": 1.0179,
+      "step": 652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951177299059362e-05,
+      "loss": 0.8792,
+      "step": 653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950923761508805e-05,
+      "loss": 0.8998,
+      "step": 654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950669570595127e-05,
+      "loss": 0.9592,
+      "step": 655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950414726351872e-05,
+      "loss": 0.8178,
+      "step": 656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950159228812671e-05,
+      "loss": 0.9418,
+      "step": 657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949903078011244e-05,
+      "loss": 0.9527,
+      "step": 658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949646273981394e-05,
+      "loss": 0.8778,
+      "step": 659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949388816757008e-05,
+      "loss": 0.8794,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949130706372064e-05,
+      "loss": 0.961,
+      "step": 661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948871942860624e-05,
+      "loss": 0.8981,
+      "step": 662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948612526256837e-05,
+      "loss": 0.8499,
+      "step": 663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948352456594938e-05,
+      "loss": 0.767,
+      "step": 664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948091733909247e-05,
+      "loss": 0.8834,
+      "step": 665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947830358234172e-05,
+      "loss": 0.9255,
+      "step": 666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947568329604204e-05,
+      "loss": 0.8937,
+      "step": 667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947305648053924e-05,
+      "loss": 0.9231,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.947042313617998e-05,
+      "loss": 0.9314,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946778326331177e-05,
+      "loss": 0.8628,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9465136862283e-05,
+      "loss": 0.83,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946248393344288e-05,
+      "loss": 0.9693,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945982447714155e-05,
+      "loss": 0.848,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945715849372995e-05,
+      "loss": 0.9086,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94544859835599e-05,
+      "loss": 0.9049,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94518069469841e-05,
+      "loss": 0.9977,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944912138435609e-05,
+      "loss": 0.9384,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944642929603028e-05,
+      "loss": 0.9477,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944373068236192e-05,
+      "loss": 0.8787,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944102554370718e-05,
+      "loss": 0.8042,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943831388042302e-05,
+      "loss": 0.9196,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94355956928673e-05,
+      "loss": 0.9208,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943287098139875e-05,
+      "loss": 0.8959,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943013974637692e-05,
+      "loss": 0.8633,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942740198816228e-05,
+      "loss": 0.8978,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942465770711607e-05,
+      "loss": 0.9005,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94219069036005e-05,
+      "loss": 0.89,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941914957797855e-05,
+      "loss": 0.9274,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941638573061414e-05,
+      "loss": 0.8983,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941361536187196e-05,
+      "loss": 0.8092,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941083847211765e-05,
+      "loss": 0.8353,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940805506171764e-05,
+      "loss": 0.8607,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940526513103929e-05,
+      "loss": 0.9017,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940246868045074e-05,
+      "loss": 0.8255,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939966571032107e-05,
+      "loss": 0.9388,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939685622102013e-05,
+      "loss": 0.8818,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939404021291873e-05,
+      "loss": 0.9476,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939121768638848e-05,
+      "loss": 0.9827,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938838864180185e-05,
+      "loss": 0.9407,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938555307953222e-05,
+      "loss": 0.7971,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938271099995373e-05,
+      "loss": 1.0053,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93798624034415e-05,
+      "loss": 0.8618,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937700729037143e-05,
+      "loss": 0.8064,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93741456611203e-05,
+      "loss": 0.8708,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937127751606576e-05,
+      "loss": 0.9649,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936840285558632e-05,
+      "loss": 0.9405,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936552168006133e-05,
+      "loss": 0.8758,
+      "step": 707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936263398987102e-05,
+      "loss": 0.8808,
+      "step": 708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935973978539647e-05,
+      "loss": 0.7658,
+      "step": 709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935683906701963e-05,
+      "loss": 1.0023,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935393183512327e-05,
+      "loss": 0.9761,
+      "step": 711
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93510180900911e-05,
+      "loss": 0.8345,
+      "step": 712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93480978323076e-05,
+      "loss": 0.9378,
+      "step": 713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934517106215819e-05,
+      "loss": 0.9021,
+      "step": 714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934223778002906e-05,
+      "loss": 0.8591,
+      "step": 715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933929798630736e-05,
+      "loss": 0.7823,
+      "step": 716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933635168138102e-05,
+      "loss": 0.9061,
+      "step": 717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933339886563886e-05,
+      "loss": 0.9333,
+      "step": 718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933043953947055e-05,
+      "loss": 0.8641,
+      "step": 719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932747370326666e-05,
+      "loss": 0.8288,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932450135741853e-05,
+      "loss": 1.004,
+      "step": 721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932152250231845e-05,
+      "loss": 0.9051,
+      "step": 722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931853713835953e-05,
+      "loss": 0.9595,
+      "step": 723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931554526593576e-05,
+      "loss": 0.8411,
+      "step": 724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931254688544193e-05,
+      "loss": 0.8292,
+      "step": 725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930954199727376e-05,
+      "loss": 0.8944,
+      "step": 726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93065306018278e-05,
+      "loss": 0.8112,
+      "step": 727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930351269950143e-05,
+      "loss": 0.9864,
+      "step": 728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930048829069294e-05,
+      "loss": 0.9227,
+      "step": 729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929745737580145e-05,
+      "loss": 0.8708,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929441995522695e-05,
+      "loss": 0.953,
+      "step": 731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929137602937027e-05,
+      "loss": 0.8474,
+      "step": 732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928832559863313e-05,
+      "loss": 0.8339,
+      "step": 733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928526866341805e-05,
+      "loss": 0.8902,
+      "step": 734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928220522412849e-05,
+      "loss": 0.831,
+      "step": 735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92791352811687e-05,
+      "loss": 0.9137,
+      "step": 736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927605883494382e-05,
+      "loss": 0.8676,
+      "step": 737
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927297588585984e-05,
+      "loss": 0.8937,
+      "step": 738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926988643432362e-05,
+      "loss": 0.8295,
+      "step": 739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926679048074288e-05,
+      "loss": 0.9167,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926368802552615e-05,
+      "loss": 0.9227,
+      "step": 741
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926057906908288e-05,
+      "loss": 0.899,
+      "step": 742
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925746361182334e-05,
+      "loss": 0.8203,
+      "step": 743
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925434165415868e-05,
+      "loss": 0.8669,
+      "step": 744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92512131965009e-05,
+      "loss": 0.8988,
+      "step": 745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924807823926284e-05,
+      "loss": 0.9142,
+      "step": 746
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924493678285825e-05,
+      "loss": 0.8482,
+      "step": 747
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924178882770165e-05,
+      "loss": 0.8766,
+      "step": 748
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92386343742085e-05,
+      "loss": 0.8886,
+      "step": 749
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923547342279509e-05,
+      "loss": 0.8962,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923230597387856e-05,
+      "loss": 0.8106,
+      "step": 751
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92291320278769e-05,
+      "loss": 0.9491,
+      "step": 752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922595158520897e-05,
+      "loss": 0.8126,
+      "step": 753
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92227646462945e-05,
+      "loss": 0.7792,
+      "step": 754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921957121155408e-05,
+      "loss": 0.8699,
+      "step": 755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921637128140908e-05,
+      "loss": 0.9217,
+      "step": 756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921316485628185e-05,
+      "loss": 0.8297,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.92099519365955e-05,
+      "loss": 0.8711,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920673252277406e-05,
+      "loss": 0.796,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920350661524236e-05,
+      "loss": 0.9135,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920027421442613e-05,
+      "loss": 0.9244,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919703532075196e-05,
+      "loss": 0.908,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919378993464725e-05,
+      "loss": 0.9921,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919053805654029e-05,
+      "loss": 0.9166,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918727968686024e-05,
+      "loss": 0.8164,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91840148260371e-05,
+      "loss": 0.8312,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918074347450173e-05,
+      "loss": 0.8323,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917746563268581e-05,
+      "loss": 0.8901,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917418130102193e-05,
+      "loss": 0.906,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917089047994354e-05,
+      "loss": 0.8186,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916759316988491e-05,
+      "loss": 0.883,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916428937128116e-05,
+      "loss": 0.819,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91609790845683e-05,
+      "loss": 0.8204,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915766231018318e-05,
+      "loss": 0.9228,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915433904856351e-05,
+      "loss": 0.9328,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915100930014785e-05,
+      "loss": 0.798,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91476730653756e-05,
+      "loss": 0.8367,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914433034468708e-05,
+      "loss": 0.874,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91409811385234e-05,
+      "loss": 0.8082,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913762544732655e-05,
+      "loss": 0.8254,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913426327153937e-05,
+      "loss": 0.89,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913089461160555e-05,
+      "loss": 0.8681,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912751946796964e-05,
+      "loss": 0.9747,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912413784107708e-05,
+      "loss": 0.8857,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912074973137412e-05,
+      "loss": 0.9238,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91173551393079e-05,
+      "loss": 0.8399,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911395406532637e-05,
+      "loss": 0.8739,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911054650987837e-05,
+      "loss": 0.7345,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91071324734136e-05,
+      "loss": 0.774,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910371195638259e-05,
+      "loss": 0.8933,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910028495923673e-05,
+      "loss": 0.8744,
+      "step": 791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909685148242831e-05,
+      "loss": 0.8435,
+      "step": 792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909341152641041e-05,
+      "loss": 0.8312,
+      "step": 793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908996509163699e-05,
+      "loss": 0.9284,
+      "step": 794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908651217856289e-05,
+      "loss": 0.9359,
+      "step": 795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908305278764376e-05,
+      "loss": 0.8234,
+      "step": 796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907958691933615e-05,
+      "loss": 0.9314,
+      "step": 797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907611457409744e-05,
+      "loss": 0.9056,
+      "step": 798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907263575238585e-05,
+      "loss": 0.8716,
+      "step": 799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90691504546605e-05,
+      "loss": 0.8165,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90656586813813e-05,
+      "loss": 0.7985,
+      "step": 801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906216043300909e-05,
+      "loss": 0.8669,
+      "step": 802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905865571000552e-05,
+      "loss": 0.8942,
+      "step": 803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905514451283308e-05,
+      "loss": 0.9217,
+      "step": 804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905162684195514e-05,
+      "loss": 0.9603,
+      "step": 805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904810269783596e-05,
+      "loss": 0.8787,
+      "step": 806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904457208094054e-05,
+      "loss": 0.8064,
+      "step": 807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904103499173487e-05,
+      "loss": 0.9251,
+      "step": 808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903749143068573e-05,
+      "loss": 0.9304,
+      "step": 809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903394139826071e-05,
+      "loss": 0.8789,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903038489492834e-05,
+      "loss": 0.9178,
+      "step": 811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902682192115795e-05,
+      "loss": 0.8593,
+      "step": 812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902325247741975e-05,
+      "loss": 0.9503,
+      "step": 813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901967656418478e-05,
+      "loss": 0.8506,
+      "step": 814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901609418192494e-05,
+      "loss": 0.8821,
+      "step": 815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901250533111302e-05,
+      "loss": 0.879,
+      "step": 816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900891001222258e-05,
+      "loss": 0.9625,
+      "step": 817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900530822572815e-05,
+      "loss": 0.8519,
+      "step": 818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900169997210503e-05,
+      "loss": 0.9073,
+      "step": 819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899808525182935e-05,
+      "loss": 0.893,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899446406537817e-05,
+      "loss": 0.9841,
+      "step": 821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89908364132294e-05,
+      "loss": 0.8274,
+      "step": 822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898720229586171e-05,
+      "loss": 0.8849,
+      "step": 823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898356171375473e-05,
+      "loss": 0.8315,
+      "step": 824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897991466738892e-05,
+      "loss": 0.9546,
+      "step": 825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897626115724552e-05,
+      "loss": 0.9679,
+      "step": 826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89726011838067e-05,
+      "loss": 0.9145,
+      "step": 827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896893474755546e-05,
+      "loss": 0.877,
+      "step": 828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896526184897566e-05,
+      "loss": 0.9128,
+      "step": 829
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896158248855198e-05,
+      "loss": 0.9206,
+      "step": 830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895789666677002e-05,
+      "loss": 0.7754,
+      "step": 831
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895420438411616e-05,
+      "loss": 0.7817,
+      "step": 832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895050564107764e-05,
+      "loss": 0.8583,
+      "step": 833
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894680043814262e-05,
+      "loss": 0.8594,
+      "step": 834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894308877580006e-05,
+      "loss": 0.906,
+      "step": 835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893937065453976e-05,
+      "loss": 0.9686,
+      "step": 836
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893564607485239e-05,
+      "loss": 0.8355,
+      "step": 837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893191503722952e-05,
+      "loss": 0.8649,
+      "step": 838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892817754216348e-05,
+      "loss": 0.8731,
+      "step": 839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892443359014751e-05,
+      "loss": 0.9228,
+      "step": 840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89206831816757e-05,
+      "loss": 0.8132,
+      "step": 841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891692631724299e-05,
+      "loss": 0.9061,
+      "step": 842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891316299734515e-05,
+      "loss": 0.9578,
+      "step": 843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890939322247881e-05,
+      "loss": 0.9953,
+      "step": 844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890561699314149e-05,
+      "loss": 0.902,
+      "step": 845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890183430983151e-05,
+      "loss": 0.8402,
+      "step": 846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.889804517304806e-05,
+      "loss": 0.8849,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88942495832912e-05,
+      "loss": 0.8762,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889044754106181e-05,
+      "loss": 0.9393,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888663904686167e-05,
+      "loss": 0.8396,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888282410119331e-05,
+      "loss": 0.8756,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887900270456025e-05,
+      "loss": 0.8896,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887517485746678e-05,
+      "loss": 0.9235,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887134056041802e-05,
+      "loss": 0.9294,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886749981392e-05,
+      "loss": 0.9264,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886365261847957e-05,
+      "loss": 0.939,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885979897460444e-05,
+      "loss": 0.8714,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885593888280316e-05,
+      "loss": 0.8958,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885207234358514e-05,
+      "loss": 0.9035,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884819935746062e-05,
+      "loss": 0.8514,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884431992494075e-05,
+      "loss": 0.9046,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884043404653746e-05,
+      "loss": 0.828,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88365417227636e-05,
+      "loss": 0.9364,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883264295413277e-05,
+      "loss": 0.8262,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882873774115954e-05,
+      "loss": 0.9288,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882482608435923e-05,
+      "loss": 0.8996,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882090798424808e-05,
+      "loss": 0.8838,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881698344134316e-05,
+      "loss": 0.8621,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881305245616237e-05,
+      "loss": 0.8925,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880911502922445e-05,
+      "loss": 0.8373,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880517116104907e-05,
+      "loss": 0.8354,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880122085215662e-05,
+      "loss": 0.8988,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879726410306849e-05,
+      "loss": 0.8026,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879330091430679e-05,
+      "loss": 0.8149,
+      "step": 874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878933128639456e-05,
+      "loss": 0.915,
+      "step": 875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878535521985567e-05,
+      "loss": 0.9123,
+      "step": 876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87813727152148e-05,
+      "loss": 0.9148,
+      "step": 877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877738377299755e-05,
+      "loss": 0.8893,
+      "step": 878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877338839373032e-05,
+      "loss": 0.8348,
+      "step": 879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876938657794034e-05,
+      "loss": 0.9433,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876537832615578e-05,
+      "loss": 0.864,
+      "step": 881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876136363890555e-05,
+      "loss": 0.8646,
+      "step": 882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875734251671949e-05,
+      "loss": 0.8873,
+      "step": 883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875331496012823e-05,
+      "loss": 0.84,
+      "step": 884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874928096966332e-05,
+      "loss": 0.9209,
+      "step": 885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874524054585708e-05,
+      "loss": 0.8841,
+      "step": 886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874119368924273e-05,
+      "loss": 0.9397,
+      "step": 887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873714040035433e-05,
+      "loss": 0.8771,
+      "step": 888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873308067972678e-05,
+      "loss": 0.9125,
+      "step": 889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872901452789585e-05,
+      "loss": 0.8788,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87249419453981e-05,
+      "loss": 0.8457,
+      "step": 891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8720862932771e-05,
+      "loss": 0.8504,
+      "step": 892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871677749055288e-05,
+      "loss": 0.9013,
+      "step": 893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871268561928284e-05,
+      "loss": 0.9103,
+      "step": 894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870858731950091e-05,
+      "loss": 0.827,
+      "step": 895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870448259174792e-05,
+      "loss": 0.8473,
+      "step": 896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870037143656556e-05,
+      "loss": 0.8367,
+      "step": 897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869625385449636e-05,
+      "loss": 0.7744,
+      "step": 898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869212984608374e-05,
+      "loss": 0.8205,
+      "step": 899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868799941187192e-05,
+      "loss": 0.9329,
+      "step": 900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868386255240598e-05,
+      "loss": 0.8734,
+      "step": 901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867971926823186e-05,
+      "loss": 0.8373,
+      "step": 902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867556955989632e-05,
+      "loss": 0.8778,
+      "step": 903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867141342794702e-05,
+      "loss": 0.8371,
+      "step": 904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866725087293242e-05,
+      "loss": 0.843,
+      "step": 905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866308189540185e-05,
+      "loss": 0.8494,
+      "step": 906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865890649590548e-05,
+      "loss": 0.8786,
+      "step": 907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865472467499431e-05,
+      "loss": 0.8294,
+      "step": 908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865053643322024e-05,
+      "loss": 0.9281,
+      "step": 909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864634177113596e-05,
+      "loss": 0.9004,
+      "step": 910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864214068929502e-05,
+      "loss": 0.8541,
+      "step": 911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863793318825186e-05,
+      "loss": 1.0041,
+      "step": 912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863371926856172e-05,
+      "loss": 0.9416,
+      "step": 913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86294989307807e-05,
+      "loss": 0.857,
+      "step": 914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862527217546576e-05,
+      "loss": 0.8635,
+      "step": 915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862103900317468e-05,
+      "loss": 0.861,
+      "step": 916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86167994144661e-05,
+      "loss": 0.8922,
+      "step": 917
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861255340989953e-05,
+      "loss": 0.8149,
+      "step": 918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860830099003528e-05,
+      "loss": 0.8679,
+      "step": 919
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860404215543457e-05,
+      "loss": 0.8326,
+      "step": 920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85997769066594e-05,
+      "loss": 0.9174,
+      "step": 921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859550524427266e-05,
+      "loss": 0.8903,
+      "step": 922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859122716883807e-05,
+      "loss": 0.8962,
+      "step": 923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858694268092019e-05,
+      "loss": 0.9318,
+      "step": 924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858265178108444e-05,
+      "loss": 0.8033,
+      "step": 925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857835446989707e-05,
+      "loss": 0.8849,
+      "step": 926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857405074792523e-05,
+      "loss": 0.8858,
+      "step": 927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856974061573681e-05,
+      "loss": 0.753,
+      "step": 928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856542407390066e-05,
+      "loss": 0.7553,
+      "step": 929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856110112298642e-05,
+      "loss": 0.8454,
+      "step": 930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855677176356454e-05,
+      "loss": 0.9597,
+      "step": 931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85524359962064e-05,
+      "loss": 0.8991,
+      "step": 932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854809382148416e-05,
+      "loss": 0.8876,
+      "step": 933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854374523997086e-05,
+      "loss": 0.875,
+      "step": 934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853939025224035e-05,
+      "loss": 0.8875,
+      "step": 935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853502885886738e-05,
+      "loss": 0.8577,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853066106042752e-05,
+      "loss": 0.8479,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852628685749713e-05,
+      "loss": 0.8624,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85219062506535e-05,
+      "loss": 0.9823,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851751924047472e-05,
+      "loss": 0.8929,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851312582753975e-05,
+      "loss": 0.8002,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850872601242837e-05,
+      "loss": 0.8479,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85043197957212e-05,
+      "loss": 0.8103,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849990717799973e-05,
+      "loss": 0.8513,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849548815984631e-05,
+      "loss": 0.9355,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849106274184406e-05,
+      "loss": 0.8384,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848663092457702e-05,
+      "loss": 0.9512,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848219270863006e-05,
+      "loss": 0.8662,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847774809458885e-05,
+      "loss": 0.8578,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847329708303996e-05,
+      "loss": 0.7759,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846883967457077e-05,
+      "loss": 0.886,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846437586976952e-05,
+      "loss": 0.9703,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84599056692253e-05,
+      "loss": 0.8802,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8455429073528e-05,
+      "loss": 0.8865,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845094608326843e-05,
+      "loss": 0.954,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844645669903816e-05,
+      "loss": 0.8318,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844196092142968e-05,
+      "loss": 0.9536,
+      "step": 957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843745875103627e-05,
+      "loss": 0.7531,
+      "step": 958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843295018845207e-05,
+      "loss": 0.8709,
+      "step": 959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842843523427207e-05,
+      "loss": 0.8969,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84239138890921e-05,
+      "loss": 0.895,
+      "step": 961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841938615350883e-05,
+      "loss": 0.9542,
+      "step": 962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841485202811978e-05,
+      "loss": 0.8605,
+      "step": 963
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841031151352331e-05,
+      "loss": 0.8456,
+      "step": 964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840576461031861e-05,
+      "loss": 0.8513,
+      "step": 965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840121131910576e-05,
+      "loss": 0.8388,
+      "step": 966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839665164048559e-05,
+      "loss": 0.851,
+      "step": 967
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839208557505987e-05,
+      "loss": 0.889,
+      "step": 968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83875131234312e-05,
+      "loss": 0.899,
+      "step": 969
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838293428620294e-05,
+      "loss": 0.7503,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837834906397937e-05,
+      "loss": 0.8769,
+      "step": 971
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837375745736562e-05,
+      "loss": 0.8767,
+      "step": 972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836915946696759e-05,
+      "loss": 0.8409,
+      "step": 973
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836455509339209e-05,
+      "loss": 0.8242,
+      "step": 974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835994433724676e-05,
+      "loss": 0.9157,
+      "step": 975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835532719914005e-05,
+      "loss": 0.9248,
+      "step": 976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835070367968129e-05,
+      "loss": 0.9471,
+      "step": 977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834607377948062e-05,
+      "loss": 0.7924,
+      "step": 978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834143749914906e-05,
+      "loss": 0.8884,
+      "step": 979
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833679483929845e-05,
+      "loss": 0.8186,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833214580054145e-05,
+      "loss": 0.8879,
+      "step": 981
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83274903834916e-05,
+      "loss": 0.8394,
+      "step": 982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832282858876325e-05,
+      "loss": 0.8502,
+      "step": 983
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831816041697163e-05,
+      "loss": 0.8512,
+      "step": 984
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831348586873277e-05,
+      "loss": 0.8821,
+      "step": 985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830880494466358e-05,
+      "loss": 0.9919,
+      "step": 986
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830411764538177e-05,
+      "loss": 0.8022,
+      "step": 987
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829942397150594e-05,
+      "loss": 0.784,
+      "step": 988
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829472392365547e-05,
+      "loss": 0.8579,
+      "step": 989
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829001750245064e-05,
+      "loss": 0.8787,
+      "step": 990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828530470851254e-05,
+      "loss": 0.9192,
+      "step": 991
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828058554246308e-05,
+      "loss": 0.8144,
+      "step": 992
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82758600049251e-05,
+      "loss": 0.8976,
+      "step": 993
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82711280965222e-05,
+      "loss": 0.9179,
+      "step": 994
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826638981787879e-05,
+      "loss": 0.8999,
+      "step": 995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826164516962022e-05,
+      "loss": 0.8218,
+      "step": 996
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82568941523726e-05,
+      "loss": 0.8004,
+      "step": 997
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825213676676295e-05,
+      "loss": 0.8373,
+      "step": 998
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824737301341906e-05,
+      "loss": 0.9222,
+      "step": 999
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824260289296962e-05,
+      "loss": 0.8641,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82378264060441e-05,
+      "loss": 0.8862,
+      "step": 1001
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823304355327288e-05,
+      "loss": 0.8827,
+      "step": 1002
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82282543352871e-05,
+      "loss": 0.8462,
+      "step": 1003
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822345875271883e-05,
+      "loss": 0.8121,
+      "step": 1004
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82186568062009e-05,
+      "loss": 0.8746,
+      "step": 1005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821384849636703e-05,
+      "loss": 0.8502,
+      "step": 1006
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820903382385174e-05,
+      "loss": 0.7189,
+      "step": 1007
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820421278929044e-05,
+      "loss": 0.8675,
+      "step": 1008
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819938539331934e-05,
+      "loss": 0.8369,
+      "step": 1009
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819455163657548e-05,
+      "loss": 0.9162,
+      "step": 1010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81897115196968e-05,
+      "loss": 0.9496,
+      "step": 1011
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818486504332203e-05,
+      "loss": 0.9616,
+      "step": 1012
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818001220809072e-05,
+      "loss": 0.8674,
+      "step": 1013
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817515301464332e-05,
+      "loss": 0.7875,
+      "step": 1014
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817028746362107e-05,
+      "loss": 0.9275,
+      "step": 1015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81654155556661e-05,
+      "loss": 0.8333,
+      "step": 1016
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81605372914213e-05,
+      "loss": 0.8321,
+      "step": 1017
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815565267153047e-05,
+      "loss": 0.8574,
+      "step": 1018
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815076169663819e-05,
+      "loss": 0.9558,
+      "step": 1019
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814586436738998e-05,
+      "loss": 0.8792,
+      "step": 1020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814096068443206e-05,
+      "loss": 0.8889,
+      "step": 1021
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813605064841161e-05,
+      "loss": 0.913,
+      "step": 1022
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813113425997655e-05,
+      "loss": 0.8655,
+      "step": 1023
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812621151977574e-05,
+      "loss": 0.8082,
+      "step": 1024
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812128242845877e-05,
+      "loss": 0.8656,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811634698667616e-05,
+      "loss": 0.9359,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811140519507921e-05,
+      "loss": 0.8933,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.81064570543201e-05,
+      "loss": 1.0076,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810150256505179e-05,
+      "loss": 0.8833,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809654172792815e-05,
+      "loss": 0.8687,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809157454360383e-05,
+      "loss": 0.8379,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808660101273434e-05,
+      "loss": 0.8949,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808162113597605e-05,
+      "loss": 0.8152,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80766349139861e-05,
+      "loss": 0.8267,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807164234742257e-05,
+      "loss": 0.853,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806664343694425e-05,
+      "loss": 0.8952,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806163818321089e-05,
+      "loss": 0.8156,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805662658688299e-05,
+      "loss": 0.85,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805160864862195e-05,
+      "loss": 0.7986,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804658436908996e-05,
+      "loss": 0.763,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804155374895006e-05,
+      "loss": 0.838,
+      "step": 1041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803651678886614e-05,
+      "loss": 0.8752,
+      "step": 1042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80314734895029e-05,
+      "loss": 0.832,
+      "step": 1043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802642385152591e-05,
+      "loss": 0.8228,
+      "step": 1044
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802136787560157e-05,
+      "loss": 0.8498,
+      "step": 1045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801630556239707e-05,
+      "loss": 0.9043,
+      "step": 1046
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80112369125805e-05,
+      "loss": 0.8506,
+      "step": 1047
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800616192682077e-05,
+      "loss": 0.8513,
+      "step": 1048
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800108060578759e-05,
+      "loss": 0.8879,
+      "step": 1049
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799599295015154e-05,
+      "loss": 0.8582,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799089896058405e-05,
+      "loss": 0.9275,
+      "step": 1051
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798579863775732e-05,
+      "loss": 0.9384,
+      "step": 1052
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798069198234445e-05,
+      "loss": 0.9137,
+      "step": 1053
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797557899501937e-05,
+      "loss": 0.847,
+      "step": 1054
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79704596764568e-05,
+      "loss": 0.8974,
+      "step": 1055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796533402733234e-05,
+      "loss": 0.8393,
+      "step": 1056
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796020204832244e-05,
+      "loss": 0.8375,
+      "step": 1057
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79550637401043e-05,
+      "loss": 0.8395,
+      "step": 1058
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794991910335607e-05,
+      "loss": 0.897,
+      "step": 1059
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794476813875663e-05,
+      "loss": 0.8164,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793961084698576e-05,
+      "loss": 0.8151,
+      "step": 1061
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793444722872407e-05,
+      "loss": 0.9135,
+      "step": 1062
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792927728465298e-05,
+      "loss": 0.8827,
+      "step": 1063
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792410101545476e-05,
+      "loss": 0.875,
+      "step": 1064
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791891842181249e-05,
+      "loss": 0.7958,
+      "step": 1065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791372950441014e-05,
+      "loss": 0.9024,
+      "step": 1066
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790853426393245e-05,
+      "loss": 0.8641,
+      "step": 1067
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790333270106504e-05,
+      "loss": 0.9242,
+      "step": 1068
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789812481649437e-05,
+      "loss": 0.948,
+      "step": 1069
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789291061090767e-05,
+      "loss": 1.007,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788769008499308e-05,
+      "loss": 0.8347,
+      "step": 1071
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788246323943953e-05,
+      "loss": 0.9582,
+      "step": 1072
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787723007493682e-05,
+      "loss": 0.9549,
+      "step": 1073
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78719905921755e-05,
+      "loss": 0.8635,
+      "step": 1074
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78667447918471e-05,
+      "loss": 0.9144,
+      "step": 1075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786149267464382e-05,
+      "loss": 0.8997,
+      "step": 1076
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78562342412588e-05,
+      "loss": 0.7857,
+      "step": 1077
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7850969492386e-05,
+      "loss": 0.8302,
+      "step": 1078
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784569842872018e-05,
+      "loss": 0.8428,
+      "step": 1079
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784042105095694e-05,
+      "loss": 0.8218,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783513735979274e-05,
+      "loss": 0.8575,
+      "step": 1081
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782984735592489e-05,
+      "loss": 0.8866,
+      "step": 1082
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782455104005144e-05,
+      "loss": 0.9245,
+      "step": 1083
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781924841287136e-05,
+      "loss": 0.8505,
+      "step": 1084
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781393947508443e-05,
+      "loss": 0.7876,
+      "step": 1085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780862422739126e-05,
+      "loss": 0.9839,
+      "step": 1086
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780330267049331e-05,
+      "loss": 0.8256,
+      "step": 1087
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77979748050928e-05,
+      "loss": 0.8699,
+      "step": 1088
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77926406318929e-05,
+      "loss": 0.9183,
+      "step": 1089
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778730015159751e-05,
+      "loss": 0.9131,
+      "step": 1090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778195336491142e-05,
+      "loss": 0.8648,
+      "step": 1091
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777660027254022e-05,
+      "loss": 0.8799,
+      "step": 1092
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777124087519037e-05,
+      "loss": 0.8531,
+      "step": 1093
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77658751735691e-05,
+      "loss": 0.8031,
+      "step": 1094
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776050316838455e-05,
+      "loss": 0.9049,
+      "step": 1095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775512486034563e-05,
+      "loss": 0.8124,
+      "step": 1096
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77497402501621e-05,
+      "loss": 0.7605,
+      "step": 1097
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774434933854459e-05,
+      "loss": 0.8877,
+      "step": 1098
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77389521262045e-05,
+      "loss": 0.8904,
+      "step": 1099
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773354861385407e-05,
+      "loss": 0.8013,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772813880220643e-05,
+      "loss": 0.877,
+      "step": 1101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772272269197549e-05,
+      "loss": 0.8742,
+      "step": 1102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771730028387598e-05,
+      "loss": 0.8611,
+      "step": 1103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771187157862351e-05,
+      "loss": 0.7204,
+      "step": 1104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770643657693449e-05,
+      "loss": 0.7866,
+      "step": 1105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770099527952617e-05,
+      "loss": 0.799,
+      "step": 1106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76955476871166e-05,
+      "loss": 0.8355,
+      "step": 1107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769009380042471e-05,
+      "loss": 0.8013,
+      "step": 1108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768463362017023e-05,
+      "loss": 0.821,
+      "step": 1109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767916714707374e-05,
+      "loss": 0.763,
+      "step": 1110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767369438185663e-05,
+      "loss": 0.8634,
+      "step": 1111
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766821532524113e-05,
+      "loss": 0.9425,
+      "step": 1112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766272997795032e-05,
+      "loss": 0.8593,
+      "step": 1113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765723834070804e-05,
+      "loss": 0.839,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765174041423905e-05,
+      "loss": 0.9054,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764623619926892e-05,
+      "loss": 0.8773,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764072569652396e-05,
+      "loss": 0.7686,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763520890673144e-05,
+      "loss": 0.8616,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762968583061938e-05,
+      "loss": 0.9328,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762415646891665e-05,
+      "loss": 0.8082,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761862082235293e-05,
+      "loss": 0.8771,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761307889165878e-05,
+      "loss": 0.8701,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760753067756555e-05,
+      "loss": 0.9346,
+      "step": 1123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76019761808054e-05,
+      "loss": 0.9555,
+      "step": 1124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759641540211137e-05,
+      "loss": 0.826,
+      "step": 1125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75908483422173e-05,
+      "loss": 0.8394,
+      "step": 1126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758527500185786e-05,
+      "loss": 0.8697,
+      "step": 1127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757969538176856e-05,
+      "loss": 0.8533,
+      "step": 1128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757410948268572e-05,
+      "loss": 0.8544,
+      "step": 1129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756851730534648e-05,
+      "loss": 0.8657,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75629188504889e-05,
+      "loss": 0.8331,
+      "step": 1131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75573141188517e-05,
+      "loss": 0.9218,
+      "step": 1132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75517031111746e-05,
+      "loss": 0.8262,
+      "step": 1133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754608582819804e-05,
+      "loss": 0.8038,
+      "step": 1134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754046227066332e-05,
+      "loss": 0.7566,
+      "step": 1135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753483243931258e-05,
+      "loss": 0.8038,
+      "step": 1136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752919633488878e-05,
+      "loss": 0.8956,
+      "step": 1137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752355395813569e-05,
+      "loss": 0.928,
+      "step": 1138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751790530979795e-05,
+      "loss": 0.7742,
+      "step": 1139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751225039062095e-05,
+      "loss": 0.8447,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750658920135101e-05,
+      "loss": 0.797,
+      "step": 1141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750092174273521e-05,
+      "loss": 0.8824,
+      "step": 1142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749524801552145e-05,
+      "loss": 0.8427,
+      "step": 1143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748956802045851e-05,
+      "loss": 0.8214,
+      "step": 1144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748388175829595e-05,
+      "loss": 0.8264,
+      "step": 1145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747818922978418e-05,
+      "loss": 0.7668,
+      "step": 1146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747249043567443e-05,
+      "loss": 0.8266,
+      "step": 1147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746678537671876e-05,
+      "loss": 0.8434,
+      "step": 1148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746107405367004e-05,
+      "loss": 0.8391,
+      "step": 1149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745535646728201e-05,
+      "loss": 0.9054,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74496326183092e-05,
+      "loss": 0.7702,
+      "step": 1151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744390250750694e-05,
+      "loss": 0.8085,
+      "step": 1152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743816613563147e-05,
+      "loss": 0.9388,
+      "step": 1153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743242350343977e-05,
+      "loss": 0.9181,
+      "step": 1154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742667461168971e-05,
+      "loss": 0.7772,
+      "step": 1155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742091946113994e-05,
+      "loss": 0.7553,
+      "step": 1156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741515805254997e-05,
+      "loss": 0.8748,
+      "step": 1157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74093903866801e-05,
+      "loss": 0.7515,
+      "step": 1158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740361646429151e-05,
+      "loss": 0.8981,
+      "step": 1159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739783628614614e-05,
+      "loss": 0.8884,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739204985300679e-05,
+      "loss": 0.9459,
+      "step": 1161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738625716563711e-05,
+      "loss": 0.8264,
+      "step": 1162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738045822480153e-05,
+      "loss": 0.9188,
+      "step": 1163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737465303126533e-05,
+      "loss": 0.6678,
+      "step": 1164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73688415857946e-05,
+      "loss": 0.9034,
+      "step": 1165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736302388915627e-05,
+      "loss": 0.7522,
+      "step": 1166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735719994211812e-05,
+      "loss": 0.8922,
+      "step": 1167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735136974544866e-05,
+      "loss": 0.9197,
+      "step": 1168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734553329991734e-05,
+      "loss": 0.6918,
+      "step": 1169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733969060629436e-05,
+      "loss": 0.9349,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733384166535078e-05,
+      "loss": 0.8053,
+      "step": 1171
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732798647785846e-05,
+      "loss": 0.9163,
+      "step": 1172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732212504459012e-05,
+      "loss": 0.8845,
+      "step": 1173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731625736631926e-05,
+      "loss": 0.796,
+      "step": 1174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731038344382024e-05,
+      "loss": 0.8752,
+      "step": 1175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73045032778682e-05,
+      "loss": 0.8621,
+      "step": 1176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729861686923916e-05,
+      "loss": 0.8371,
+      "step": 1177
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729272421870993e-05,
+      "loss": 0.8597,
+      "step": 1178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728682532705818e-05,
+      "loss": 0.7925,
+      "step": 1179
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728092019506232e-05,
+      "loss": 0.7986,
+      "step": 1180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727500882350167e-05,
+      "loss": 0.8751,
+      "step": 1181
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726909121315634e-05,
+      "loss": 0.8481,
+      "step": 1182
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726316736480726e-05,
+      "loss": 0.8327,
+      "step": 1183
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72572372792362e-05,
+      "loss": 0.8843,
+      "step": 1184
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72513009572257e-05,
+      "loss": 0.7863,
+      "step": 1185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724535839955921e-05,
+      "loss": 0.7387,
+      "step": 1186
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723940960702093e-05,
+      "loss": 0.8315,
+      "step": 1187
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723345458039594e-05,
+      "loss": 0.8593,
+      "step": 1188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72274933204701e-05,
+      "loss": 0.8169,
+      "step": 1189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722152582803008e-05,
+      "loss": 0.873,
+      "step": 1190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721555210386341e-05,
+      "loss": 0.9237,
+      "step": 1191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720957214875845e-05,
+      "loss": 0.7528,
+      "step": 1192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720358596350435e-05,
+      "loss": 0.8305,
+      "step": 1193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719759354889109e-05,
+      "loss": 0.8215,
+      "step": 1194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719159490570946e-05,
+      "loss": 0.7429,
+      "step": 1195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718559003475113e-05,
+      "loss": 0.8241,
+      "step": 1196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717957893680851e-05,
+      "loss": 0.8785,
+      "step": 1197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717356161267492e-05,
+      "loss": 0.8937,
+      "step": 1198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716753806314439e-05,
+      "loss": 0.9505,
+      "step": 1199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716150828901189e-05,
+      "loss": 0.8517,
+      "step": 1200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715547229107314e-05,
+      "loss": 0.8647,
+      "step": 1201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714943007012467e-05,
+      "loss": 0.8247,
+      "step": 1202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71433816269639e-05,
+      "loss": 0.8184,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713732696238901e-05,
+      "loss": 0.8873,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713126607719903e-05,
+      "loss": 0.9615,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712519897219379e-05,
+      "loss": 0.8459,
+      "step": 1206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711912564817396e-05,
+      "loss": 0.837,
+      "step": 1207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711304610594104e-05,
+      "loss": 0.8969,
+      "step": 1208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710696034629729e-05,
+      "loss": 0.9158,
+      "step": 1209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710086837004589e-05,
+      "loss": 0.7427,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709477017799076e-05,
+      "loss": 0.8788,
+      "step": 1211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708866577093665e-05,
+      "loss": 0.8513,
+      "step": 1212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708255514968917e-05,
+      "loss": 0.8667,
+      "step": 1213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707643831505472e-05,
+      "loss": 0.8333,
+      "step": 1214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707031526784051e-05,
+      "loss": 0.6511,
+      "step": 1215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706418600885463e-05,
+      "loss": 0.8584,
+      "step": 1216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705805053890588e-05,
+      "loss": 0.8421,
+      "step": 1217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.7051908858804e-05,
+      "loss": 0.872,
+      "step": 1218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704576096935947e-05,
+      "loss": 0.8675,
+      "step": 1219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703960687138362e-05,
+      "loss": 0.8065,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70334465656886e-05,
+      "loss": 0.9518,
+      "step": 1221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702728005308735e-05,
+      "loss": 0.8161,
+      "step": 1222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702110733439368e-05,
+      "loss": 0.9115,
+      "step": 1223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701492841042217e-05,
+      "loss": 0.8942,
+      "step": 1224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700874328198826e-05,
+      "loss": 0.8407,
+      "step": 1225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700255194990815e-05,
+      "loss": 0.78,
+      "step": 1226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699635441499894e-05,
+      "loss": 0.9034,
+      "step": 1227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699015067807851e-05,
+      "loss": 0.8277,
+      "step": 1228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69839407399655e-05,
+      "loss": 0.8363,
+      "step": 1229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697772460147948e-05,
+      "loss": 0.8454,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697150226344074e-05,
+      "loss": 0.7805,
+      "step": 1231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696527372667046e-05,
+      "loss": 0.8102,
+      "step": 1232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695903899199058e-05,
+      "loss": 0.7742,
+      "step": 1233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69527980602239e-05,
+      "loss": 0.8166,
+      "step": 1234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694655093219402e-05,
+      "loss": 0.8128,
+      "step": 1235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694029760872539e-05,
+      "loss": 0.8432,
+      "step": 1236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693403809064319e-05,
+      "loss": 0.8381,
+      "step": 1237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69277723787735e-05,
+      "loss": 0.951,
+      "step": 1238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692150047394322e-05,
+      "loss": 0.8285,
+      "step": 1239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691522237698e-05,
+      "loss": 0.9056,
+      "step": 1240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690893808871237e-05,
+      "loss": 0.8724,
+      "step": 1241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690264760996967e-05,
+      "loss": 0.92,
+      "step": 1242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6896350941582e-05,
+      "loss": 0.931,
+      "step": 1243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689004808438035e-05,
+      "loss": 0.8778,
+      "step": 1244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688373903919647e-05,
+      "loss": 0.9189,
+      "step": 1245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687742380686299e-05,
+      "loss": 0.9044,
+      "step": 1246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687110238821329e-05,
+      "loss": 0.821,
+      "step": 1247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686477478408158e-05,
+      "loss": 0.9188,
+      "step": 1248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685844099530294e-05,
+      "loss": 0.9104,
+      "step": 1249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68521010227132e-05,
+      "loss": 0.823,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684575486714904e-05,
+      "loss": 0.8089,
+      "step": 1251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683940252944794e-05,
+      "loss": 0.8648,
+      "step": 1252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683304401044823e-05,
+      "loss": 0.8756,
+      "step": 1253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6826679310989e-05,
+      "loss": 0.7648,
+      "step": 1254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682030843191022e-05,
+      "loss": 0.8266,
+      "step": 1255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681393137405259e-05,
+      "loss": 0.8381,
+      "step": 1256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680754813825773e-05,
+      "loss": 0.851,
+      "step": 1257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6801158725368e-05,
+      "loss": 0.8451,
+      "step": 1258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67947631362266e-05,
+      "loss": 0.8839,
+      "step": 1259
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678836137167753e-05,
+      "loss": 0.9257,
+      "step": 1260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678195343256564e-05,
+      "loss": 0.8951,
+      "step": 1261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677553931973657e-05,
+      "loss": 0.9125,
+      "step": 1262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676911903403675e-05,
+      "loss": 0.9018,
+      "step": 1263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676269257631347e-05,
+      "loss": 0.9692,
+      "step": 1264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675625994741484e-05,
+      "loss": 0.8005,
+      "step": 1265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67498211481897e-05,
+      "loss": 0.8479,
+      "step": 1266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674337617948785e-05,
+      "loss": 0.8553,
+      "step": 1267
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673692504215974e-05,
+      "loss": 0.8157,
+      "step": 1268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673046773705675e-05,
+      "loss": 0.75,
+      "step": 1269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672400426503103e-05,
+      "loss": 0.8567,
+      "step": 1270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671753462693555e-05,
+      "loss": 0.8404,
+      "step": 1271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671105882362412e-05,
+      "loss": 0.8001,
+      "step": 1272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670457685595129e-05,
+      "loss": 0.7501,
+      "step": 1273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669808872477251e-05,
+      "loss": 0.8622,
+      "step": 1274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669159443094399e-05,
+      "loss": 0.7627,
+      "step": 1275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668509397532278e-05,
+      "loss": 0.8885,
+      "step": 1276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66785873587667e-05,
+      "loss": 0.9732,
+      "step": 1277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667207458213445e-05,
+      "loss": 0.8592,
+      "step": 1278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66655556462855e-05,
+      "loss": 0.9087,
+      "step": 1279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665903055208014e-05,
+      "loss": 0.8526,
+      "step": 1280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665249930037945e-05,
+      "loss": 0.8632,
+      "step": 1281
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664596189204538e-05,
+      "loss": 0.7747,
+      "step": 1282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663941832794063e-05,
+      "loss": 0.8775,
+      "step": 1283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663286860892877e-05,
+      "loss": 0.8701,
+      "step": 1284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662631273587412e-05,
+      "loss": 0.8637,
+      "step": 1285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661975070964186e-05,
+      "loss": 0.8215,
+      "step": 1286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661318253109799e-05,
+      "loss": 0.7838,
+      "step": 1287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660660820110926e-05,
+      "loss": 0.7382,
+      "step": 1288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660002772054329e-05,
+      "loss": 0.9499,
+      "step": 1289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659344109026849e-05,
+      "loss": 0.8736,
+      "step": 1290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658684831115408e-05,
+      "loss": 0.8386,
+      "step": 1291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.65802493840701e-05,
+      "loss": 0.7815,
+      "step": 1292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65736443098874e-05,
+      "loss": 0.8274,
+      "step": 1293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656703308947764e-05,
+      "loss": 0.741,
+      "step": 1294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656041572371328e-05,
+      "loss": 0.8435,
+      "step": 1295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655379221346759e-05,
+      "loss": 0.8075,
+      "step": 1296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654716255961468e-05,
+      "loss": 0.8184,
+      "step": 1297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654052676302944e-05,
+      "loss": 0.9003,
+      "step": 1298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65338848245876e-05,
+      "loss": 0.9108,
+      "step": 1299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652723674516566e-05,
+      "loss": 0.8517,
+      "step": 1300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652058252564097e-05,
+      "loss": 0.8226,
+      "step": 1301
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651392216689165e-05,
+      "loss": 0.9091,
+      "step": 1302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65072556697967e-05,
+      "loss": 0.8501,
+      "step": 1303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650058303523584e-05,
+      "loss": 0.874,
+      "step": 1304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649390426408965e-05,
+      "loss": 0.8805,
+      "step": 1305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648721935723954e-05,
+      "loss": 0.8465,
+      "step": 1306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648052831556767e-05,
+      "loss": 0.9067,
+      "step": 1307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647383113995707e-05,
+      "loss": 0.8653,
+      "step": 1308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646712783129153e-05,
+      "loss": 0.8603,
+      "step": 1309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646041839045567e-05,
+      "loss": 0.8494,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645370281833496e-05,
+      "loss": 0.8694,
+      "step": 1311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64469811158156e-05,
+      "loss": 0.8168,
+      "step": 1312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644025328378466e-05,
+      "loss": 0.8395,
+      "step": 1313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643351932312997e-05,
+      "loss": 0.8931,
+      "step": 1314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642677923474024e-05,
+      "loss": 0.8296,
+      "step": 1315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642003301950491e-05,
+      "loss": 0.753,
+      "step": 1316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641328067831428e-05,
+      "loss": 0.8432,
+      "step": 1317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640652221205944e-05,
+      "loss": 0.8831,
+      "step": 1318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639975762163228e-05,
+      "loss": 0.8402,
+      "step": 1319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639298690792553e-05,
+      "loss": 0.8071,
+      "step": 1320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638621007183269e-05,
+      "loss": 0.9192,
+      "step": 1321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63794271142481e-05,
+      "loss": 0.901,
+      "step": 1322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637263803606687e-05,
+      "loss": 0.7825,
+      "step": 1323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636584283818496e-05,
+      "loss": 0.8414,
+      "step": 1324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63590415214991e-05,
+      "loss": 0.8738,
+      "step": 1325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635223408690688e-05,
+      "loss": 0.9218,
+      "step": 1326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634542053530662e-05,
+      "loss": 0.7877,
+      "step": 1327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633860086759754e-05,
+      "loss": 0.816,
+      "step": 1328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633177508467957e-05,
+      "loss": 0.7176,
+      "step": 1329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632494318745351e-05,
+      "loss": 0.8768,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631810517682096e-05,
+      "loss": 0.8331,
+      "step": 1331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63112610536843e-05,
+      "loss": 0.777,
+      "step": 1332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630441081894675e-05,
+      "loss": 0.8408,
+      "step": 1333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629755447351233e-05,
+      "loss": 0.7976,
+      "step": 1334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629069201828584e-05,
+      "loss": 0.8795,
+      "step": 1335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62838234541729e-05,
+      "loss": 0.8567,
+      "step": 1336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627694878207996e-05,
+      "loss": 0.8621,
+      "step": 1337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627006800291425e-05,
+      "loss": 0.7983,
+      "step": 1338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62631811175838e-05,
+      "loss": 0.8842,
+      "step": 1339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625628812699748e-05,
+      "loss": 0.8514,
+      "step": 1340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624938903206491e-05,
+      "loss": 0.793,
+      "step": 1341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624248383369657e-05,
+      "loss": 0.9279,
+      "step": 1342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623557253280373e-05,
+      "loss": 0.8642,
+      "step": 1343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622865513029847e-05,
+      "loss": 0.8785,
+      "step": 1344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622173162709362e-05,
+      "loss": 0.8412,
+      "step": 1345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62148020241029e-05,
+      "loss": 0.8727,
+      "step": 1346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620786632224079e-05,
+      "loss": 0.7754,
+      "step": 1347
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620092452242257e-05,
+      "loss": 0.9225,
+      "step": 1348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619397662556435e-05,
+      "loss": 0.8429,
+      "step": 1349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6187022632583e-05,
+      "loss": 0.9124,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618006254439625e-05,
+      "loss": 0.9055,
+      "step": 1351
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617309636192261e-05,
+      "loss": 0.8696,
+      "step": 1352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616612408608138e-05,
+      "loss": 0.873,
+      "step": 1353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615914571779268e-05,
+      "loss": 0.887,
+      "step": 1354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615216125797744e-05,
+      "loss": 0.8094,
+      "step": 1355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614517070755735e-05,
+      "loss": 0.8749,
+      "step": 1356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613817406745498e-05,
+      "loss": 0.8675,
+      "step": 1357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613117133859367e-05,
+      "loss": 0.8156,
+      "step": 1358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61241625218975e-05,
+      "loss": 0.8312,
+      "step": 1359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611714761829145e-05,
+      "loss": 0.7591,
+      "step": 1360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611012662870124e-05,
+      "loss": 0.8327,
+      "step": 1361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610309955405346e-05,
+      "loss": 0.8613,
+      "step": 1362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609606639527541e-05,
+      "loss": 0.8609,
+      "step": 1363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608902715329526e-05,
+      "loss": 0.8414,
+      "step": 1364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608198182904195e-05,
+      "loss": 0.8549,
+      "step": 1365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607493042344528e-05,
+      "loss": 0.817,
+      "step": 1366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606787293743576e-05,
+      "loss": 0.8291,
+      "step": 1367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606080937194478e-05,
+      "loss": 0.8222,
+      "step": 1368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605373972790448e-05,
+      "loss": 0.8184,
+      "step": 1369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604666400624785e-05,
+      "loss": 0.8472,
+      "step": 1370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603958220790865e-05,
+      "loss": 0.8201,
+      "step": 1371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603249433382144e-05,
+      "loss": 0.7178,
+      "step": 1372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60254003849216e-05,
+      "loss": 0.8437,
+      "step": 1373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601830036214531e-05,
+      "loss": 0.8629,
+      "step": 1374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601119426642956e-05,
+      "loss": 0.8341,
+      "step": 1375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60040820987121e-05,
+      "loss": 0.8565,
+      "step": 1376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59969638599315e-05,
+      "loss": 0.9131,
+      "step": 1377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598983955102716e-05,
+      "loss": 0.9389,
+      "step": 1378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598270917293926e-05,
+      "loss": 0.8187,
+      "step": 1379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597557272660878e-05,
+      "loss": 0.8867,
+      "step": 1380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596843021297748e-05,
+      "loss": 0.9127,
+      "step": 1381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.596128163298797e-05,
+      "loss": 0.8324,
+      "step": 1382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595412698758363e-05,
+      "loss": 0.8754,
+      "step": 1383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594696627770862e-05,
+      "loss": 0.8855,
+      "step": 1384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593979950430796e-05,
+      "loss": 0.873,
+      "step": 1385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593262666832741e-05,
+      "loss": 0.8414,
+      "step": 1386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592544777071355e-05,
+      "loss": 0.8992,
+      "step": 1387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591826281241378e-05,
+      "loss": 0.8181,
+      "step": 1388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591107179437628e-05,
+      "loss": 0.7939,
+      "step": 1389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590387471755004e-05,
+      "loss": 0.8696,
+      "step": 1390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589667158288482e-05,
+      "loss": 0.8963,
+      "step": 1391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588946239133123e-05,
+      "loss": 0.9409,
+      "step": 1392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588224714384064e-05,
+      "loss": 0.8252,
+      "step": 1393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587502584136523e-05,
+      "loss": 0.9012,
+      "step": 1394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586779848485797e-05,
+      "loss": 0.8964,
+      "step": 1395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586056507527266e-05,
+      "loss": 0.808,
+      "step": 1396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585332561356385e-05,
+      "loss": 0.878,
+      "step": 1397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584608010068696e-05,
+      "loss": 0.779,
+      "step": 1398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583882853759813e-05,
+      "loss": 0.8954,
+      "step": 1399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583157092525435e-05,
+      "loss": 0.8123,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582430726461338e-05,
+      "loss": 0.8713,
+      "step": 1401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581703755663381e-05,
+      "loss": 0.8034,
+      "step": 1402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580976180227498e-05,
+      "loss": 0.8926,
+      "step": 1403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580248000249708e-05,
+      "loss": 0.8381,
+      "step": 1404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579519215826105e-05,
+      "loss": 0.8556,
+      "step": 1405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578789827052868e-05,
+      "loss": 0.7856,
+      "step": 1406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578059834026252e-05,
+      "loss": 0.8433,
+      "step": 1407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57732923684259e-05,
+      "loss": 0.7706,
+      "step": 1408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576598035598301e-05,
+      "loss": 0.7941,
+      "step": 1409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575866230389879e-05,
+      "loss": 0.8849,
+      "step": 1410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575133821313899e-05,
+      "loss": 0.901,
+      "step": 1411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574400808467015e-05,
+      "loss": 0.8843,
+      "step": 1412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57366719194596e-05,
+      "loss": 0.9413,
+      "step": 1413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572932971847551e-05,
+      "loss": 0.8851,
+      "step": 1414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572198148268678e-05,
+      "loss": 0.9,
+      "step": 1415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571462721306314e-05,
+      "loss": 0.8226,
+      "step": 1416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570726691057517e-05,
+      "loss": 0.8104,
+      "step": 1417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569990057619414e-05,
+      "loss": 0.8579,
+      "step": 1418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56925282108922e-05,
+      "loss": 0.7653,
+      "step": 1419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568514981564225e-05,
+      "loss": 0.8295,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5677765391418e-05,
+      "loss": 0.8674,
+      "step": 1421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567037493919396e-05,
+      "loss": 0.8739,
+      "step": 1422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566297845994545e-05,
+      "loss": 0.8555,
+      "step": 1423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565557595464854e-05,
+      "loss": 0.9055,
+      "step": 1424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564816742428014e-05,
+      "loss": 0.7759,
+      "step": 1425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564075286981795e-05,
+      "loss": 0.7613,
+      "step": 1426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563333229224042e-05,
+      "loss": 0.8972,
+      "step": 1427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562590569252686e-05,
+      "loss": 0.7858,
+      "step": 1428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561847307165732e-05,
+      "loss": 0.9375,
+      "step": 1429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561103443061267e-05,
+      "loss": 0.8886,
+      "step": 1430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56035897703746e-05,
+      "loss": 0.8268,
+      "step": 1431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559613909192553e-05,
+      "loss": 0.8944,
+      "step": 1432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558868239624874e-05,
+      "loss": 0.9083,
+      "step": 1433
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558121968432824e-05,
+      "loss": 0.8428,
+      "step": 1434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557375095714891e-05,
+      "loss": 0.7938,
+      "step": 1435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556627621569637e-05,
+      "loss": 0.9037,
+      "step": 1436
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5558795460957e-05,
+      "loss": 0.853,
+      "step": 1437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555130869391809e-05,
+      "loss": 0.8316,
+      "step": 1438
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55438159155676e-05,
+      "loss": 0.8358,
+      "step": 1439
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553631712689437e-05,
+      "loss": 0.8138,
+      "step": 1440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552881232888799e-05,
+      "loss": 0.8977,
+      "step": 1441
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552130152253884e-05,
+      "loss": 0.8974,
+      "step": 1442
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551378470883812e-05,
+      "loss": 0.754,
+      "step": 1443
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55062618887778e-05,
+      "loss": 0.7951,
+      "step": 1444
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549873306335065e-05,
+      "loss": 0.7601,
+      "step": 1445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549119823355023e-05,
+      "loss": 0.8126,
+      "step": 1446
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548365740037092e-05,
+      "loss": 0.7748,
+      "step": 1447
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547611056480786e-05,
+      "loss": 0.8321,
+      "step": 1448
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546855772785695e-05,
+      "loss": 0.8885,
+      "step": 1449
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546099889051498e-05,
+      "loss": 0.8323,
+      "step": 1450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545343405377943e-05,
+      "loss": 0.8363,
+      "step": 1451
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544586321864864e-05,
+      "loss": 0.8141,
+      "step": 1452
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543828638612172e-05,
+      "loss": 0.8337,
+      "step": 1453
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543070355719856e-05,
+      "loss": 0.8319,
+      "step": 1454
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542311473287984e-05,
+      "loss": 0.7937,
+      "step": 1455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541551991416704e-05,
+      "loss": 0.8404,
+      "step": 1456
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540791910206246e-05,
+      "loss": 0.8985,
+      "step": 1457
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540031229756915e-05,
+      "loss": 0.8152,
+      "step": 1458
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539269950169096e-05,
+      "loss": 0.833,
+      "step": 1459
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538508071543253e-05,
+      "loss": 0.9,
+      "step": 1460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53774559397993e-05,
+      "loss": 0.8046,
+      "step": 1461
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536982517579751e-05,
+      "loss": 0.8488,
+      "step": 1462
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536218842443414e-05,
+      "loss": 0.9576,
+      "step": 1463
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535454568671704e-05,
+      "loss": 0.8373,
+      "step": 1464
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534689696365478e-05,
+      "loss": 0.8715,
+      "step": 1465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533924225625675e-05,
+      "loss": 0.7602,
+      "step": 1466
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533158156553313e-05,
+      "loss": 0.9015,
+      "step": 1467
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532391489249487e-05,
+      "loss": 0.8514,
+      "step": 1468
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531624223815376e-05,
+      "loss": 0.8383,
+      "step": 1469
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530856360352232e-05,
+      "loss": 0.8233,
+      "step": 1470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530087898961388e-05,
+      "loss": 0.9198,
+      "step": 1471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529318839744257e-05,
+      "loss": 0.959,
+      "step": 1472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528549182802329e-05,
+      "loss": 0.8215,
+      "step": 1473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527778928237176e-05,
+      "loss": 0.9057,
+      "step": 1474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527008076150446e-05,
+      "loss": 0.7794,
+      "step": 1475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526236626643866e-05,
+      "loss": 0.8215,
+      "step": 1476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525464579819244e-05,
+      "loss": 0.886,
+      "step": 1477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524691935778463e-05,
+      "loss": 0.8875,
+      "step": 1478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52391869462349e-05,
+      "loss": 0.831,
+      "step": 1479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523144856456368e-05,
+      "loss": 0.825,
+      "step": 1480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522370421379216e-05,
+      "loss": 0.838,
+      "step": 1481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521595389494238e-05,
+      "loss": 0.842,
+      "step": 1482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52081976090371e-05,
+      "loss": 0.8333,
+      "step": 1483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520043535709992e-05,
+      "loss": 0.8097,
+      "step": 1484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519266714015522e-05,
+      "loss": 0.8942,
+      "step": 1485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518489295922813e-05,
+      "loss": 0.9107,
+      "step": 1486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51771128153446e-05,
+      "loss": 0.8909,
+      "step": 1487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516932670953137e-05,
+      "loss": 0.8829,
+      "step": 1488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516153464281595e-05,
+      "loss": 0.7745,
+      "step": 1489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515373661622664e-05,
+      "loss": 0.8703,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514593263079254e-05,
+      "loss": 0.8407,
+      "step": 1491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51381226875435e-05,
+      "loss": 0.8293,
+      "step": 1492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513030678751021e-05,
+      "loss": 0.9536,
+      "step": 1493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51224849317241e-05,
+      "loss": 0.8121,
+      "step": 1494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511465712121742e-05,
+      "loss": 0.8632,
+      "step": 1495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510682335702319e-05,
+      "loss": 0.8134,
+      "step": 1496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509898364017519e-05,
+      "loss": 0.8426,
+      "step": 1497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509113797170803e-05,
+      "loss": 0.846,
+      "step": 1498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508328635265709e-05,
+      "loss": 0.8889,
+      "step": 1499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507542878405853e-05,
+      "loss": 0.8792,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506756526694929e-05,
+      "loss": 0.865,
+      "step": 1501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50596958023671e-05,
+      "loss": 0.9142,
+      "step": 1502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505182039135047e-05,
+      "loss": 0.8065,
+      "step": 1503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504393903493875e-05,
+      "loss": 0.8852,
+      "step": 1504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503605173417196e-05,
+      "loss": 0.8033,
+      "step": 1505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502815849009102e-05,
+      "loss": 0.7831,
+      "step": 1506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502025930373755e-05,
+      "loss": 0.8765,
+      "step": 1507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501235417615401e-05,
+      "loss": 0.7751,
+      "step": 1508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500444310838363e-05,
+      "loss": 0.8631,
+      "step": 1509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49965261014704e-05,
+      "loss": 0.7979,
+      "step": 1510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498860315645911e-05,
+      "loss": 0.8653,
+      "step": 1511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498067427439534e-05,
+      "loss": 0.9172,
+      "step": 1512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497273945632546e-05,
+      "loss": 0.8084,
+      "step": 1513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49647987032966e-05,
+      "loss": 0.8902,
+      "step": 1514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495685201635667e-05,
+      "loss": 0.7422,
+      "step": 1515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49488993965544e-05,
+      "loss": 0.9119,
+      "step": 1516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494094084493929e-05,
+      "loss": 0.8761,
+      "step": 1517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493297636256158e-05,
+      "loss": 0.8695,
+      "step": 1518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.492500595047235e-05,
+      "loss": 0.9375,
+      "step": 1519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491702960972342e-05,
+      "loss": 0.8321,
+      "step": 1520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490904734136743e-05,
+      "loss": 0.7879,
+      "step": 1521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49010591464578e-05,
+      "loss": 0.838,
+      "step": 1522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.489306502604866e-05,
+      "loss": 0.8412,
+      "step": 1523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488506498119502e-05,
+      "loss": 0.8564,
+      "step": 1524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487705901295263e-05,
+      "loss": 0.8482,
+      "step": 1525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486904712237801e-05,
+      "loss": 0.8642,
+      "step": 1526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486102931052845e-05,
+      "loss": 0.7994,
+      "step": 1527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485300557846208e-05,
+      "loss": 0.909,
+      "step": 1528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484497592723776e-05,
+      "loss": 0.7937,
+      "step": 1529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483694035791516e-05,
+      "loss": 0.9169,
+      "step": 1530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48288988715547e-05,
+      "loss": 0.9009,
+      "step": 1531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482085146921759e-05,
+      "loss": 0.792,
+      "step": 1532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481279815196587e-05,
+      "loss": 0.8277,
+      "step": 1533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480473892086226e-05,
+      "loss": 0.8579,
+      "step": 1534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479667377697037e-05,
+      "loss": 0.7804,
+      "step": 1535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478860272135453e-05,
+      "loss": 0.8344,
+      "step": 1536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478052575507982e-05,
+      "loss": 0.8348,
+      "step": 1537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47724428792122e-05,
+      "loss": 0.7992,
+      "step": 1538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476435409481834e-05,
+      "loss": 0.8489,
+      "step": 1539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475625940296566e-05,
+      "loss": 0.7701,
+      "step": 1540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474815880472243e-05,
+      "loss": 0.7477,
+      "step": 1541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474005230115766e-05,
+      "loss": 0.9647,
+      "step": 1542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473193989334115e-05,
+      "loss": 0.901,
+      "step": 1543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.472382158234348e-05,
+      "loss": 0.8328,
+      "step": 1544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4715697369236e-05,
+      "loss": 0.7881,
+      "step": 1545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470756725509087e-05,
+      "loss": 0.8805,
+      "step": 1546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469943124098096e-05,
+      "loss": 0.7827,
+      "step": 1547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469128932798001e-05,
+      "loss": 0.8159,
+      "step": 1548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468314151716245e-05,
+      "loss": 0.7428,
+      "step": 1549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467498780960356e-05,
+      "loss": 0.8109,
+      "step": 1550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466682820637935e-05,
+      "loss": 0.7533,
+      "step": 1551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465866270856665e-05,
+      "loss": 0.7169,
+      "step": 1552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465049131724302e-05,
+      "loss": 0.7596,
+      "step": 1553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464231403348682e-05,
+      "loss": 0.8924,
+      "step": 1554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46341308583772e-05,
+      "loss": 0.8002,
+      "step": 1555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462594179299406e-05,
+      "loss": 0.8526,
+      "step": 1556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461774683841813e-05,
+      "loss": 0.8606,
+      "step": 1557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.460954599573085e-05,
+      "loss": 0.8403,
+      "step": 1558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.460133926601448e-05,
+      "loss": 0.8863,
+      "step": 1559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459312665035202e-05,
+      "loss": 0.7846,
+      "step": 1560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458490814982733e-05,
+      "loss": 0.8271,
+      "step": 1561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457668376552494e-05,
+      "loss": 0.931,
+      "step": 1562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456845349853021e-05,
+      "loss": 0.8224,
+      "step": 1563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456021734992928e-05,
+      "loss": 0.8937,
+      "step": 1564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455197532080906e-05,
+      "loss": 0.8157,
+      "step": 1565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454372741225723e-05,
+      "loss": 0.9543,
+      "step": 1566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453547362536224e-05,
+      "loss": 0.8671,
+      "step": 1567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452721396121333e-05,
+      "loss": 0.8742,
+      "step": 1568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451894842090054e-05,
+      "loss": 0.8807,
+      "step": 1569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451067700551461e-05,
+      "loss": 0.9028,
+      "step": 1570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450239971614711e-05,
+      "loss": 0.8882,
+      "step": 1571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449411655389042e-05,
+      "loss": 0.9101,
+      "step": 1572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448582751983761e-05,
+      "loss": 0.8873,
+      "step": 1573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447753261508257e-05,
+      "loss": 0.8653,
+      "step": 1574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446923184071997e-05,
+      "loss": 0.9017,
+      "step": 1575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446092519784526e-05,
+      "loss": 0.8114,
+      "step": 1576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44526126875546e-05,
+      "loss": 0.8826,
+      "step": 1577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444429431094503e-05,
+      "loss": 0.8278,
+      "step": 1578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443597006911431e-05,
+      "loss": 0.7186,
+      "step": 1579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442763996316092e-05,
+      "loss": 0.861,
+      "step": 1580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441930399418421e-05,
+      "loss": 0.8909,
+      "step": 1581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441096216328425e-05,
+      "loss": 0.8766,
+      "step": 1582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.440261447156188e-05,
+      "loss": 0.8517,
+      "step": 1583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439426092011875e-05,
+      "loss": 0.8193,
+      "step": 1584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438590151005726e-05,
+      "loss": 0.8311,
+      "step": 1585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437753624248058e-05,
+      "loss": 0.8443,
+      "step": 1586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436916511849263e-05,
+      "loss": 0.9542,
+      "step": 1587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436078813919818e-05,
+      "loss": 0.8838,
+      "step": 1588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435240530570266e-05,
+      "loss": 0.835,
+      "step": 1589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434401661911238e-05,
+      "loss": 0.8608,
+      "step": 1590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43356220805344e-05,
+      "loss": 0.8008,
+      "step": 1591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432722169107646e-05,
+      "loss": 0.7029,
+      "step": 1592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431881545184719e-05,
+      "loss": 0.9407,
+      "step": 1593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431040336395592e-05,
+      "loss": 0.8497,
+      "step": 1594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43019854285128e-05,
+      "loss": 0.8087,
+      "step": 1595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429356164662873e-05,
+      "loss": 0.9279,
+      "step": 1596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428513201941535e-05,
+      "loss": 0.813,
+      "step": 1597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427669654798512e-05,
+      "loss": 0.8412,
+      "step": 1598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.426825523345124e-05,
+      "loss": 0.7864,
+      "step": 1599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42598080769277e-05,
+      "loss": 0.8272,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425135507952926e-05,
+      "loss": 0.8792,
+      "step": 1601
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.424289624237144e-05,
+      "loss": 0.8648,
+      "step": 1602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423443156657052e-05,
+      "loss": 0.8294,
+      "step": 1603
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422596105324359e-05,
+      "loss": 0.8799,
+      "step": 1604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421748470350847e-05,
+      "loss": 0.8723,
+      "step": 1605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420900251848377e-05,
+      "loss": 0.8016,
+      "step": 1606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420051449928889e-05,
+      "loss": 0.8473,
+      "step": 1607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419202064704393e-05,
+      "loss": 0.9015,
+      "step": 1608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418352096286982e-05,
+      "loss": 0.8234,
+      "step": 1609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417501544788827e-05,
+      "loss": 0.8764,
+      "step": 1610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416650410322172e-05,
+      "loss": 0.7855,
+      "step": 1611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41579869299934e-05,
+      "loss": 0.8173,
+      "step": 1612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41494639293273e-05,
+      "loss": 0.7977,
+      "step": 1613
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414093510234818e-05,
+      "loss": 0.9009,
+      "step": 1614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413240045018157e-05,
+      "loss": 0.8461,
+      "step": 1615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412385997395376e-05,
+      "loss": 0.898,
+      "step": 1616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411531367479185e-05,
+      "loss": 0.8961,
+      "step": 1617
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410676155382365e-05,
+      "loss": 0.8688,
+      "step": 1618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409820361217778e-05,
+      "loss": 0.8116,
+      "step": 1619
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40896398509836e-05,
+      "loss": 0.9212,
+      "step": 1620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408107027137128e-05,
+      "loss": 0.8843,
+      "step": 1621
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40724948744717e-05,
+      "loss": 0.8144,
+      "step": 1622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406391366141654e-05,
+      "loss": 0.7723,
+      "step": 1623
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405532663333826e-05,
+      "loss": 0.8202,
+      "step": 1624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404673379137006e-05,
+      "loss": 0.8257,
+      "step": 1625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403813513664592e-05,
+      "loss": 0.8275,
+      "step": 1626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402953067030061e-05,
+      "loss": 0.8036,
+      "step": 1627
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402092039346961e-05,
+      "loss": 0.819,
+      "step": 1628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401230430728923e-05,
+      "loss": 0.7916,
+      "step": 1629
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400368241289648e-05,
+      "loss": 0.7364,
+      "step": 1630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39950547114292e-05,
+      "loss": 0.8809,
+      "step": 1631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398642120402597e-05,
+      "loss": 0.939,
+      "step": 1632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397778189182613e-05,
+      "loss": 0.9132,
+      "step": 1633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396913677596979e-05,
+      "loss": 0.695,
+      "step": 1634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396048585759785e-05,
+      "loss": 0.7853,
+      "step": 1635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39518291378519e-05,
+      "loss": 0.6758,
+      "step": 1636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394316661787441e-05,
+      "loss": 0.8044,
+      "step": 1637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393449829880853e-05,
+      "loss": 0.9012,
+      "step": 1638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392582418179819e-05,
+      "loss": 0.8685,
+      "step": 1639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39171442679881e-05,
+      "loss": 0.8115,
+      "step": 1640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390845855852375e-05,
+      "loss": 0.7294,
+      "step": 1641
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.389976705455135e-05,
+      "loss": 0.847,
+      "step": 1642
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38910697572179e-05,
+      "loss": 0.791,
+      "step": 1643
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388236666767118e-05,
+      "loss": 0.8418,
+      "step": 1644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387365778705973e-05,
+      "loss": 0.7939,
+      "step": 1645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38649431165328e-05,
+      "loss": 0.8669,
+      "step": 1646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.385622265724048e-05,
+      "loss": 0.8968,
+      "step": 1647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.384749641033359e-05,
+      "loss": 0.728,
+      "step": 1648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.383876437696368e-05,
+      "loss": 0.7725,
+      "step": 1649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.383002655828316e-05,
+      "loss": 0.7434,
+      "step": 1650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.38212829554451e-05,
+      "loss": 0.8233,
+      "step": 1651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381253356960338e-05,
+      "loss": 0.8465,
+      "step": 1652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380377840191264e-05,
+      "loss": 0.844,
+      "step": 1653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379501745352827e-05,
+      "loss": 0.8533,
+      "step": 1654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378625072560645e-05,
+      "loss": 0.8855,
+      "step": 1655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377747821930412e-05,
+      "loss": 0.8222,
+      "step": 1656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376869993577893e-05,
+      "loss": 0.8638,
+      "step": 1657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375991587618938e-05,
+      "loss": 0.7985,
+      "step": 1658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375112604169464e-05,
+      "loss": 0.7779,
+      "step": 1659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374233043345471e-05,
+      "loss": 0.8013,
+      "step": 1660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373352905263031e-05,
+      "loss": 0.808,
+      "step": 1661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372472190038298e-05,
+      "loss": 0.8624,
+      "step": 1662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371590897787492e-05,
+      "loss": 0.8675,
+      "step": 1663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370709028626921e-05,
+      "loss": 0.7624,
+      "step": 1664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369826582672962e-05,
+      "loss": 0.8345,
+      "step": 1665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368943560042067e-05,
+      "loss": 0.7808,
+      "step": 1666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368059960850769e-05,
+      "loss": 0.9115,
+      "step": 1667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367175785215674e-05,
+      "loss": 0.8654,
+      "step": 1668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.366291033253465e-05,
+      "loss": 0.8429,
+      "step": 1669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365405705080902e-05,
+      "loss": 0.7893,
+      "step": 1670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364519800814817e-05,
+      "loss": 0.8469,
+      "step": 1671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363633320572124e-05,
+      "loss": 0.7636,
+      "step": 1672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362746264469808e-05,
+      "loss": 0.8484,
+      "step": 1673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361858632624934e-05,
+      "loss": 0.7923,
+      "step": 1674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360970425154638e-05,
+      "loss": 0.8036,
+      "step": 1675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360081642176136e-05,
+      "loss": 0.8765,
+      "step": 1676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35919228380672e-05,
+      "loss": 0.9486,
+      "step": 1677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358302350163757e-05,
+      "loss": 0.8867,
+      "step": 1678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357411841364687e-05,
+      "loss": 0.8612,
+      "step": 1679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356520757527032e-05,
+      "loss": 0.8237,
+      "step": 1680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355629098768384e-05,
+      "loss": 0.8136,
+      "step": 1681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.354736865206415e-05,
+      "loss": 0.9517,
+      "step": 1682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353844056958869e-05,
+      "loss": 0.8352,
+      "step": 1683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35295067414357e-05,
+      "loss": 0.7465,
+      "step": 1684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352056716878416e-05,
+      "loss": 0.8937,
+      "step": 1685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35116218528138e-05,
+      "loss": 0.7953,
+      "step": 1686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35026707947051e-05,
+      "loss": 0.8162,
+      "step": 1687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.349371399563934e-05,
+      "loss": 0.8916,
+      "step": 1688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348475145679851e-05,
+      "loss": 0.8466,
+      "step": 1689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34757831793654e-05,
+      "loss": 0.8889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34668091645235e-05,
+      "loss": 0.8406,
+      "step": 1691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345782941345713e-05,
+      "loss": 0.7887,
+      "step": 1692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344884392735132e-05,
+      "loss": 0.88,
+      "step": 1693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343985270739182e-05,
+      "loss": 0.8438,
+      "step": 1694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343085575476525e-05,
+      "loss": 0.8208,
+      "step": 1695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342185307065888e-05,
+      "loss": 0.8136,
+      "step": 1696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34128446562608e-05,
+      "loss": 0.7177,
+      "step": 1697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340383051275979e-05,
+      "loss": 0.8919,
+      "step": 1698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.339481064134547e-05,
+      "loss": 0.8539,
+      "step": 1699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338578504320814e-05,
+      "loss": 0.8663,
+      "step": 1700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337675371953894e-05,
+      "loss": 0.7809,
+      "step": 1701
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.336771667152965e-05,
+      "loss": 0.8938,
+      "step": 1702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335867390037292e-05,
+      "loss": 0.8167,
+      "step": 1703
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334962540726207e-05,
+      "loss": 0.8493,
+      "step": 1704
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334057119339125e-05,
+      "loss": 0.8481,
+      "step": 1705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.33315112599553e-05,
+      "loss": 0.9344,
+      "step": 1706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332244560814984e-05,
+      "loss": 0.844,
+      "step": 1707
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331337423917126e-05,
+      "loss": 0.7812,
+      "step": 1708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330429715421667e-05,
+      "loss": 0.829,
+      "step": 1709
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329521435448396e-05,
+      "loss": 0.8462,
+      "step": 1710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32861258411718e-05,
+      "loss": 0.8645,
+      "step": 1711
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327703161547952e-05,
+      "loss": 0.7626,
+      "step": 1712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326793167860732e-05,
+      "loss": 0.8221,
+      "step": 1713
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325882603175609e-05,
+      "loss": 0.8309,
+      "step": 1714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324971467612746e-05,
+      "loss": 0.7591,
+      "step": 1715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324059761292385e-05,
+      "loss": 0.8784,
+      "step": 1716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323147484334843e-05,
+      "loss": 0.7931,
+      "step": 1717
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32223463686051e-05,
+      "loss": 0.8019,
+      "step": 1718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321321218989851e-05,
+      "loss": 0.7947,
+      "step": 1719
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320407230843413e-05,
+      "loss": 0.8396,
+      "step": 1720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319492672541807e-05,
+      "loss": 0.9024,
+      "step": 1721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318577544205731e-05,
+      "loss": 0.7661,
+      "step": 1722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317661845955947e-05,
+      "loss": 0.8594,
+      "step": 1723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.316745577913303e-05,
+      "loss": 0.8295,
+      "step": 1724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315828740198714e-05,
+      "loss": 0.7891,
+      "step": 1725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314911332933174e-05,
+      "loss": 0.8416,
+      "step": 1726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.313993356237751e-05,
+      "loss": 0.7281,
+      "step": 1727
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.313074810233588e-05,
+      "loss": 0.8915,
+      "step": 1728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312155695041904e-05,
+      "loss": 0.8051,
+      "step": 1729
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311236010783995e-05,
+      "loss": 0.8383,
+      "step": 1730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.310315757581225e-05,
+      "loss": 0.8832,
+      "step": 1731
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309394935555043e-05,
+      "loss": 0.7917,
+      "step": 1732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308473544826966e-05,
+      "loss": 0.8706,
+      "step": 1733
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307551585518586e-05,
+      "loss": 0.735,
+      "step": 1734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.306629057751574e-05,
+      "loss": 0.728,
+      "step": 1735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305705961647673e-05,
+      "loss": 0.8119,
+      "step": 1736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.304782297328704e-05,
+      "loss": 0.7524,
+      "step": 1737
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30385806491656e-05,
+      "loss": 0.8728,
+      "step": 1738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302933264533209e-05,
+      "loss": 0.8452,
+      "step": 1739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302007896300698e-05,
+      "loss": 0.8238,
+      "step": 1740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301081960341143e-05,
+      "loss": 0.7667,
+      "step": 1741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300155456776739e-05,
+      "loss": 0.8631,
+      "step": 1742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299228385729752e-05,
+      "loss": 0.837,
+      "step": 1743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298300747322529e-05,
+      "loss": 0.7928,
+      "step": 1744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297372541677489e-05,
+      "loss": 0.8074,
+      "step": 1745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.296443768917123e-05,
+      "loss": 0.8591,
+      "step": 1746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295514429164e-05,
+      "loss": 0.8104,
+      "step": 1747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294584522540766e-05,
+      "loss": 0.8636,
+      "step": 1748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293654049170133e-05,
+      "loss": 0.85,
+      "step": 1749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.292723009174897e-05,
+      "loss": 0.7515,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291791402677926e-05,
+      "loss": 0.865,
+      "step": 1751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290859229802161e-05,
+      "loss": 0.6854,
+      "step": 1752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28992649067062e-05,
+      "loss": 0.8532,
+      "step": 1753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288993185406392e-05,
+      "loss": 0.7161,
+      "step": 1754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288059314132647e-05,
+      "loss": 0.843,
+      "step": 1755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.287124876972624e-05,
+      "loss": 0.8398,
+      "step": 1756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286189874049638e-05,
+      "loss": 0.8633,
+      "step": 1757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28525430548708e-05,
+      "loss": 0.836,
+      "step": 1758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284318171408419e-05,
+      "loss": 0.8022,
+      "step": 1759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283381471937187e-05,
+      "loss": 0.7464,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282444207197006e-05,
+      "loss": 0.8413,
+      "step": 1761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281506377311557e-05,
+      "loss": 0.8658,
+      "step": 1762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280567982404611e-05,
+      "loss": 0.7794,
+      "step": 1763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279629022600001e-05,
+      "loss": 0.8835,
+      "step": 1764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.278689498021642e-05,
+      "loss": 0.9211,
+      "step": 1765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277749408793521e-05,
+      "loss": 0.7438,
+      "step": 1766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.276808755039699e-05,
+      "loss": 0.8782,
+      "step": 1767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275867536884311e-05,
+      "loss": 0.7985,
+      "step": 1768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27492575445157e-05,
+      "loss": 0.8669,
+      "step": 1769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273983407865759e-05,
+      "loss": 0.8973,
+      "step": 1770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27304049725124e-05,
+      "loss": 0.8688,
+      "step": 1771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272097022732443e-05,
+      "loss": 0.7957,
+      "step": 1772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27115298443388e-05,
+      "loss": 0.7703,
+      "step": 1773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270208382480133e-05,
+      "loss": 0.7533,
+      "step": 1774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269263216995857e-05,
+      "loss": 0.6581,
+      "step": 1775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268317488105787e-05,
+      "loss": 0.8486,
+      "step": 1776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267371195934726e-05,
+      "loss": 0.8721,
+      "step": 1777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266424340607555e-05,
+      "loss": 0.7921,
+      "step": 1778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265476922249229e-05,
+      "loss": 0.8691,
+      "step": 1779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.264528940984776e-05,
+      "loss": 0.8682,
+      "step": 1780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263580396939301e-05,
+      "loss": 0.8503,
+      "step": 1781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.262631290237979e-05,
+      "loss": 0.7854,
+      "step": 1782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261681621006065e-05,
+      "loss": 0.874,
+      "step": 1783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26073138936888e-05,
+      "loss": 0.8556,
+      "step": 1784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259780595451827e-05,
+      "loss": 0.8215,
+      "step": 1785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25882923938038e-05,
+      "loss": 0.7571,
+      "step": 1786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25787732128009e-05,
+      "loss": 0.8715,
+      "step": 1787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256924841276575e-05,
+      "loss": 0.8521,
+      "step": 1788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.255971799495534e-05,
+      "loss": 0.75,
+      "step": 1789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.255018196062738e-05,
+      "loss": 0.7557,
+      "step": 1790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.254064031104033e-05,
+      "loss": 0.883,
+      "step": 1791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253109304745335e-05,
+      "loss": 0.8302,
+      "step": 1792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252154017112642e-05,
+      "loss": 0.8627,
+      "step": 1793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251198168332017e-05,
+      "loss": 0.8028,
+      "step": 1794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.250241758529603e-05,
+      "loss": 0.9048,
+      "step": 1795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249284787831616e-05,
+      "loss": 0.826,
+      "step": 1796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248327256364346e-05,
+      "loss": 0.887,
+      "step": 1797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247369164254153e-05,
+      "loss": 0.9126,
+      "step": 1798
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.246410511627479e-05,
+      "loss": 0.9024,
+      "step": 1799
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245451298610832e-05,
+      "loss": 0.8544,
+      "step": 1800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2444915253308e-05,
+      "loss": 0.8294,
+      "step": 1801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.243531191914038e-05,
+      "loss": 0.8096,
+      "step": 1802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.242570298487285e-05,
+      "loss": 0.7701,
+      "step": 1803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241608845177344e-05,
+      "loss": 0.8205,
+      "step": 1804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240646832111097e-05,
+      "loss": 0.9765,
+      "step": 1805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2396842594155e-05,
+      "loss": 0.8618,
+      "step": 1806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.238721127217579e-05,
+      "loss": 0.8631,
+      "step": 1807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237757435644439e-05,
+      "loss": 0.8462,
+      "step": 1808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236793184823256e-05,
+      "loss": 0.8424,
+      "step": 1809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.235828374881279e-05,
+      "loss": 0.8821,
+      "step": 1810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.234863005945835e-05,
+      "loss": 0.7903,
+      "step": 1811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233897078144317e-05,
+      "loss": 0.8014,
+      "step": 1812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2329305916042e-05,
+      "loss": 0.6881,
+      "step": 1813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231963546453029e-05,
+      "loss": 0.8264,
+      "step": 1814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.230995942818421e-05,
+      "loss": 0.8453,
+      "step": 1815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23002778082807e-05,
+      "loss": 0.9061,
+      "step": 1816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229059060609742e-05,
+      "loss": 0.9066,
+      "step": 1817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228089782291278e-05,
+      "loss": 0.7726,
+      "step": 1818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22711994600059e-05,
+      "loss": 0.7735,
+      "step": 1819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226149551865665e-05,
+      "loss": 0.9277,
+      "step": 1820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.225178600014567e-05,
+      "loss": 0.7973,
+      "step": 1821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224207090575428e-05,
+      "loss": 0.8916,
+      "step": 1822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223235023676457e-05,
+      "loss": 0.8068,
+      "step": 1823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222262399445933e-05,
+      "loss": 0.7543,
+      "step": 1824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.221289218012216e-05,
+      "loss": 0.9635,
+      "step": 1825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22031547950373e-05,
+      "loss": 0.8536,
+      "step": 1826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.21934118404898e-05,
+      "loss": 0.8936,
+      "step": 1827
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.218366331776543e-05,
+      "loss": 0.8683,
+      "step": 1828
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217390922815063e-05,
+      "loss": 0.8354,
+      "step": 1829
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216414957293269e-05,
+      "loss": 0.8876,
+      "step": 1830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.215438435339954e-05,
+      "loss": 0.9451,
+      "step": 1831
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214461357083985e-05,
+      "loss": 0.8478,
+      "step": 1832
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213483722654311e-05,
+      "loss": 0.9138,
+      "step": 1833
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212505532179943e-05,
+      "loss": 0.8267,
+      "step": 1834
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.211526785789973e-05,
+      "loss": 0.8814,
+      "step": 1835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210547483613565e-05,
+      "loss": 0.7742,
+      "step": 1836
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209567625779953e-05,
+      "loss": 0.9566,
+      "step": 1837
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20858721241845e-05,
+      "loss": 0.8309,
+      "step": 1838
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.207606243658438e-05,
+      "loss": 0.8735,
+      "step": 1839
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20662471962937e-05,
+      "loss": 0.8407,
+      "step": 1840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20564264046078e-05,
+      "loss": 0.7853,
+      "step": 1841
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20466000628227e-05,
+      "loss": 0.8185,
+      "step": 1842
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203676817223516e-05,
+      "loss": 0.8749,
+      "step": 1843
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.202693073414266e-05,
+      "loss": 0.8328,
+      "step": 1844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.201708774984344e-05,
+      "loss": 0.9381,
+      "step": 1845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.200723922063645e-05,
+      "loss": 0.8537,
+      "step": 1846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199738514782138e-05,
+      "loss": 0.8125,
+      "step": 1847
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198752553269867e-05,
+      "loss": 0.8474,
+      "step": 1848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197766037656946e-05,
+      "loss": 0.7337,
+      "step": 1849
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.196778968073563e-05,
+      "loss": 0.7181,
+      "step": 1850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19579134464998e-05,
+      "loss": 0.8338,
+      "step": 1851
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19480316751653e-05,
+      "loss": 0.9152,
+      "step": 1852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193814436803624e-05,
+      "loss": 0.7315,
+      "step": 1853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19282515264174e-05,
+      "loss": 0.8379,
+      "step": 1854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191835315161433e-05,
+      "loss": 0.9267,
+      "step": 1855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.190844924493329e-05,
+      "loss": 0.8934,
+      "step": 1856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189853980768128e-05,
+      "loss": 0.6947,
+      "step": 1857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188862484116604e-05,
+      "loss": 0.8787,
+      "step": 1858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.187870434669602e-05,
+      "loss": 0.8064,
+      "step": 1859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.18687783255804e-05,
+      "loss": 0.7682,
+      "step": 1860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.18588467791291e-05,
+      "loss": 0.7366,
+      "step": 1861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.184890970865277e-05,
+      "loss": 0.8341,
+      "step": 1862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183896711546278e-05,
+      "loss": 0.854,
+      "step": 1863
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.182901900087125e-05,
+      "loss": 0.8394,
+      "step": 1864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181906536619099e-05,
+      "loss": 0.8026,
+      "step": 1865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.180910621273555e-05,
+      "loss": 0.8492,
+      "step": 1866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.179914154181925e-05,
+      "loss": 0.8442,
+      "step": 1867
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.178917135475708e-05,
+      "loss": 0.8875,
+      "step": 1868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.177919565286482e-05,
+      "loss": 0.8035,
+      "step": 1869
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17692144374589e-05,
+      "loss": 0.9382,
+      "step": 1870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.175922770985656e-05,
+      "loss": 0.7846,
+      "step": 1871
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17492354713757e-05,
+      "loss": 0.8944,
+      "step": 1872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.173923772333496e-05,
+      "loss": 0.8622,
+      "step": 1873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.172923446705379e-05,
+      "loss": 0.7861,
+      "step": 1874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171922570385224e-05,
+      "loss": 0.7667,
+      "step": 1875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.170921143505113e-05,
+      "loss": 0.7782,
+      "step": 1876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169919166197209e-05,
+      "loss": 0.8456,
+      "step": 1877
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168916638593736e-05,
+      "loss": 0.8299,
+      "step": 1878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.167913560826995e-05,
+      "loss": 0.8748,
+      "step": 1879
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166909933029364e-05,
+      "loss": 0.7808,
+      "step": 1880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165905755333287e-05,
+      "loss": 0.9109,
+      "step": 1881
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.164901027871282e-05,
+      "loss": 0.816,
+      "step": 1882
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163895750775943e-05,
+      "loss": 0.8697,
+      "step": 1883
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162889924179933e-05,
+      "loss": 0.8304,
+      "step": 1884
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161883548215993e-05,
+      "loss": 0.8066,
+      "step": 1885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.160876623016925e-05,
+      "loss": 0.8262,
+      "step": 1886
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159869148715618e-05,
+      "loss": 0.7769,
+      "step": 1887
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15886112544502e-05,
+      "loss": 0.8194,
+      "step": 1888
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.157852553338162e-05,
+      "loss": 0.7728,
+      "step": 1889
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156843432528144e-05,
+      "loss": 0.8638,
+      "step": 1890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155833763148133e-05,
+      "loss": 0.845,
+      "step": 1891
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.154823545331376e-05,
+      "loss": 0.8677,
+      "step": 1892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.153812779211188e-05,
+      "loss": 0.8553,
+      "step": 1893
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15280146492096e-05,
+      "loss": 0.865,
+      "step": 1894
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151789602594152e-05,
+      "loss": 0.8086,
+      "step": 1895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.150777192364296e-05,
+      "loss": 0.9121,
+      "step": 1896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149764234364999e-05,
+      "loss": 0.8918,
+      "step": 1897
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148750728729938e-05,
+      "loss": 0.8522,
+      "step": 1898
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.147736675592864e-05,
+      "loss": 0.7997,
+      "step": 1899
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.146722075087599e-05,
+      "loss": 0.8223,
+      "step": 1900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145706927348039e-05,
+      "loss": 0.8823,
+      "step": 1901
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14469123250815e-05,
+      "loss": 0.828,
+      "step": 1902
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14367499070197e-05,
+      "loss": 0.9084,
+      "step": 1903
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142658202063612e-05,
+      "loss": 0.8748,
+      "step": 1904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14164086672726e-05,
+      "loss": 0.8155,
+      "step": 1905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140622984827167e-05,
+      "loss": 0.8526,
+      "step": 1906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139604556497661e-05,
+      "loss": 0.8757,
+      "step": 1907
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.138585581873145e-05,
+      "loss": 0.8432,
+      "step": 1908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13756606108809e-05,
+      "loss": 0.8497,
+      "step": 1909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136545994277038e-05,
+      "loss": 0.9023,
+      "step": 1910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135525381574607e-05,
+      "loss": 0.8804,
+      "step": 1911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.134504223115483e-05,
+      "loss": 0.8825,
+      "step": 1912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133482519034428e-05,
+      "loss": 0.8237,
+      "step": 1913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132460269466277e-05,
+      "loss": 0.8315,
+      "step": 1914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131437474545926e-05,
+      "loss": 0.8444,
+      "step": 1915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.130414134408358e-05,
+      "loss": 0.8351,
+      "step": 1916
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12939024918862e-05,
+      "loss": 0.8714,
+      "step": 1917
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12836581902183e-05,
+      "loss": 0.8279,
+      "step": 1918
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.127340844043182e-05,
+      "loss": 0.9314,
+      "step": 1919
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.126315324387937e-05,
+      "loss": 0.889,
+      "step": 1920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125289260191434e-05,
+      "loss": 0.8666,
+      "step": 1921
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12426265158908e-05,
+      "loss": 0.7549,
+      "step": 1922
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.123235498716352e-05,
+      "loss": 0.8306,
+      "step": 1923
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122207801708802e-05,
+      "loss": 0.7875,
+      "step": 1924
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.121179560702055e-05,
+      "loss": 0.8617,
+      "step": 1925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120150775831804e-05,
+      "loss": 0.8248,
+      "step": 1926
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.119121447233817e-05,
+      "loss": 0.8917,
+      "step": 1927
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11809157504393e-05,
+      "loss": 0.7087,
+      "step": 1928
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.117061159398056e-05,
+      "loss": 0.8358,
+      "step": 1929
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116030200432175e-05,
+      "loss": 0.7779,
+      "step": 1930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114998698282341e-05,
+      "loss": 0.7883,
+      "step": 1931
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11396665308468e-05,
+      "loss": 0.7667,
+      "step": 1932
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.112934064975388e-05,
+      "loss": 0.8416,
+      "step": 1933
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111900934090731e-05,
+      "loss": 0.8279,
+      "step": 1934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110867260567054e-05,
+      "loss": 0.8777,
+      "step": 1935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.109833044540765e-05,
+      "loss": 0.8052,
+      "step": 1936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10879828614835e-05,
+      "loss": 0.8481,
+      "step": 1937
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.107762985526363e-05,
+      "loss": 0.8033,
+      "step": 1938
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.106727142811429e-05,
+      "loss": 0.8172,
+      "step": 1939
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105690758140246e-05,
+      "loss": 0.7809,
+      "step": 1940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.104653831649588e-05,
+      "loss": 0.8547,
+      "step": 1941
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10361636347629e-05,
+      "loss": 0.8606,
+      "step": 1942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.102578353757268e-05,
+      "loss": 0.8469,
+      "step": 1943
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101539802629505e-05,
+      "loss": 0.8266,
+      "step": 1944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100500710230056e-05,
+      "loss": 0.8239,
+      "step": 1945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09946107669605e-05,
+      "loss": 0.8846,
+      "step": 1946
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.098420902164684e-05,
+      "loss": 0.8626,
+      "step": 1947
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.097380186773226e-05,
+      "loss": 0.8052,
+      "step": 1948
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09633893065902e-05,
+      "loss": 0.7752,
+      "step": 1949
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095297133959476e-05,
+      "loss": 0.8858,
+      "step": 1950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094254796812079e-05,
+      "loss": 0.7731,
+      "step": 1951
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.093211919354383e-05,
+      "loss": 0.8908,
+      "step": 1952
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092168501724015e-05,
+      "loss": 0.8408,
+      "step": 1953
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091124544058673e-05,
+      "loss": 0.865,
+      "step": 1954
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.090080046496127e-05,
+      "loss": 0.8632,
+      "step": 1955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089035009174214e-05,
+      "loss": 0.7093,
+      "step": 1956
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087989432230849e-05,
+      "loss": 0.923,
+      "step": 1957
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086943315804012e-05,
+      "loss": 0.8865,
+      "step": 1958
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085896660031757e-05,
+      "loss": 0.8512,
+      "step": 1959
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08484946505221e-05,
+      "loss": 0.8647,
+      "step": 1960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.083801731003567e-05,
+      "loss": 0.8416,
+      "step": 1961
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.082753458024096e-05,
+      "loss": 0.8486,
+      "step": 1962
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081704646252134e-05,
+      "loss": 0.8778,
+      "step": 1963
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080655295826091e-05,
+      "loss": 0.9255,
+      "step": 1964
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.079605406884446e-05,
+      "loss": 0.8474,
+      "step": 1965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.078554979565755e-05,
+      "loss": 0.7948,
+      "step": 1966
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077504014008636e-05,
+      "loss": 0.8157,
+      "step": 1967
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.076452510351786e-05,
+      "loss": 0.8613,
+      "step": 1968
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075400468733967e-05,
+      "loss": 0.8069,
+      "step": 1969
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.074347889294016e-05,
+      "loss": 0.8123,
+      "step": 1970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.073294772170843e-05,
+      "loss": 0.8397,
+      "step": 1971
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072241117503421e-05,
+      "loss": 0.8377,
+      "step": 1972
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071186925430798e-05,
+      "loss": 0.9074,
+      "step": 1973
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.070132196092099e-05,
+      "loss": 0.8183,
+      "step": 1974
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06907692962651e-05,
+      "loss": 0.8928,
+      "step": 1975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068021126173294e-05,
+      "loss": 0.889,
+      "step": 1976
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066964785871783e-05,
+      "loss": 0.7065,
+      "step": 1977
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065907908861378e-05,
+      "loss": 0.8396,
+      "step": 1978
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064850495281557e-05,
+      "loss": 0.8222,
+      "step": 1979
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.063792545271859e-05,
+      "loss": 0.8654,
+      "step": 1980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.062734058971905e-05,
+      "loss": 0.8135,
+      "step": 1981
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061675036521379e-05,
+      "loss": 0.7198,
+      "step": 1982
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.060615478060037e-05,
+      "loss": 0.8092,
+      "step": 1983
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.059555383727709e-05,
+      "loss": 0.8381,
+      "step": 1984
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058494753664291e-05,
+      "loss": 0.8489,
+      "step": 1985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057433588009754e-05,
+      "loss": 0.8456,
+      "step": 1986
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056371886904137e-05,
+      "loss": 0.7964,
+      "step": 1987
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.055309650487551e-05,
+      "loss": 0.8262,
+      "step": 1988
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054246878900177e-05,
+      "loss": 0.7458,
+      "step": 1989
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053183572282266e-05,
+      "loss": 0.7221,
+      "step": 1990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05211973077414e-05,
+      "loss": 0.803,
+      "step": 1991
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.051055354516194e-05,
+      "loss": 0.807,
+      "step": 1992
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049990443648891e-05,
+      "loss": 0.7837,
+      "step": 1993
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.048924998312762e-05,
+      "loss": 0.8877,
+      "step": 1994
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047859018648417e-05,
+      "loss": 0.8445,
+      "step": 1995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.046792504796526e-05,
+      "loss": 0.8212,
+      "step": 1996
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045725456897838e-05,
+      "loss": 0.7973,
+      "step": 1997
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044657875093167e-05,
+      "loss": 0.8011,
+      "step": 1998
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0435897595234e-05,
+      "loss": 0.9231,
+      "step": 1999
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042521110329496e-05,
+      "loss": 0.784,
+      "step": 2000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.041451927652481e-05,
+      "loss": 0.8405,
+      "step": 2001
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04038221163345e-05,
+      "loss": 0.8862,
+      "step": 2002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.039311962413577e-05,
+      "loss": 0.8952,
+      "step": 2003
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.038241180134095e-05,
+      "loss": 0.7907,
+      "step": 2004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037169864936317e-05,
+      "loss": 0.8786,
+      "step": 2005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.036098016961619e-05,
+      "loss": 0.7892,
+      "step": 2006
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.035025636351452e-05,
+      "loss": 0.8058,
+      "step": 2007
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033952723247338e-05,
+      "loss": 0.7613,
+      "step": 2008
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.032879277790865e-05,
+      "loss": 0.8007,
+      "step": 2009
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031805300123693e-05,
+      "loss": 0.8451,
+      "step": 2010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.030730790387552e-05,
+      "loss": 0.8343,
+      "step": 2011
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029655748724244e-05,
+      "loss": 0.8512,
+      "step": 2012
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028580175275641e-05,
+      "loss": 0.7327,
+      "step": 2013
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027504070183681e-05,
+      "loss": 0.8666,
+      "step": 2014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.026427433590378e-05,
+      "loss": 0.87,
+      "step": 2015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025350265637815e-05,
+      "loss": 0.7953,
+      "step": 2016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02427256646814e-05,
+      "loss": 0.8284,
+      "step": 2017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.023194336223578e-05,
+      "loss": 0.8674,
+      "step": 2018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.022115575046416e-05,
+      "loss": 0.7601,
+      "step": 2019
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021036283079021e-05,
+      "loss": 0.8126,
+      "step": 2020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.019956460463822e-05,
+      "loss": 0.8815,
+      "step": 2021
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.018876107343322e-05,
+      "loss": 0.9032,
+      "step": 2022
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.017795223860093e-05,
+      "loss": 0.8549,
+      "step": 2023
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016713810156776e-05,
+      "loss": 0.8556,
+      "step": 2024
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.015631866376085e-05,
+      "loss": 0.8687,
+      "step": 2025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0145493926608e-05,
+      "loss": 0.8557,
+      "step": 2026
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013466389153773e-05,
+      "loss": 0.8704,
+      "step": 2027
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.012382855997926e-05,
+      "loss": 0.8514,
+      "step": 2028
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01129879333625e-05,
+      "loss": 0.8278,
+      "step": 2029
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.010214201311809e-05,
+      "loss": 0.8456,
+      "step": 2030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.009129080067731e-05,
+      "loss": 0.7607,
+      "step": 2031
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008043429747218e-05,
+      "loss": 0.8792,
+      "step": 2032
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.006957250493543e-05,
+      "loss": 0.8008,
+      "step": 2033
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005870542450045e-05,
+      "loss": 0.7899,
+      "step": 2034
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.004783305760136e-05,
+      "loss": 0.879,
+      "step": 2035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003695540567294e-05,
+      "loss": 0.783,
+      "step": 2036
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002607247015071e-05,
+      "loss": 0.8577,
+      "step": 2037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.001518425247088e-05,
+      "loss": 0.8737,
+      "step": 2038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00042907540703e-05,
+      "loss": 0.7936,
+      "step": 2039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999339197638663e-05,
+      "loss": 0.8129,
+      "step": 2040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998248792085809e-05,
+      "loss": 0.9322,
+      "step": 2041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.997157858892372e-05,
+      "loss": 0.809,
+      "step": 2042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.996066398202316e-05,
+      "loss": 0.8048,
+      "step": 2043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.994974410159683e-05,
+      "loss": 0.8813,
+      "step": 2044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993881894908576e-05,
+      "loss": 0.8694,
+      "step": 2045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.992788852593174e-05,
+      "loss": 0.7393,
+      "step": 2046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.991695283357724e-05,
+      "loss": 0.8474,
+      "step": 2047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.990601187346542e-05,
+      "loss": 0.7427,
+      "step": 2048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.989506564704012e-05,
+      "loss": 0.8047,
+      "step": 2049
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988411415574591e-05,
+      "loss": 0.7828,
+      "step": 2050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987315740102803e-05,
+      "loss": 0.9052,
+      "step": 2051
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.986219538433243e-05,
+      "loss": 0.825,
+      "step": 2052
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.985122810710572e-05,
+      "loss": 0.8663,
+      "step": 2053
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984025557079523e-05,
+      "loss": 0.7477,
+      "step": 2054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.982927777684902e-05,
+      "loss": 0.8452,
+      "step": 2055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.981829472671576e-05,
+      "loss": 0.8091,
+      "step": 2056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980730642184489e-05,
+      "loss": 0.7816,
+      "step": 2057
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97963128636865e-05,
+      "loss": 0.8104,
+      "step": 2058
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97853140536914e-05,
+      "loss": 0.8022,
+      "step": 2059
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977430999331106e-05,
+      "loss": 0.8463,
+      "step": 2060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97633006839977e-05,
+      "loss": 0.9269,
+      "step": 2061
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975228612720416e-05,
+      "loss": 0.8915,
+      "step": 2062
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.974126632438401e-05,
+      "loss": 0.793,
+      "step": 2063
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.973024127699152e-05,
+      "loss": 0.7662,
+      "step": 2064
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971921098648166e-05,
+      "loss": 0.7859,
+      "step": 2065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.970817545431005e-05,
+      "loss": 0.8794,
+      "step": 2066
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969713468193301e-05,
+      "loss": 0.846,
+      "step": 2067
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968608867080762e-05,
+      "loss": 0.7987,
+      "step": 2068
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967503742239155e-05,
+      "loss": 0.8206,
+      "step": 2069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966398093814324e-05,
+      "loss": 0.7943,
+      "step": 2070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965291921952178e-05,
+      "loss": 0.8428,
+      "step": 2071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.964185226798696e-05,
+      "loss": 0.8153,
+      "step": 2072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.963078008499928e-05,
+      "loss": 0.9058,
+      "step": 2073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.961970267201989e-05,
+      "loss": 0.8154,
+      "step": 2074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960862003051064e-05,
+      "loss": 0.8085,
+      "step": 2075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.959753216193414e-05,
+      "loss": 0.8233,
+      "step": 2076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95864390677536e-05,
+      "loss": 0.783,
+      "step": 2077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957534074943292e-05,
+      "loss": 0.911,
+      "step": 2078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956423720843677e-05,
+      "loss": 0.929,
+      "step": 2079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955312844623045e-05,
+      "loss": 0.7462,
+      "step": 2080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.954201446427994e-05,
+      "loss": 0.9079,
+      "step": 2081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.953089526405196e-05,
+      "loss": 0.7697,
+      "step": 2082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.951977084701386e-05,
+      "loss": 0.8375,
+      "step": 2083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.950864121463373e-05,
+      "loss": 0.9175,
+      "step": 2084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94975063683803e-05,
+      "loss": 0.7754,
+      "step": 2085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.948636630972302e-05,
+      "loss": 0.7572,
+      "step": 2086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947522104013205e-05,
+      "loss": 0.7461,
+      "step": 2087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.946407056107815e-05,
+      "loss": 0.7052,
+      "step": 2088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.945291487403288e-05,
+      "loss": 0.7602,
+      "step": 2089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944175398046842e-05,
+      "loss": 0.7944,
+      "step": 2090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94305878818576e-05,
+      "loss": 0.7358,
+      "step": 2091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.941941657967408e-05,
+      "loss": 0.849,
+      "step": 2092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940824007539203e-05,
+      "loss": 0.9284,
+      "step": 2093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.939705837048642e-05,
+      "loss": 0.83,
+      "step": 2094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.93858714664329e-05,
+      "loss": 0.7753,
+      "step": 2095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.937467936470773e-05,
+      "loss": 0.9123,
+      "step": 2096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.936348206678797e-05,
+      "loss": 0.7171,
+      "step": 2097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.935227957415126e-05,
+      "loss": 0.8613,
+      "step": 2098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.934107188827598e-05,
+      "loss": 0.8434,
+      "step": 2099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93298590106412e-05,
+      "loss": 0.8299,
+      "step": 2100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.931864094272663e-05,
+      "loss": 0.7736,
+      "step": 2101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930741768601272e-05,
+      "loss": 0.8967,
+      "step": 2102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.929618924198057e-05,
+      "loss": 0.8399,
+      "step": 2103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.928495561211199e-05,
+      "loss": 0.8739,
+      "step": 2104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927371679788943e-05,
+      "loss": 0.8459,
+      "step": 2105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926247280079609e-05,
+      "loss": 0.7943,
+      "step": 2106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.925122362231578e-05,
+      "loss": 0.8082,
+      "step": 2107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923996926393305e-05,
+      "loss": 0.8221,
+      "step": 2108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.922870972713311e-05,
+      "loss": 0.9331,
+      "step": 2109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.921744501340187e-05,
+      "loss": 0.7974,
+      "step": 2110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920617512422588e-05,
+      "loss": 0.8337,
+      "step": 2111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.919490006109243e-05,
+      "loss": 0.8926,
+      "step": 2112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.918361982548948e-05,
+      "loss": 0.8629,
+      "step": 2113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.917233441890562e-05,
+      "loss": 0.8346,
+      "step": 2114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91610438428302e-05,
+      "loss": 0.8223,
+      "step": 2115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.914974809875318e-05,
+      "loss": 0.7996,
+      "step": 2116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.913844718816524e-05,
+      "loss": 0.9266,
+      "step": 2117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.912714111255776e-05,
+      "loss": 0.8051,
+      "step": 2118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911582987342278e-05,
+      "loss": 0.8126,
+      "step": 2119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.910451347225298e-05,
+      "loss": 0.7903,
+      "step": 2120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.909319191054182e-05,
+      "loss": 0.8185,
+      "step": 2121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.908186518978332e-05,
+      "loss": 0.8261,
+      "step": 2122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907053331147231e-05,
+      "loss": 0.893,
+      "step": 2123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.905919627710419e-05,
+      "loss": 0.8745,
+      "step": 2124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904785408817509e-05,
+      "loss": 0.8532,
+      "step": 2125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.903650674618182e-05,
+      "loss": 0.7164,
+      "step": 2126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.902515425262187e-05,
+      "loss": 0.8515,
+      "step": 2127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.901379660899343e-05,
+      "loss": 0.8311,
+      "step": 2128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.900243381679528e-05,
+      "loss": 0.812,
+      "step": 2129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.899106587752699e-05,
+      "loss": 0.7188,
+      "step": 2130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.897969279268877e-05,
+      "loss": 0.7785,
+      "step": 2131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.896831456378147e-05,
+      "loss": 0.7928,
+      "step": 2132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895693119230667e-05,
+      "loss": 0.7508,
+      "step": 2133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89455426797666e-05,
+      "loss": 0.8311,
+      "step": 2134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89341490276642e-05,
+      "loss": 0.8845,
+      "step": 2135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.892275023750305e-05,
+      "loss": 0.7918,
+      "step": 2136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.891134631078741e-05,
+      "loss": 0.8478,
+      "step": 2137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.889993724902228e-05,
+      "loss": 0.9215,
+      "step": 2138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.888852305371324e-05,
+      "loss": 0.799,
+      "step": 2139
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.887710372636662e-05,
+      "loss": 0.7862,
+      "step": 2140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88656792684894e-05,
+      "loss": 0.8173,
+      "step": 2141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.885424968158925e-05,
+      "loss": 0.7931,
+      "step": 2142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.884281496717449e-05,
+      "loss": 0.9289,
+      "step": 2143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.883137512675415e-05,
+      "loss": 0.8365,
+      "step": 2144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.881993016183794e-05,
+      "loss": 0.7893,
+      "step": 2145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.880848007393619e-05,
+      "loss": 0.804,
+      "step": 2146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.879702486455998e-05,
+      "loss": 0.7627,
+      "step": 2147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8785564535221e-05,
+      "loss": 0.7154,
+      "step": 2148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877409908743166e-05,
+      "loss": 0.8013,
+      "step": 2149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.876262852270502e-05,
+      "loss": 0.8537,
+      "step": 2150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875115284255486e-05,
+      "loss": 0.8864,
+      "step": 2151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.873967204849555e-05,
+      "loss": 0.7907,
+      "step": 2152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.872818614204225e-05,
+      "loss": 0.7433,
+      "step": 2153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.871669512471068e-05,
+      "loss": 0.769,
+      "step": 2154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.870519899801731e-05,
+      "loss": 0.7664,
+      "step": 2155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.869369776347924e-05,
+      "loss": 0.7769,
+      "step": 2156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868219142261427e-05,
+      "loss": 0.8626,
+      "step": 2157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.867067997694089e-05,
+      "loss": 0.8467,
+      "step": 2158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865916342797821e-05,
+      "loss": 0.8485,
+      "step": 2159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.864764177724607e-05,
+      "loss": 0.7718,
+      "step": 2160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.863611502626494e-05,
+      "loss": 0.6962,
+      "step": 2161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.862458317655601e-05,
+      "loss": 0.7724,
+      "step": 2162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.861304622964109e-05,
+      "loss": 0.9255,
+      "step": 2163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.860150418704268e-05,
+      "loss": 0.9055,
+      "step": 2164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.858995705028397e-05,
+      "loss": 0.7733,
+      "step": 2165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.857840482088881e-05,
+      "loss": 0.8955,
+      "step": 2166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.856684750038173e-05,
+      "loss": 0.7982,
+      "step": 2167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855528509028793e-05,
+      "loss": 0.869,
+      "step": 2168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.854371759213327e-05,
+      "loss": 0.8668,
+      "step": 2169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.853214500744428e-05,
+      "loss": 0.8431,
+      "step": 2170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852056733774818e-05,
+      "loss": 0.8381,
+      "step": 2171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.850898458457283e-05,
+      "loss": 0.8349,
+      "step": 2172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.849739674944683e-05,
+      "loss": 0.7443,
+      "step": 2173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848580383389935e-05,
+      "loss": 0.8629,
+      "step": 2174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.847420583946034e-05,
+      "loss": 0.7909,
+      "step": 2175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.846260276766028e-05,
+      "loss": 0.7755,
+      "step": 2176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.845099462003048e-05,
+      "loss": 0.8537,
+      "step": 2177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.843938139810282e-05,
+      "loss": 0.7763,
+      "step": 2178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.842776310340986e-05,
+      "loss": 0.8309,
+      "step": 2179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.841613973748486e-05,
+      "loss": 0.8147,
+      "step": 2180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84045113018617e-05,
+      "loss": 0.8874,
+      "step": 2181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.839287779807501e-05,
+      "loss": 0.8322,
+      "step": 2182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.838123922766e-05,
+      "loss": 0.8205,
+      "step": 2183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.836959559215259e-05,
+      "loss": 0.7959,
+      "step": 2184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.835794689308938e-05,
+      "loss": 0.9372,
+      "step": 2185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.834629313200762e-05,
+      "loss": 0.8401,
+      "step": 2186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.833463431044525e-05,
+      "loss": 0.7965,
+      "step": 2187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832297042994083e-05,
+      "loss": 0.773,
+      "step": 2188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.831130149203364e-05,
+      "loss": 0.7781,
+      "step": 2189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82996274982636e-05,
+      "loss": 0.8365,
+      "step": 2190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.828794845017128e-05,
+      "loss": 0.7161,
+      "step": 2191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827626434929796e-05,
+      "loss": 0.7909,
+      "step": 2192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.826457519718558e-05,
+      "loss": 0.8069,
+      "step": 2193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.825288099537672e-05,
+      "loss": 0.7971,
+      "step": 2194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.824118174541464e-05,
+      "loss": 0.8095,
+      "step": 2195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822947744884325e-05,
+      "loss": 0.8922,
+      "step": 2196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.821776810720718e-05,
+      "loss": 0.8236,
+      "step": 2197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.820605372205164e-05,
+      "loss": 0.7957,
+      "step": 2198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.819433429492258e-05,
+      "loss": 0.8266,
+      "step": 2199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.818260982736661e-05,
+      "loss": 0.8225,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.817088032093094e-05,
+      "loss": 0.7964,
+      "step": 2201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815914577716351e-05,
+      "loss": 0.8992,
+      "step": 2202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.814740619761291e-05,
+      "loss": 0.8183,
+      "step": 2203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.813566158382835e-05,
+      "loss": 0.7787,
+      "step": 2204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.812391193735981e-05,
+      "loss": 0.7766,
+      "step": 2205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.811215725975779e-05,
+      "loss": 0.7869,
+      "step": 2206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.810039755257358e-05,
+      "loss": 0.8327,
+      "step": 2207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808863281735909e-05,
+      "loss": 0.8665,
+      "step": 2208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.807686305566685e-05,
+      "loss": 0.7212,
+      "step": 2209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.806508826905013e-05,
+      "loss": 0.7481,
+      "step": 2210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80533084590628e-05,
+      "loss": 0.891,
+      "step": 2211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80415236272594e-05,
+      "loss": 0.7904,
+      "step": 2212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80297337751952e-05,
+      "loss": 0.8442,
+      "step": 2213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801793890442602e-05,
+      "loss": 0.901,
+      "step": 2214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800613901650847e-05,
+      "loss": 0.7948,
+      "step": 2215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.799433411299972e-05,
+      "loss": 0.8203,
+      "step": 2216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.798252419545763e-05,
+      "loss": 0.8168,
+      "step": 2217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.797070926544076e-05,
+      "loss": 0.7764,
+      "step": 2218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.795888932450827e-05,
+      "loss": 0.8347,
+      "step": 2219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.794706437422003e-05,
+      "loss": 0.7391,
+      "step": 2220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.793523441613656e-05,
+      "loss": 0.7821,
+      "step": 2221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.792339945181904e-05,
+      "loss": 0.8385,
+      "step": 2222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791155948282928e-05,
+      "loss": 0.7469,
+      "step": 2223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.789971451072978e-05,
+      "loss": 0.7988,
+      "step": 2224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788786453708373e-05,
+      "loss": 0.8385,
+      "step": 2225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78760095634549e-05,
+      "loss": 0.8094,
+      "step": 2226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.786414959140781e-05,
+      "loss": 0.8545,
+      "step": 2227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.785228462250755e-05,
+      "loss": 0.9607,
+      "step": 2228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784041465831995e-05,
+      "loss": 0.9584,
+      "step": 2229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.782853970041146e-05,
+      "loss": 0.8365,
+      "step": 2230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781665975034919e-05,
+      "loss": 0.8969,
+      "step": 2231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.780477480970091e-05,
+      "loss": 0.9358,
+      "step": 2232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.779288488003504e-05,
+      "loss": 0.887,
+      "step": 2233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778098996292068e-05,
+      "loss": 0.8838,
+      "step": 2234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77690900599276e-05,
+      "loss": 0.9047,
+      "step": 2235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.775718517262617e-05,
+      "loss": 0.8523,
+      "step": 2236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.774527530258746e-05,
+      "loss": 0.7945,
+      "step": 2237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77333604513832e-05,
+      "loss": 0.8926,
+      "step": 2238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.772144062058576e-05,
+      "loss": 0.801,
+      "step": 2239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77095158117682e-05,
+      "loss": 0.8026,
+      "step": 2240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769758602650418e-05,
+      "loss": 0.8286,
+      "step": 2241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768565126636806e-05,
+      "loss": 0.8668,
+      "step": 2242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.767371153293485e-05,
+      "loss": 0.808,
+      "step": 2243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766176682778022e-05,
+      "loss": 0.76,
+      "step": 2244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.764981715248046e-05,
+      "loss": 0.7412,
+      "step": 2245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.763786250861256e-05,
+      "loss": 0.8551,
+      "step": 2246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.762590289775418e-05,
+      "loss": 0.7521,
+      "step": 2247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761393832148354e-05,
+      "loss": 0.8113,
+      "step": 2248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.760196878137965e-05,
+      "loss": 0.7519,
+      "step": 2249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.758999427902205e-05,
+      "loss": 0.7802,
+      "step": 2250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.757801481599101e-05,
+      "loss": 0.8893,
+      "step": 2251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.756603039386744e-05,
+      "loss": 0.8453,
+      "step": 2252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75540410142329e-05,
+      "loss": 0.7595,
+      "step": 2253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.754204667866959e-05,
+      "loss": 0.6937,
+      "step": 2254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.753004738876041e-05,
+      "loss": 0.8156,
+      "step": 2255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.751804314608883e-05,
+      "loss": 0.8178,
+      "step": 2256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.750603395223906e-05,
+      "loss": 0.9124,
+      "step": 2257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.749401980879592e-05,
+      "loss": 0.8847,
+      "step": 2258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.748200071734491e-05,
+      "loss": 0.7403,
+      "step": 2259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.746997667947215e-05,
+      "loss": 0.7244,
+      "step": 2260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.745794769676439e-05,
+      "loss": 0.8049,
+      "step": 2261
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.744591377080913e-05,
+      "loss": 0.7767,
+      "step": 2262
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.743387490319444e-05,
+      "loss": 0.8985,
+      "step": 2263
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.742183109550905e-05,
+      "loss": 0.7523,
+      "step": 2264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740978234934236e-05,
+      "loss": 0.8363,
+      "step": 2265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739772866628445e-05,
+      "loss": 0.8484,
+      "step": 2266
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.738567004792598e-05,
+      "loss": 0.8549,
+      "step": 2267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73736064958583e-05,
+      "loss": 0.7919,
+      "step": 2268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.736153801167347e-05,
+      "loss": 0.8444,
+      "step": 2269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.734946459696407e-05,
+      "loss": 0.7882,
+      "step": 2270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733738625332343e-05,
+      "loss": 0.8286,
+      "step": 2271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.732530298234552e-05,
+      "loss": 0.8299,
+      "step": 2272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.731321478562494e-05,
+      "loss": 0.8829,
+      "step": 2273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.73011216647569e-05,
+      "loss": 0.8726,
+      "step": 2274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.728902362133737e-05,
+      "loss": 0.8208,
+      "step": 2275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.727692065696287e-05,
+      "loss": 0.8983,
+      "step": 2276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72648127732306e-05,
+      "loss": 0.7845,
+      "step": 2277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.725269997173843e-05,
+      "loss": 0.7494,
+      "step": 2278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.724058225408484e-05,
+      "loss": 0.8761,
+      "step": 2279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.7228459621869e-05,
+      "loss": 0.8219,
+      "step": 2280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72163320766907e-05,
+      "loss": 0.7343,
+      "step": 2281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720419962015038e-05,
+      "loss": 0.866,
+      "step": 2282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.719206225384914e-05,
+      "loss": 0.7776,
+      "step": 2283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.717991997938873e-05,
+      "loss": 0.8033,
+      "step": 2284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.716777279837152e-05,
+      "loss": 0.8169,
+      "step": 2285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.715562071240057e-05,
+      "loss": 0.7895,
+      "step": 2286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.714346372307956e-05,
+      "loss": 0.9078,
+      "step": 2287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.713130183201284e-05,
+      "loss": 0.7867,
+      "step": 2288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711913504080534e-05,
+      "loss": 0.866,
+      "step": 2289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710696335106274e-05,
+      "loss": 0.7592,
+      "step": 2290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709478676439128e-05,
+      "loss": 0.8617,
+      "step": 2291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.708260528239788e-05,
+      "loss": 0.8112,
+      "step": 2292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.707041890669012e-05,
+      "loss": 0.8766,
+      "step": 2293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70582276388762e-05,
+      "loss": 0.8216,
+      "step": 2294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.704603148056499e-05,
+      "loss": 0.8878,
+      "step": 2295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.703383043336598e-05,
+      "loss": 0.859,
+      "step": 2296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.702162449888931e-05,
+      "loss": 0.8018,
+      "step": 2297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.700941367874579e-05,
+      "loss": 0.8203,
+      "step": 2298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.699719797454684e-05,
+      "loss": 0.7377,
+      "step": 2299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.698497738790456e-05,
+      "loss": 0.8838,
+      "step": 2300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.697275192043165e-05,
+      "loss": 0.8858,
+      "step": 2301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.696052157374149e-05,
+      "loss": 0.7778,
+      "step": 2302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.694828634944812e-05,
+      "loss": 0.8679,
+      "step": 2303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.693604624916614e-05,
+      "loss": 0.9041,
+      "step": 2304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69238012745109e-05,
+      "loss": 0.8411,
+      "step": 2305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.691155142709834e-05,
+      "loss": 0.847,
+      "step": 2306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.689929670854502e-05,
+      "loss": 0.8423,
+      "step": 2307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68870371204682e-05,
+      "loss": 0.8374,
+      "step": 2308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.687477266448572e-05,
+      "loss": 0.863,
+      "step": 2309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686250334221612e-05,
+      "loss": 0.7579,
+      "step": 2310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.685022915527854e-05,
+      "loss": 0.7622,
+      "step": 2311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68379501052928e-05,
+      "loss": 0.7982,
+      "step": 2312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.682566619387933e-05,
+      "loss": 0.824,
+      "step": 2313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681337742265919e-05,
+      "loss": 0.7343,
+      "step": 2314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.680108379325414e-05,
+      "loss": 0.8457,
+      "step": 2315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.678878530728652e-05,
+      "loss": 0.8091,
+      "step": 2316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.677648196637935e-05,
+      "loss": 0.695,
+      "step": 2317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676417377215629e-05,
+      "loss": 0.8409,
+      "step": 2318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.675186072624158e-05,
+      "loss": 0.8174,
+      "step": 2319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673954283026019e-05,
+      "loss": 0.8086,
+      "step": 2320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.672722008583767e-05,
+      "loss": 0.8813,
+      "step": 2321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.671489249460023e-05,
+      "loss": 0.9171,
+      "step": 2322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.670256005817472e-05,
+      "loss": 0.7977,
+      "step": 2323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669022277818861e-05,
+      "loss": 0.9757,
+      "step": 2324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.667788065627003e-05,
+      "loss": 0.8561,
+      "step": 2325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.666553369404777e-05,
+      "loss": 0.8304,
+      "step": 2326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66531818931512e-05,
+      "loss": 0.8187,
+      "step": 2327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.664082525521038e-05,
+      "loss": 0.7564,
+      "step": 2328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662846378185599e-05,
+      "loss": 0.8464,
+      "step": 2329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.661609747471935e-05,
+      "loss": 0.7998,
+      "step": 2330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660372633543239e-05,
+      "loss": 0.8189,
+      "step": 2331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659135036562773e-05,
+      "loss": 0.7167,
+      "step": 2332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65789695669386e-05,
+      "loss": 0.7662,
+      "step": 2333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.656658394099885e-05,
+      "loss": 0.8629,
+      "step": 2334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.655419348944301e-05,
+      "loss": 0.8648,
+      "step": 2335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.654179821390621e-05,
+      "loss": 0.7094,
+      "step": 2336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.652939811602424e-05,
+      "loss": 0.8928,
+      "step": 2337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.651699319743347e-05,
+      "loss": 0.7742,
+      "step": 2338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.6504583459771e-05,
+      "loss": 0.8273,
+      "step": 2339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64921689046745e-05,
+      "loss": 0.7164,
+      "step": 2340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64797495337823e-05,
+      "loss": 0.8535,
+      "step": 2341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646732534873336e-05,
+      "loss": 0.8045,
+      "step": 2342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.645489635116726e-05,
+      "loss": 0.8199,
+      "step": 2343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.644246254272424e-05,
+      "loss": 0.7833,
+      "step": 2344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.643002392504516e-05,
+      "loss": 0.8593,
+      "step": 2345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64175804997715e-05,
+      "loss": 0.7744,
+      "step": 2346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.640513226854542e-05,
+      "loss": 0.9042,
+      "step": 2347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.639267923300969e-05,
+      "loss": 0.7922,
+      "step": 2348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63802213948077e-05,
+      "loss": 0.825,
+      "step": 2349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636775875558348e-05,
+      "loss": 0.9111,
+      "step": 2350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63552913169817e-05,
+      "loss": 0.9031,
+      "step": 2351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.634281908064768e-05,
+      "loss": 0.8714,
+      "step": 2352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.633034204822733e-05,
+      "loss": 0.8009,
+      "step": 2353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.631786022136724e-05,
+      "loss": 0.9126,
+      "step": 2354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63053736017146e-05,
+      "loss": 0.8218,
+      "step": 2355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.629288219091722e-05,
+      "loss": 0.791,
+      "step": 2356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62803859906236e-05,
+      "loss": 0.8191,
+      "step": 2357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626788500248283e-05,
+      "loss": 0.8074,
+      "step": 2358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.625537922814463e-05,
+      "loss": 0.7608,
+      "step": 2359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.624286866925937e-05,
+      "loss": 0.7498,
+      "step": 2360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.623035332747804e-05,
+      "loss": 0.7603,
+      "step": 2361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.621783320445227e-05,
+      "loss": 0.8813,
+      "step": 2362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.62053083018343e-05,
+      "loss": 0.8747,
+      "step": 2363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.619277862127702e-05,
+      "loss": 0.8876,
+      "step": 2364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.618024416443396e-05,
+      "loss": 0.8544,
+      "step": 2365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.616770493295925e-05,
+      "loss": 0.7483,
+      "step": 2366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.615516092850767e-05,
+      "loss": 0.8001,
+      "step": 2367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614261215273462e-05,
+      "loss": 0.7825,
+      "step": 2368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.613005860729614e-05,
+      "loss": 0.8547,
+      "step": 2369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.611750029384891e-05,
+      "loss": 0.8047,
+      "step": 2370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.610493721405021e-05,
+      "loss": 0.7473,
+      "step": 2371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.609236936955797e-05,
+      "loss": 0.9052,
+      "step": 2372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.607979676203073e-05,
+      "loss": 0.7859,
+      "step": 2373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.606721939312769e-05,
+      "loss": 0.7099,
+      "step": 2374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.605463726450865e-05,
+      "loss": 0.79,
+      "step": 2375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.604205037783404e-05,
+      "loss": 0.7377,
+      "step": 2376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.602945873476494e-05,
+      "loss": 0.7474,
+      "step": 2377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.601686233696302e-05,
+      "loss": 0.8827,
+      "step": 2378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.600426118609062e-05,
+      "loss": 0.7257,
+      "step": 2379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599165528381068e-05,
+      "loss": 0.8295,
+      "step": 2380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.597904463178679e-05,
+      "loss": 0.8036,
+      "step": 2381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.596642923168314e-05,
+      "loss": 0.8355,
+      "step": 2382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.595380908516454e-05,
+      "loss": 0.7832,
+      "step": 2383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.594118419389647e-05,
+      "loss": 0.7203,
+      "step": 2384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.592855455954501e-05,
+      "loss": 0.7814,
+      "step": 2385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.591592018377685e-05,
+      "loss": 0.8636,
+      "step": 2386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.590328106825933e-05,
+      "loss": 0.7003,
+      "step": 2387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.589063721466042e-05,
+      "loss": 0.7887,
+      "step": 2388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.587798862464869e-05,
+      "loss": 0.8197,
+      "step": 2389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.586533529989334e-05,
+      "loss": 0.8941,
+      "step": 2390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.585267724206423e-05,
+      "loss": 0.8613,
+      "step": 2391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.584001445283179e-05,
+      "loss": 0.8388,
+      "step": 2392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.582734693386711e-05,
+      "loss": 0.7942,
+      "step": 2393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58146746868419e-05,
+      "loss": 0.8957,
+      "step": 2394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.580199771342849e-05,
+      "loss": 0.8461,
+      "step": 2395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.578931601529984e-05,
+      "loss": 0.8624,
+      "step": 2396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57766295941295e-05,
+      "loss": 0.8234,
+      "step": 2397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.576393845159172e-05,
+      "loss": 0.8425,
+      "step": 2398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.575124258936128e-05,
+      "loss": 0.8914,
+      "step": 2399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.573854200911364e-05,
+      "loss": 0.9245,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.572583671252488e-05,
+      "loss": 0.8107,
+      "step": 2401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.571312670127168e-05,
+      "loss": 0.8323,
+      "step": 2402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.570041197703134e-05,
+      "loss": 0.877,
+      "step": 2403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.568769254148182e-05,
+      "loss": 0.8875,
+      "step": 2404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.567496839630167e-05,
+      "loss": 0.8954,
+      "step": 2405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.566223954317007e-05,
+      "loss": 0.7947,
+      "step": 2406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.564950598376681e-05,
+      "loss": 0.8695,
+      "step": 2407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.563676771977233e-05,
+      "loss": 0.7363,
+      "step": 2408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.562402475286768e-05,
+      "loss": 0.816,
+      "step": 2409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.561127708473447e-05,
+      "loss": 0.814,
+      "step": 2410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.559852471705505e-05,
+      "loss": 0.8824,
+      "step": 2411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.558576765151228e-05,
+      "loss": 0.7383,
+      "step": 2412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557300588978968e-05,
+      "loss": 0.8977,
+      "step": 2413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556023943357141e-05,
+      "loss": 0.7969,
+      "step": 2414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.554746828454225e-05,
+      "loss": 0.8529,
+      "step": 2415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553469244438754e-05,
+      "loss": 0.7885,
+      "step": 2416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.552191191479334e-05,
+      "loss": 0.7789,
+      "step": 2417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.550912669744619e-05,
+      "loss": 0.844,
+      "step": 2418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54963367940334e-05,
+      "loss": 0.7889,
+      "step": 2419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.548354220624279e-05,
+      "loss": 0.7097,
+      "step": 2420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.547074293576284e-05,
+      "loss": 0.81,
+      "step": 2421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545793898428265e-05,
+      "loss": 0.7636,
+      "step": 2422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.544513035349194e-05,
+      "loss": 0.9157,
+      "step": 2423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543231704508102e-05,
+      "loss": 0.9652,
+      "step": 2424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541949906074084e-05,
+      "loss": 0.7769,
+      "step": 2425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.540667640216297e-05,
+      "loss": 0.7308,
+      "step": 2426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.539384907103959e-05,
+      "loss": 0.7675,
+      "step": 2427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.538101706906348e-05,
+      "loss": 0.8737,
+      "step": 2428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.536818039792807e-05,
+      "loss": 0.8823,
+      "step": 2429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.535533905932738e-05,
+      "loss": 0.8858,
+      "step": 2430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.534249305495606e-05,
+      "loss": 0.8392,
+      "step": 2431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532964238650937e-05,
+      "loss": 0.8292,
+      "step": 2432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.531678705568318e-05,
+      "loss": 0.7589,
+      "step": 2433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.530392706417398e-05,
+      "loss": 0.9448,
+      "step": 2434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52910624136789e-05,
+      "loss": 0.8854,
+      "step": 2435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.527819310589563e-05,
+      "loss": 0.8254,
+      "step": 2436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.526531914252251e-05,
+      "loss": 0.8312,
+      "step": 2437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.525244052525851e-05,
+      "loss": 0.8114,
+      "step": 2438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.523955725580319e-05,
+      "loss": 0.8935,
+      "step": 2439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.522666933585672e-05,
+      "loss": 0.7981,
+      "step": 2440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.521377676711988e-05,
+      "loss": 0.8,
+      "step": 2441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52008795512941e-05,
+      "loss": 0.8647,
+      "step": 2442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.518797769008138e-05,
+      "loss": 0.8562,
+      "step": 2443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.517507118518436e-05,
+      "loss": 0.884,
+      "step": 2444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516216003830629e-05,
+      "loss": 0.8125,
+      "step": 2445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.514924425115102e-05,
+      "loss": 0.8606,
+      "step": 2446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.513632382542303e-05,
+      "loss": 0.8542,
+      "step": 2447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.512339876282738e-05,
+      "loss": 0.7634,
+      "step": 2448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.511046906506979e-05,
+      "loss": 0.8841,
+      "step": 2449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.509753473385653e-05,
+      "loss": 0.7625,
+      "step": 2450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.508459577089456e-05,
+      "loss": 0.8025,
+      "step": 2451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.507165217789138e-05,
+      "loss": 0.8062,
+      "step": 2452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.505870395655513e-05,
+      "loss": 0.8904,
+      "step": 2453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.504575110859456e-05,
+      "loss": 0.8615,
+      "step": 2454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.503279363571905e-05,
+      "loss": 0.7582,
+      "step": 2455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501983153963854e-05,
+      "loss": 0.871,
+      "step": 2456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.500686482206366e-05,
+      "loss": 0.7172,
+      "step": 2457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.499389348470555e-05,
+      "loss": 0.7585,
+      "step": 2458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.498091752927603e-05,
+      "loss": 0.8426,
+      "step": 2459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.496793695748751e-05,
+      "loss": 0.8713,
+      "step": 2460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.495495177105299e-05,
+      "loss": 0.8216,
+      "step": 2461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.494196197168613e-05,
+      "loss": 0.8752,
+      "step": 2462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.492896756110116e-05,
+      "loss": 0.8746,
+      "step": 2463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.491596854101292e-05,
+      "loss": 0.7611,
+      "step": 2464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.490296491313685e-05,
+      "loss": 0.8415,
+      "step": 2465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.488995667918903e-05,
+      "loss": 0.8295,
+      "step": 2466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.487694384088611e-05,
+      "loss": 0.8016,
+      "step": 2467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.486392639994541e-05,
+      "loss": 0.7671,
+      "step": 2468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.485090435808476e-05,
+      "loss": 0.7623,
+      "step": 2469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.483787771702268e-05,
+      "loss": 0.7903,
+      "step": 2470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.482484647847828e-05,
+      "loss": 0.8559,
+      "step": 2471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.481181064417123e-05,
+      "loss": 0.8579,
+      "step": 2472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479877021582188e-05,
+      "loss": 0.9005,
+      "step": 2473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.478572519515112e-05,
+      "loss": 0.8425,
+      "step": 2474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477267558388049e-05,
+      "loss": 0.807,
+      "step": 2475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.475962138373213e-05,
+      "loss": 0.8426,
+      "step": 2476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.474656259642873e-05,
+      "loss": 0.7458,
+      "step": 2477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.473349922369369e-05,
+      "loss": 0.8182,
+      "step": 2478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.472043126725092e-05,
+      "loss": 0.7757,
+      "step": 2479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.470735872882498e-05,
+      "loss": 0.8342,
+      "step": 2480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.469428161014102e-05,
+      "loss": 0.8907,
+      "step": 2481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.468119991292481e-05,
+      "loss": 0.7134,
+      "step": 2482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46681136389027e-05,
+      "loss": 0.8044,
+      "step": 2483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.465502278980168e-05,
+      "loss": 0.9024,
+      "step": 2484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.464192736734931e-05,
+      "loss": 0.6841,
+      "step": 2485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.462882737327376e-05,
+      "loss": 0.8085,
+      "step": 2486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.461572280930383e-05,
+      "loss": 0.8079,
+      "step": 2487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.460261367716887e-05,
+      "loss": 0.8671,
+      "step": 2488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.45894999785989e-05,
+      "loss": 0.9516,
+      "step": 2489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.457638171532447e-05,
+      "loss": 0.8198,
+      "step": 2490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.45632588890768e-05,
+      "loss": 0.777,
+      "step": 2491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.455013150158768e-05,
+      "loss": 0.824,
+      "step": 2492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.453699955458948e-05,
+      "loss": 0.956,
+      "step": 2493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.452386304981521e-05,
+      "loss": 0.7888,
+      "step": 2494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.451072198899847e-05,
+      "loss": 0.8502,
+      "step": 2495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.449757637387348e-05,
+      "loss": 0.7684,
+      "step": 2496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4484426206175e-05,
+      "loss": 0.8836,
+      "step": 2497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.447127148763845e-05,
+      "loss": 0.756,
+      "step": 2498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.445811221999982e-05,
+      "loss": 0.9334,
+      "step": 2499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.444494840499573e-05,
+      "loss": 0.8087,
+      "step": 2500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.443178004436336e-05,
+      "loss": 0.8646,
+      "step": 2501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.441860713984054e-05,
+      "loss": 0.8115,
+      "step": 2502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.440542969316563e-05,
+      "loss": 0.8142,
+      "step": 2503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.439224770607767e-05,
+      "loss": 0.7908,
+      "step": 2504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.437906118031624e-05,
+      "loss": 0.756,
+      "step": 2505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.436587011762153e-05,
+      "loss": 0.8035,
+      "step": 2506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.435267451973435e-05,
+      "loss": 0.8283,
+      "step": 2507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43394743883961e-05,
+      "loss": 0.846,
+      "step": 2508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.432626972534876e-05,
+      "loss": 0.8743,
+      "step": 2509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.431306053233494e-05,
+      "loss": 0.8394,
+      "step": 2510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.429984681109782e-05,
+      "loss": 0.8861,
+      "step": 2511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.428662856338117e-05,
+      "loss": 0.8759,
+      "step": 2512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.427340579092941e-05,
+      "loss": 0.7784,
+      "step": 2513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.426017849548748e-05,
+      "loss": 0.8454,
+      "step": 2514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4246946678801e-05,
+      "loss": 0.7332,
+      "step": 2515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.423371034261611e-05,
+      "loss": 0.8817,
+      "step": 2516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422046948867963e-05,
+      "loss": 0.7598,
+      "step": 2517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.420722411873887e-05,
+      "loss": 0.7566,
+      "step": 2518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.419397423454183e-05,
+      "loss": 0.7358,
+      "step": 2519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418071983783707e-05,
+      "loss": 0.7173,
+      "step": 2520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.416746093037371e-05,
+      "loss": 0.8117,
+      "step": 2521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.415419751390155e-05,
+      "loss": 0.7792,
+      "step": 2522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.414092959017089e-05,
+      "loss": 0.7435,
+      "step": 2523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.412765716093272e-05,
+      "loss": 0.7905,
+      "step": 2524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.411438022793852e-05,
+      "loss": 0.8572,
+      "step": 2525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.410109879294046e-05,
+      "loss": 0.8007,
+      "step": 2526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.408781285769124e-05,
+      "loss": 0.8384,
+      "step": 2527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.407452242394419e-05,
+      "loss": 0.7694,
+      "step": 2528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406122749345321e-05,
+      "loss": 0.7103,
+      "step": 2529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.404792806797283e-05,
+      "loss": 0.8889,
+      "step": 2530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.403462414925812e-05,
+      "loss": 0.7804,
+      "step": 2531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.402131573906479e-05,
+      "loss": 0.9002,
+      "step": 2532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.400800283914911e-05,
+      "loss": 0.8059,
+      "step": 2533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.399468545126795e-05,
+      "loss": 0.8325,
+      "step": 2534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.398136357717879e-05,
+      "loss": 0.7979,
+      "step": 2535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.396803721863969e-05,
+      "loss": 0.8095,
+      "step": 2536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.395470637740929e-05,
+      "loss": 0.847,
+      "step": 2537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.394137105524685e-05,
+      "loss": 0.8854,
+      "step": 2538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.392803125391219e-05,
+      "loss": 0.8715,
+      "step": 2539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.391468697516576e-05,
+      "loss": 0.8622,
+      "step": 2540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.390133822076854e-05,
+      "loss": 0.8436,
+      "step": 2541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.388798499248216e-05,
+      "loss": 0.8193,
+      "step": 2542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.387462729206882e-05,
+      "loss": 0.7574,
+      "step": 2543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.386126512129131e-05,
+      "loss": 0.7781,
+      "step": 2544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.384789848191299e-05,
+      "loss": 0.6886,
+      "step": 2545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.383452737569784e-05,
+      "loss": 0.7362,
+      "step": 2546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.382115180441043e-05,
+      "loss": 0.8937,
+      "step": 2547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.380777176981587e-05,
+      "loss": 0.8203,
+      "step": 2548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.379438727367993e-05,
+      "loss": 0.8712,
+      "step": 2549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.378099831776892e-05,
+      "loss": 0.7901,
+      "step": 2550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.376760490384973e-05,
+      "loss": 0.8205,
+      "step": 2551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.375420703368993e-05,
+      "loss": 0.8092,
+      "step": 2552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374080470905754e-05,
+      "loss": 0.7388,
+      "step": 2553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.372739793172126e-05,
+      "loss": 0.7866,
+      "step": 2554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.371398670345037e-05,
+      "loss": 0.8417,
+      "step": 2555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.370057102601467e-05,
+      "loss": 0.8597,
+      "step": 2556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.368715090118467e-05,
+      "loss": 0.8681,
+      "step": 2557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.367372633073136e-05,
+      "loss": 0.8128,
+      "step": 2558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.366029731642636e-05,
+      "loss": 0.9294,
+      "step": 2559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.364686386004184e-05,
+      "loss": 0.7326,
+      "step": 2560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.363342596335061e-05,
+      "loss": 0.7513,
+      "step": 2561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.361998362812607e-05,
+      "loss": 0.8559,
+      "step": 2562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.360653685614213e-05,
+      "loss": 0.7987,
+      "step": 2563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.359308564917335e-05,
+      "loss": 0.8518,
+      "step": 2564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.357963000899486e-05,
+      "loss": 0.7718,
+      "step": 2565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.356616993738238e-05,
+      "loss": 0.8354,
+      "step": 2566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.355270543611218e-05,
+      "loss": 0.8548,
+      "step": 2567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.353923650696118e-05,
+      "loss": 0.7715,
+      "step": 2568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.352576315170681e-05,
+      "loss": 0.7699,
+      "step": 2569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.351228537212717e-05,
+      "loss": 0.8319,
+      "step": 2570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.349880317000082e-05,
+      "loss": 0.7703,
+      "step": 2571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.348531654710705e-05,
+      "loss": 0.7641,
+      "step": 2572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.347182550522563e-05,
+      "loss": 0.7791,
+      "step": 2573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.345833004613695e-05,
+      "loss": 0.883,
+      "step": 2574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.344483017162195e-05,
+      "loss": 0.815,
+      "step": 2575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.343132588346223e-05,
+      "loss": 0.8014,
+      "step": 2576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34178171834399e-05,
+      "loss": 0.8331,
+      "step": 2577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.340430407333765e-05,
+      "loss": 0.8011,
+      "step": 2578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.339078655493884e-05,
+      "loss": 0.7764,
+      "step": 2579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.337726463002728e-05,
+      "loss": 0.7846,
+      "step": 2580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.336373830038747e-05,
+      "loss": 0.849,
+      "step": 2581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.335020756780442e-05,
+      "loss": 0.7963,
+      "step": 2582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33366724340638e-05,
+      "loss": 0.759,
+      "step": 2583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33231329009518e-05,
+      "loss": 0.8293,
+      "step": 2584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.330958897025516e-05,
+      "loss": 0.7504,
+      "step": 2585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.329604064376129e-05,
+      "loss": 0.8839,
+      "step": 2586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.328248792325812e-05,
+      "loss": 0.8239,
+      "step": 2587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.326893081053418e-05,
+      "loss": 0.8124,
+      "step": 2588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.325536930737856e-05,
+      "loss": 0.8262,
+      "step": 2589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.324180341558094e-05,
+      "loss": 0.912,
+      "step": 2590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32282331369316e-05,
+      "loss": 0.8464,
+      "step": 2591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32146584732214e-05,
+      "loss": 0.8811,
+      "step": 2592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.320107942624173e-05,
+      "loss": 0.783,
+      "step": 2593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.318749599778458e-05,
+      "loss": 0.8687,
+      "step": 2594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.317390818964255e-05,
+      "loss": 0.8749,
+      "step": 2595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.316031600360879e-05,
+      "loss": 0.831,
+      "step": 2596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.314671944147702e-05,
+      "loss": 0.8104,
+      "step": 2597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313311850504157e-05,
+      "loss": 0.8015,
+      "step": 2598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.311951319609731e-05,
+      "loss": 0.7981,
+      "step": 2599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31059035164397e-05,
+      "loss": 0.7898,
+      "step": 2600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.309228946786481e-05,
+      "loss": 0.7578,
+      "step": 2601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.307867105216924e-05,
+      "loss": 0.8497,
+      "step": 2602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.306504827115018e-05,
+      "loss": 0.8222,
+      "step": 2603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.30514211266054e-05,
+      "loss": 0.8255,
+      "step": 2604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.303778962033324e-05,
+      "loss": 0.8846,
+      "step": 2605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.302415375413265e-05,
+      "loss": 0.8614,
+      "step": 2606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.301051352980312e-05,
+      "loss": 0.8721,
+      "step": 2607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.299686894914468e-05,
+      "loss": 0.9166,
+      "step": 2608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.298322001395802e-05,
+      "loss": 0.7906,
+      "step": 2609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.296956672604435e-05,
+      "loss": 0.8209,
+      "step": 2610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.295590908720545e-05,
+      "loss": 0.8235,
+      "step": 2611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.294224709924373e-05,
+      "loss": 0.8646,
+      "step": 2612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.292858076396211e-05,
+      "loss": 0.8367,
+      "step": 2613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.291491008316409e-05,
+      "loss": 0.9451,
+      "step": 2614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.290123505865377e-05,
+      "loss": 0.723,
+      "step": 2615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.288755569223585e-05,
+      "loss": 0.8059,
+      "step": 2616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.287387198571552e-05,
+      "loss": 0.8856,
+      "step": 2617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.286018394089863e-05,
+      "loss": 0.8387,
+      "step": 2618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.284649155959155e-05,
+      "loss": 0.8471,
+      "step": 2619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28327948436012e-05,
+      "loss": 0.8323,
+      "step": 2620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.281909379473514e-05,
+      "loss": 0.8103,
+      "step": 2621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280538841480148e-05,
+      "loss": 0.8939,
+      "step": 2622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.279167870560887e-05,
+      "loss": 0.7819,
+      "step": 2623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.277796466896657e-05,
+      "loss": 0.8022,
+      "step": 2624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.276424630668435e-05,
+      "loss": 0.7944,
+      "step": 2625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.275052362057265e-05,
+      "loss": 0.8514,
+      "step": 2626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27367966124424e-05,
+      "loss": 0.8491,
+      "step": 2627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272306528410511e-05,
+      "loss": 0.7373,
+      "step": 2628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27093296373729e-05,
+      "loss": 0.7988,
+      "step": 2629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.269558967405841e-05,
+      "loss": 0.9085,
+      "step": 2630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.268184539597488e-05,
+      "loss": 0.7628,
+      "step": 2631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.266809680493614e-05,
+      "loss": 0.8097,
+      "step": 2632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.265434390275654e-05,
+      "loss": 0.7951,
+      "step": 2633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.264058669125102e-05,
+      "loss": 0.8484,
+      "step": 2634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.26268251722351e-05,
+      "loss": 0.7918,
+      "step": 2635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.261305934752486e-05,
+      "loss": 0.7661,
+      "step": 2636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.259928921893694e-05,
+      "loss": 0.9161,
+      "step": 2637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.258551478828856e-05,
+      "loss": 0.8103,
+      "step": 2638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.257173605739749e-05,
+      "loss": 0.77,
+      "step": 2639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.255795302808211e-05,
+      "loss": 0.7895,
+      "step": 2640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.254416570216132e-05,
+      "loss": 0.8547,
+      "step": 2641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.253037408145459e-05,
+      "loss": 0.8397,
+      "step": 2642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2516578167782e-05,
+      "loss": 0.8027,
+      "step": 2643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.250277796296413e-05,
+      "loss": 0.8629,
+      "step": 2644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.248897346882221e-05,
+      "loss": 0.8336,
+      "step": 2645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.247516468717797e-05,
+      "loss": 0.7841,
+      "step": 2646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.246135161985371e-05,
+      "loss": 0.8311,
+      "step": 2647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.244753426867233e-05,
+      "loss": 0.6947,
+      "step": 2648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.243371263545727e-05,
+      "loss": 0.7276,
+      "step": 2649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.241988672203256e-05,
+      "loss": 0.809,
+      "step": 2650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.240605653022275e-05,
+      "loss": 0.7658,
+      "step": 2651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.239222206185302e-05,
+      "loss": 0.887,
+      "step": 2652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.237838331874903e-05,
+      "loss": 0.8994,
+      "step": 2653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.236454030273705e-05,
+      "loss": 0.913,
+      "step": 2654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.235069301564397e-05,
+      "loss": 0.8831,
+      "step": 2655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.233684145929714e-05,
+      "loss": 0.8076,
+      "step": 2656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.232298563552453e-05,
+      "loss": 0.8363,
+      "step": 2657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.230912554615469e-05,
+      "loss": 0.813,
+      "step": 2658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.229526119301666e-05,
+      "loss": 0.7849,
+      "step": 2659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.228139257794012e-05,
+      "loss": 0.8334,
+      "step": 2660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.226751970275528e-05,
+      "loss": 0.8723,
+      "step": 2661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.225364256929291e-05,
+      "loss": 0.7995,
+      "step": 2662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.223976117938437e-05,
+      "loss": 0.8753,
+      "step": 2663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.222587553486151e-05,
+      "loss": 0.8461,
+      "step": 2664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.221198563755682e-05,
+      "loss": 0.6721,
+      "step": 2665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.219809148930333e-05,
+      "loss": 0.8576,
+      "step": 2666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21841930919346e-05,
+      "loss": 0.8518,
+      "step": 2667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.217029044728477e-05,
+      "loss": 0.7637,
+      "step": 2668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.215638355718858e-05,
+      "loss": 0.8326,
+      "step": 2669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.214247242348125e-05,
+      "loss": 0.7691,
+      "step": 2670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.212855704799863e-05,
+      "loss": 0.8289,
+      "step": 2671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.211463743257709e-05,
+      "loss": 0.8882,
+      "step": 2672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.210071357905358e-05,
+      "loss": 0.8161,
+      "step": 2673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.208678548926559e-05,
+      "loss": 0.8161,
+      "step": 2674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.20728531650512e-05,
+      "loss": 0.7891,
+      "step": 2675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205891660824902e-05,
+      "loss": 0.8013,
+      "step": 2676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.204497582069823e-05,
+      "loss": 0.763,
+      "step": 2677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.203103080423857e-05,
+      "loss": 0.7952,
+      "step": 2678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.201708156071033e-05,
+      "loss": 0.7213,
+      "step": 2679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.200312809195437e-05,
+      "loss": 0.8226,
+      "step": 2680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.198917039981206e-05,
+      "loss": 0.7119,
+      "step": 2681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197520848612543e-05,
+      "loss": 0.8353,
+      "step": 2682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.196124235273697e-05,
+      "loss": 0.8542,
+      "step": 2683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.194727200148977e-05,
+      "loss": 0.7919,
+      "step": 2684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.193329743422746e-05,
+      "loss": 0.7756,
+      "step": 2685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.191931865279424e-05,
+      "loss": 0.8669,
+      "step": 2686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.190533565903486e-05,
+      "loss": 0.8594,
+      "step": 2687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.189134845479462e-05,
+      "loss": 0.8593,
+      "step": 2688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.187735704191942e-05,
+      "loss": 0.8323,
+      "step": 2689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.186336142225562e-05,
+      "loss": 0.7034,
+      "step": 2690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184936159765024e-05,
+      "loss": 0.8268,
+      "step": 2691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.183535756995077e-05,
+      "loss": 0.8061,
+      "step": 2692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.182134934100532e-05,
+      "loss": 0.7495,
+      "step": 2693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.180733691266251e-05,
+      "loss": 0.8687,
+      "step": 2694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.179332028677156e-05,
+      "loss": 0.8418,
+      "step": 2695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.177929946518216e-05,
+      "loss": 0.7841,
+      "step": 2696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.176527444974468e-05,
+      "loss": 0.8453,
+      "step": 2697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.175124524230991e-05,
+      "loss": 0.802,
+      "step": 2698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.173721184472928e-05,
+      "loss": 0.7731,
+      "step": 2699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.172317425885476e-05,
+      "loss": 0.7905,
+      "step": 2700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.170913248653884e-05,
+      "loss": 0.8582,
+      "step": 2701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16950865296346e-05,
+      "loss": 0.7749,
+      "step": 2702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.168103638999564e-05,
+      "loss": 0.8427,
+      "step": 2703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.166698206947614e-05,
+      "loss": 0.7782,
+      "step": 2704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.165292356993082e-05,
+      "loss": 0.8174,
+      "step": 2705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.163886089321493e-05,
+      "loss": 0.7719,
+      "step": 2706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.162479404118433e-05,
+      "loss": 0.8248,
+      "step": 2707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.161072301569534e-05,
+      "loss": 0.7427,
+      "step": 2708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.159664781860495e-05,
+      "loss": 0.8196,
+      "step": 2709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.158256845177059e-05,
+      "loss": 0.8433,
+      "step": 2710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15684849170503e-05,
+      "loss": 0.8048,
+      "step": 2711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.155439721630264e-05,
+      "loss": 0.8114,
+      "step": 2712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.154030535138675e-05,
+      "loss": 0.7954,
+      "step": 2713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.152620932416231e-05,
+      "loss": 0.7827,
+      "step": 2714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.151210913648952e-05,
+      "loss": 0.8673,
+      "step": 2715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149800479022915e-05,
+      "loss": 0.8055,
+      "step": 2716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.148389628724256e-05,
+      "loss": 0.8167,
+      "step": 2717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.146978362939159e-05,
+      "loss": 0.8019,
+      "step": 2718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.145566681853867e-05,
+      "loss": 0.8107,
+      "step": 2719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.144154585654675e-05,
+      "loss": 0.8574,
+      "step": 2720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.142742074527936e-05,
+      "loss": 0.9328,
+      "step": 2721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.141329148660055e-05,
+      "loss": 0.8355,
+      "step": 2722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.139915808237493e-05,
+      "loss": 0.8116,
+      "step": 2723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.138502053446766e-05,
+      "loss": 0.735,
+      "step": 2724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.137087884474446e-05,
+      "loss": 0.7282,
+      "step": 2725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.135673301507156e-05,
+      "loss": 0.8358,
+      "step": 2726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.134258304731574e-05,
+      "loss": 0.7993,
+      "step": 2727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.132842894334438e-05,
+      "loss": 0.7584,
+      "step": 2728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.131427070502535e-05,
+      "loss": 0.8711,
+      "step": 2729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.130010833422707e-05,
+      "loss": 0.7852,
+      "step": 2730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.128594183281853e-05,
+      "loss": 0.8516,
+      "step": 2731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.127177120266927e-05,
+      "loss": 0.819,
+      "step": 2732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.125759644564932e-05,
+      "loss": 0.8235,
+      "step": 2733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.124341756362933e-05,
+      "loss": 0.7584,
+      "step": 2734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.122923455848044e-05,
+      "loss": 0.809,
+      "step": 2735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.121504743207436e-05,
+      "loss": 0.6541,
+      "step": 2736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.120085618628331e-05,
+      "loss": 0.8352,
+      "step": 2737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.118666082298012e-05,
+      "loss": 0.776,
+      "step": 2738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.117246134403809e-05,
+      "loss": 0.8484,
+      "step": 2739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.115825775133113e-05,
+      "loss": 0.8756,
+      "step": 2740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.114405004673361e-05,
+      "loss": 0.8133,
+      "step": 2741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.112983823212053e-05,
+      "loss": 0.7651,
+      "step": 2742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.111562230936738e-05,
+      "loss": 0.8775,
+      "step": 2743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.110140228035021e-05,
+      "loss": 0.8121,
+      "step": 2744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10871781469456e-05,
+      "loss": 0.8345,
+      "step": 2745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.107294991103069e-05,
+      "loss": 0.8639,
+      "step": 2746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.105871757448316e-05,
+      "loss": 0.7794,
+      "step": 2747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.104448113918118e-05,
+      "loss": 0.7954,
+      "step": 2748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.103024060700355e-05,
+      "loss": 0.8012,
+      "step": 2749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.101599597982954e-05,
+      "loss": 0.7339,
+      "step": 2750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.100174725953897e-05,
+      "loss": 0.7926,
+      "step": 2751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.098749444801224e-05,
+      "loss": 0.872,
+      "step": 2752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.097323754713027e-05,
+      "loss": 0.8314,
+      "step": 2753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.095897655877449e-05,
+      "loss": 0.8181,
+      "step": 2754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09447114848269e-05,
+      "loss": 0.8372,
+      "step": 2755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093044232717004e-05,
+      "loss": 0.8156,
+      "step": 2756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.091616908768695e-05,
+      "loss": 0.757,
+      "step": 2757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.090189176826129e-05,
+      "loss": 0.6753,
+      "step": 2758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.088761037077718e-05,
+      "loss": 0.8212,
+      "step": 2759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087332489711931e-05,
+      "loss": 0.8093,
+      "step": 2760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.08590353491729e-05,
+      "loss": 0.8068,
+      "step": 2761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.08447417288237e-05,
+      "loss": 0.8253,
+      "step": 2762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.083044403795805e-05,
+      "loss": 0.8366,
+      "step": 2763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.081614227846275e-05,
+      "loss": 0.882,
+      "step": 2764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.080183645222519e-05,
+      "loss": 0.8253,
+      "step": 2765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.078752656113326e-05,
+      "loss": 0.8701,
+      "step": 2766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.077321260707544e-05,
+      "loss": 0.8312,
+      "step": 2767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.075889459194067e-05,
+      "loss": 0.8874,
+      "step": 2768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.074457251761853e-05,
+      "loss": 0.7761,
+      "step": 2769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.073024638599901e-05,
+      "loss": 0.8141,
+      "step": 2770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.071591619897274e-05,
+      "loss": 0.8069,
+      "step": 2771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.070158195843083e-05,
+      "loss": 0.8692,
+      "step": 2772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.068724366626495e-05,
+      "loss": 0.8725,
+      "step": 2773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.067290132436726e-05,
+      "loss": 0.8354,
+      "step": 2774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065855493463054e-05,
+      "loss": 0.853,
+      "step": 2775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.064420449894803e-05,
+      "loss": 0.8239,
+      "step": 2776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.062985001921351e-05,
+      "loss": 0.8128,
+      "step": 2777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.061549149732134e-05,
+      "loss": 0.6819,
+      "step": 2778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.060112893516637e-05,
+      "loss": 0.7559,
+      "step": 2779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.0586762334644e-05,
+      "loss": 0.8318,
+      "step": 2780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.057239169765015e-05,
+      "loss": 0.7976,
+      "step": 2781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05580170260813e-05,
+      "loss": 0.6006,
+      "step": 2782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.054363832183444e-05,
+      "loss": 0.837,
+      "step": 2783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.052925558680709e-05,
+      "loss": 0.7384,
+      "step": 2784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.051486882289732e-05,
+      "loss": 0.8594,
+      "step": 2785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.050047803200372e-05,
+      "loss": 0.7695,
+      "step": 2786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04860832160254e-05,
+      "loss": 0.7677,
+      "step": 2787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.047168437686203e-05,
+      "loss": 0.8335,
+      "step": 2788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.045728151641378e-05,
+      "loss": 0.7634,
+      "step": 2789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04428746365814e-05,
+      "loss": 0.7854,
+      "step": 2790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.042846373926609e-05,
+      "loss": 0.7717,
+      "step": 2791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.041404882636963e-05,
+      "loss": 0.8103,
+      "step": 2792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.039962989979437e-05,
+      "loss": 0.8162,
+      "step": 2793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.038520696144311e-05,
+      "loss": 0.8991,
+      "step": 2794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.037078001321922e-05,
+      "loss": 0.7614,
+      "step": 2795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03563490570266e-05,
+      "loss": 0.7626,
+      "step": 2796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.034191409476967e-05,
+      "loss": 0.7793,
+      "step": 2797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.032747512835337e-05,
+      "loss": 0.9185,
+      "step": 2798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.031303215968322e-05,
+      "loss": 0.7938,
+      "step": 2799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.029858519066518e-05,
+      "loss": 0.8411,
+      "step": 2800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.028413422320583e-05,
+      "loss": 0.8265,
+      "step": 2801
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.026967925921218e-05,
+      "loss": 0.8492,
+      "step": 2802
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.025522030059188e-05,
+      "loss": 0.8411,
+      "step": 2803
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.024075734925303e-05,
+      "loss": 0.7871,
+      "step": 2804
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022629040710425e-05,
+      "loss": 0.7639,
+      "step": 2805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.021181947605473e-05,
+      "loss": 0.8171,
+      "step": 2806
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.019734455801418e-05,
+      "loss": 0.8404,
+      "step": 2807
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.01828656548928e-05,
+      "loss": 0.8432,
+      "step": 2808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.016838276860137e-05,
+      "loss": 0.835,
+      "step": 2809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.015389590105114e-05,
+      "loss": 0.8416,
+      "step": 2810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.013940505415394e-05,
+      "loss": 0.7977,
+      "step": 2811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.012491022982206e-05,
+      "loss": 0.7867,
+      "step": 2812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.011041142996839e-05,
+      "loss": 0.8408,
+      "step": 2813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.009590865650628e-05,
+      "loss": 0.7449,
+      "step": 2814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.008140191134964e-05,
+      "loss": 0.6905,
+      "step": 2815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.006689119641288e-05,
+      "loss": 0.8863,
+      "step": 2816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.005237651361099e-05,
+      "loss": 0.8296,
+      "step": 2817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00378578648594e-05,
+      "loss": 0.8283,
+      "step": 2818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.002333525207412e-05,
+      "loss": 0.9345,
+      "step": 2819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.000880867717169e-05,
+      "loss": 0.8307,
+      "step": 2820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.999427814206912e-05,
+      "loss": 0.8275,
+      "step": 2821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.997974364868397e-05,
+      "loss": 0.9063,
+      "step": 2822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.996520519893438e-05,
+      "loss": 0.7406,
+      "step": 2823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.995066279473891e-05,
+      "loss": 0.7889,
+      "step": 2824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.993611643801669e-05,
+      "loss": 0.7495,
+      "step": 2825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.992156613068738e-05,
+      "loss": 0.8598,
+      "step": 2826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.990701187467117e-05,
+      "loss": 0.9363,
+      "step": 2827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.989245367188877e-05,
+      "loss": 0.7725,
+      "step": 2828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.987789152426135e-05,
+      "loss": 0.7279,
+      "step": 2829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.986332543371066e-05,
+      "loss": 0.8139,
+      "step": 2830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.984875540215897e-05,
+      "loss": 0.7314,
+      "step": 2831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.983418143152907e-05,
+      "loss": 0.8402,
+      "step": 2832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.981960352374422e-05,
+      "loss": 0.8252,
+      "step": 2833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.980502168072826e-05,
+      "loss": 0.8585,
+      "step": 2834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.979043590440551e-05,
+      "loss": 0.812,
+      "step": 2835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.977584619670083e-05,
+      "loss": 0.7508,
+      "step": 2836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.976125255953962e-05,
+      "loss": 0.808,
+      "step": 2837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.974665499484773e-05,
+      "loss": 0.7898,
+      "step": 2838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97320535045516e-05,
+      "loss": 0.818,
+      "step": 2839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.971744809057813e-05,
+      "loss": 0.8256,
+      "step": 2840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97028387548548e-05,
+      "loss": 0.7312,
+      "step": 2841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.968822549930956e-05,
+      "loss": 0.8598,
+      "step": 2842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.967360832587087e-05,
+      "loss": 0.8162,
+      "step": 2843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.965898723646776e-05,
+      "loss": 0.8501,
+      "step": 2844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.964436223302973e-05,
+      "loss": 0.7137,
+      "step": 2845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.962973331748682e-05,
+      "loss": 0.7624,
+      "step": 2846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.961510049176957e-05,
+      "loss": 0.8536,
+      "step": 2847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.960046375780903e-05,
+      "loss": 0.8004,
+      "step": 2848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.958582311753681e-05,
+      "loss": 0.7655,
+      "step": 2849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.957117857288498e-05,
+      "loss": 0.8099,
+      "step": 2850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.955653012578615e-05,
+      "loss": 0.7618,
+      "step": 2851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.954187777817346e-05,
+      "loss": 0.8822,
+      "step": 2852
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.952722153198054e-05,
+      "loss": 0.8231,
+      "step": 2853
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.951256138914156e-05,
+      "loss": 0.951,
+      "step": 2854
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.949789735159116e-05,
+      "loss": 0.7519,
+      "step": 2855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.948322942126456e-05,
+      "loss": 0.8226,
+      "step": 2856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.946855760009742e-05,
+      "loss": 0.9203,
+      "step": 2857
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.945388189002596e-05,
+      "loss": 0.847,
+      "step": 2858
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.943920229298691e-05,
+      "loss": 0.8316,
+      "step": 2859
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.942451881091752e-05,
+      "loss": 0.8163,
+      "step": 2860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.940983144575552e-05,
+      "loss": 0.8185,
+      "step": 2861
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.939514019943914e-05,
+      "loss": 0.8175,
+      "step": 2862
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.938044507390723e-05,
+      "loss": 0.8867,
+      "step": 2863
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.936574607109901e-05,
+      "loss": 0.8102,
+      "step": 2864
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93510431929543e-05,
+      "loss": 0.7837,
+      "step": 2865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.933633644141341e-05,
+      "loss": 0.8138,
+      "step": 2866
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.932162581841716e-05,
+      "loss": 0.8577,
+      "step": 2867
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.930691132590686e-05,
+      "loss": 0.7249,
+      "step": 2868
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.929219296582439e-05,
+      "loss": 0.7682,
+      "step": 2869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.927747074011207e-05,
+      "loss": 0.8594,
+      "step": 2870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.926274465071276e-05,
+      "loss": 0.7994,
+      "step": 2871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.924801469956986e-05,
+      "loss": 0.7349,
+      "step": 2872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.923328088862722e-05,
+      "loss": 0.7842,
+      "step": 2873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.921854321982924e-05,
+      "loss": 0.8378,
+      "step": 2874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.920380169512083e-05,
+      "loss": 0.8351,
+      "step": 2875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91890563164474e-05,
+      "loss": 0.8692,
+      "step": 2876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.917430708575487e-05,
+      "loss": 0.8384,
+      "step": 2877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.915955400498962e-05,
+      "loss": 0.8365,
+      "step": 2878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914479707609863e-05,
+      "loss": 0.7687,
+      "step": 2879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.913003630102933e-05,
+      "loss": 0.7416,
+      "step": 2880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.911527168172968e-05,
+      "loss": 0.7849,
+      "step": 2881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91005032201481e-05,
+      "loss": 0.7334,
+      "step": 2882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.908573091823359e-05,
+      "loss": 0.8286,
+      "step": 2883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.907095477793563e-05,
+      "loss": 0.8393,
+      "step": 2884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.905617480120413e-05,
+      "loss": 0.8108,
+      "step": 2885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.904139098998964e-05,
+      "loss": 0.8356,
+      "step": 2886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.902660334624312e-05,
+      "loss": 0.8525,
+      "step": 2887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.901181187191607e-05,
+      "loss": 0.7533,
+      "step": 2888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.899701656896048e-05,
+      "loss": 0.751,
+      "step": 2889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.898221743932888e-05,
+      "loss": 0.7948,
+      "step": 2890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.896741448497426e-05,
+      "loss": 0.8412,
+      "step": 2891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.895260770785013e-05,
+      "loss": 0.7547,
+      "step": 2892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.893779710991054e-05,
+      "loss": 0.8498,
+      "step": 2893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.892298269310998e-05,
+      "loss": 0.817,
+      "step": 2894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.890816445940349e-05,
+      "loss": 0.8529,
+      "step": 2895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.889334241074661e-05,
+      "loss": 0.8401,
+      "step": 2896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.887851654909537e-05,
+      "loss": 0.855,
+      "step": 2897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.88636868764063e-05,
+      "loss": 0.7119,
+      "step": 2898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.884885339463645e-05,
+      "loss": 0.7762,
+      "step": 2899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.883401610574336e-05,
+      "loss": 0.8606,
+      "step": 2900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.881917501168508e-05,
+      "loss": 0.8908,
+      "step": 2901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.880433011442016e-05,
+      "loss": 0.8024,
+      "step": 2902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.878948141590763e-05,
+      "loss": 0.7561,
+      "step": 2903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.877462891810707e-05,
+      "loss": 0.7429,
+      "step": 2904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.875977262297852e-05,
+      "loss": 0.7333,
+      "step": 2905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.874491253248252e-05,
+      "loss": 0.8366,
+      "step": 2906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.873004864858017e-05,
+      "loss": 0.8646,
+      "step": 2907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.871518097323299e-05,
+      "loss": 0.9092,
+      "step": 2908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.870030950840304e-05,
+      "loss": 0.8629,
+      "step": 2909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.868543425605287e-05,
+      "loss": 0.8128,
+      "step": 2910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.867055521814555e-05,
+      "loss": 0.8304,
+      "step": 2911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.865567239664463e-05,
+      "loss": 0.6922,
+      "step": 2912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.864078579351417e-05,
+      "loss": 0.8222,
+      "step": 2913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.862589541071872e-05,
+      "loss": 0.8434,
+      "step": 2914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.861100125022331e-05,
+      "loss": 0.8299,
+      "step": 2915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.859610331399354e-05,
+      "loss": 0.8999,
+      "step": 2916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.858120160399545e-05,
+      "loss": 0.7697,
+      "step": 2917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.856629612219555e-05,
+      "loss": 0.8549,
+      "step": 2918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85513868705609e-05,
+      "loss": 0.7683,
+      "step": 2919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.853647385105905e-05,
+      "loss": 0.6724,
+      "step": 2920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.852155706565806e-05,
+      "loss": 0.8015,
+      "step": 2921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.850663651632643e-05,
+      "loss": 0.8034,
+      "step": 2922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.849171220503322e-05,
+      "loss": 0.738,
+      "step": 2923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.847678413374795e-05,
+      "loss": 0.8873,
+      "step": 2924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.846185230444065e-05,
+      "loss": 0.9396,
+      "step": 2925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.844691671908184e-05,
+      "loss": 0.9403,
+      "step": 2926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.843197737964255e-05,
+      "loss": 0.8763,
+      "step": 2927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.841703428809426e-05,
+      "loss": 0.7765,
+      "step": 2928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.840208744640901e-05,
+      "loss": 0.8417,
+      "step": 2929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.83871368565593e-05,
+      "loss": 0.8446,
+      "step": 2930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.837218252051812e-05,
+      "loss": 0.8336,
+      "step": 2931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.835722444025898e-05,
+      "loss": 0.7675,
+      "step": 2932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.834226261775583e-05,
+      "loss": 0.8761,
+      "step": 2933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.832729705498319e-05,
+      "loss": 0.8403,
+      "step": 2934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8312327753916e-05,
+      "loss": 0.8432,
+      "step": 2935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.829735471652978e-05,
+      "loss": 0.8203,
+      "step": 2936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.828237794480042e-05,
+      "loss": 0.6784,
+      "step": 2937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.826739744070441e-05,
+      "loss": 0.8247,
+      "step": 2938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.825241320621871e-05,
+      "loss": 0.7657,
+      "step": 2939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.823742524332074e-05,
+      "loss": 0.7446,
+      "step": 2940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.822243355398844e-05,
+      "loss": 0.7463,
+      "step": 2941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.820743814020022e-05,
+      "loss": 0.7477,
+      "step": 2942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8192439003935e-05,
+      "loss": 0.8342,
+      "step": 2943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.817743614717218e-05,
+      "loss": 0.8462,
+      "step": 2944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.816242957189166e-05,
+      "loss": 0.9295,
+      "step": 2945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.814741928007383e-05,
+      "loss": 0.8315,
+      "step": 2946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.813240527369959e-05,
+      "loss": 0.813,
+      "step": 2947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.811738755475024e-05,
+      "loss": 0.8477,
+      "step": 2948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81023661252077e-05,
+      "loss": 0.8571,
+      "step": 2949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.808734098705428e-05,
+      "loss": 0.76,
+      "step": 2950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.807231214227284e-05,
+      "loss": 0.8421,
+      "step": 2951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.805727959284669e-05,
+      "loss": 0.808,
+      "step": 2952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.804224334075967e-05,
+      "loss": 0.7733,
+      "step": 2953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.802720338799605e-05,
+      "loss": 0.7787,
+      "step": 2954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.801215973654064e-05,
+      "loss": 0.778,
+      "step": 2955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.79971123883787e-05,
+      "loss": 0.721,
+      "step": 2956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.798206134549604e-05,
+      "loss": 0.8414,
+      "step": 2957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.796700660987888e-05,
+      "loss": 0.8242,
+      "step": 2958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.795194818351395e-05,
+      "loss": 0.8561,
+      "step": 2959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.793688606838851e-05,
+      "loss": 0.8286,
+      "step": 2960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.792182026649026e-05,
+      "loss": 0.6136,
+      "step": 2961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790675077980742e-05,
+      "loss": 0.8001,
+      "step": 2962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.789167761032865e-05,
+      "loss": 0.8396,
+      "step": 2963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.787660076004315e-05,
+      "loss": 0.8723,
+      "step": 2964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.786152023094057e-05,
+      "loss": 0.7749,
+      "step": 2965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.784643602501105e-05,
+      "loss": 0.874,
+      "step": 2966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.783134814424524e-05,
+      "loss": 0.6946,
+      "step": 2967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.781625659063423e-05,
+      "loss": 0.7843,
+      "step": 2968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.780116136616964e-05,
+      "loss": 0.8483,
+      "step": 2969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.778606247284357e-05,
+      "loss": 0.8064,
+      "step": 2970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.777095991264855e-05,
+      "loss": 0.771,
+      "step": 2971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.775585368757766e-05,
+      "loss": 0.8241,
+      "step": 2972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.774074379962443e-05,
+      "loss": 0.8514,
+      "step": 2973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.772563025078288e-05,
+      "loss": 0.7622,
+      "step": 2974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.771051304304753e-05,
+      "loss": 0.7291,
+      "step": 2975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.769539217841334e-05,
+      "loss": 0.782,
+      "step": 2976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.768026765887579e-05,
+      "loss": 0.7548,
+      "step": 2977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766513948643084e-05,
+      "loss": 0.8694,
+      "step": 2978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.76500076630749e-05,
+      "loss": 0.9036,
+      "step": 2979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.763487219080491e-05,
+      "loss": 0.847,
+      "step": 2980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.761973307161828e-05,
+      "loss": 0.7937,
+      "step": 2981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.760459030751284e-05,
+      "loss": 0.7399,
+      "step": 2982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.758944390048697e-05,
+      "loss": 0.7055,
+      "step": 2983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.757429385253954e-05,
+      "loss": 0.8648,
+      "step": 2984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.755914016566984e-05,
+      "loss": 0.7485,
+      "step": 2985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.754398284187767e-05,
+      "loss": 0.8243,
+      "step": 2986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.75288218831633e-05,
+      "loss": 0.8278,
+      "step": 2987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.751365729152754e-05,
+      "loss": 0.8688,
+      "step": 2988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.749848906897157e-05,
+      "loss": 0.7857,
+      "step": 2989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.748331721749716e-05,
+      "loss": 0.7992,
+      "step": 2990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.746814173910647e-05,
+      "loss": 0.8354,
+      "step": 2991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.745296263580217e-05,
+      "loss": 0.821,
+      "step": 2992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.743777990958746e-05,
+      "loss": 0.7295,
+      "step": 2993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.742259356246593e-05,
+      "loss": 0.7903,
+      "step": 2994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.740740359644173e-05,
+      "loss": 0.7728,
+      "step": 2995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.739221001351941e-05,
+      "loss": 0.7616,
+      "step": 2996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.737701281570407e-05,
+      "loss": 0.8417,
+      "step": 2997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73618120050012e-05,
+      "loss": 0.7326,
+      "step": 2998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.734660758341687e-05,
+      "loss": 0.7611,
+      "step": 2999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.733139955295756e-05,
+      "loss": 0.8121,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.731618791563026e-05,
+      "loss": 0.8874,
+      "step": 3001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.730097267344238e-05,
+      "loss": 0.8106,
+      "step": 3002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.728575382840186e-05,
+      "loss": 0.6695,
+      "step": 3003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.727053138251711e-05,
+      "loss": 0.8284,
+      "step": 3004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.7255305337797e-05,
+      "loss": 0.7818,
+      "step": 3005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.724007569625088e-05,
+      "loss": 0.6854,
+      "step": 3006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.722484245988857e-05,
+      "loss": 0.7729,
+      "step": 3007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.720960563072035e-05,
+      "loss": 0.8131,
+      "step": 3008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.719436521075704e-05,
+      "loss": 0.7891,
+      "step": 3009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.717912120200986e-05,
+      "loss": 0.7887,
+      "step": 3010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.716387360649051e-05,
+      "loss": 0.7926,
+      "step": 3011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.714862242621121e-05,
+      "loss": 0.7987,
+      "step": 3012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.713336766318461e-05,
+      "loss": 0.8402,
+      "step": 3013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711810931942386e-05,
+      "loss": 0.7945,
+      "step": 3014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.710284739694258e-05,
+      "loss": 0.8237,
+      "step": 3015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.708758189775485e-05,
+      "loss": 0.838,
+      "step": 3016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.707231282387521e-05,
+      "loss": 0.7893,
+      "step": 3017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.705704017731871e-05,
+      "loss": 0.7394,
+      "step": 3018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.704176396010084e-05,
+      "loss": 0.8172,
+      "step": 3019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.702648417423755e-05,
+      "loss": 0.7282,
+      "step": 3020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.701120082174531e-05,
+      "loss": 0.826,
+      "step": 3021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699591390464103e-05,
+      "loss": 0.8357,
+      "step": 3022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.698062342494207e-05,
+      "loss": 0.7575,
+      "step": 3023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.69653293846663e-05,
+      "loss": 0.7582,
+      "step": 3024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.695003178583205e-05,
+      "loss": 0.7163,
+      "step": 3025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.69347306304581e-05,
+      "loss": 0.7197,
+      "step": 3026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.691942592056374e-05,
+      "loss": 0.8136,
+      "step": 3027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.690411765816864e-05,
+      "loss": 0.8459,
+      "step": 3028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.688880584529306e-05,
+      "loss": 0.8218,
+      "step": 3029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.687349048395762e-05,
+      "loss": 0.8131,
+      "step": 3030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.685817157618349e-05,
+      "loss": 0.8177,
+      "step": 3031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.684284912399226e-05,
+      "loss": 0.7785,
+      "step": 3032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.682752312940602e-05,
+      "loss": 0.7186,
+      "step": 3033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681219359444729e-05,
+      "loss": 0.8404,
+      "step": 3034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.679686052113907e-05,
+      "loss": 0.83,
+      "step": 3035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.678152391150489e-05,
+      "loss": 0.8675,
+      "step": 3036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67661837675686e-05,
+      "loss": 0.8402,
+      "step": 3037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.675084009135467e-05,
+      "loss": 0.8035,
+      "step": 3038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.673549288488796e-05,
+      "loss": 0.78,
+      "step": 3039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.672014215019381e-05,
+      "loss": 0.7773,
+      "step": 3040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.670478788929802e-05,
+      "loss": 0.8442,
+      "step": 3041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.668943010422685e-05,
+      "loss": 0.8038,
+      "step": 3042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.667406879700705e-05,
+      "loss": 0.9285,
+      "step": 3043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.665870396966583e-05,
+      "loss": 0.8018,
+      "step": 3044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.664333562423083e-05,
+      "loss": 0.7767,
+      "step": 3045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.662796376273018e-05,
+      "loss": 0.8059,
+      "step": 3046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66125883871925e-05,
+      "loss": 0.7616,
+      "step": 3047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65972094996468e-05,
+      "loss": 0.8285,
+      "step": 3048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.658182710212265e-05,
+      "loss": 0.8274,
+      "step": 3049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656644119665001e-05,
+      "loss": 0.8576,
+      "step": 3050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.655105178525931e-05,
+      "loss": 0.7479,
+      "step": 3051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.653565886998148e-05,
+      "loss": 0.8784,
+      "step": 3052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65202624528479e-05,
+      "loss": 0.8732,
+      "step": 3053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.650486253589039e-05,
+      "loss": 0.8371,
+      "step": 3054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.648945912114124e-05,
+      "loss": 0.8357,
+      "step": 3055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64740522106332e-05,
+      "loss": 0.8206,
+      "step": 3056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64586418063995e-05,
+      "loss": 0.7352,
+      "step": 3057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644322791047382e-05,
+      "loss": 0.6298,
+      "step": 3058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64278105248903e-05,
+      "loss": 0.8906,
+      "step": 3059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.641238965168356e-05,
+      "loss": 0.8073,
+      "step": 3060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.639696529288861e-05,
+      "loss": 0.8233,
+      "step": 3061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.6381537450541e-05,
+      "loss": 0.9365,
+      "step": 3062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.636610612667672e-05,
+      "loss": 0.8505,
+      "step": 3063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.63506713233322e-05,
+      "loss": 0.8541,
+      "step": 3064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.633523304254433e-05,
+      "loss": 0.8935,
+      "step": 3065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.631979128635049e-05,
+      "loss": 0.8618,
+      "step": 3066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.630434605678847e-05,
+      "loss": 0.8472,
+      "step": 3067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.628889735589657e-05,
+      "loss": 0.9115,
+      "step": 3068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.627344518571348e-05,
+      "loss": 0.9593,
+      "step": 3069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.625798954827845e-05,
+      "loss": 0.7841,
+      "step": 3070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.62425304456311e-05,
+      "loss": 0.7666,
+      "step": 3071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.622706787981153e-05,
+      "loss": 0.6909,
+      "step": 3072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.621160185286031e-05,
+      "loss": 0.8414,
+      "step": 3073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.619613236681843e-05,
+      "loss": 0.7735,
+      "step": 3074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.618065942372743e-05,
+      "loss": 0.8006,
+      "step": 3075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.61651830256292e-05,
+      "loss": 0.7644,
+      "step": 3076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.61497031745661e-05,
+      "loss": 0.8705,
+      "step": 3077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.613421987258105e-05,
+      "loss": 0.7604,
+      "step": 3078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.611873312171728e-05,
+      "loss": 0.7686,
+      "step": 3079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.61032429240186e-05,
+      "loss": 0.7414,
+      "step": 3080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.608774928152917e-05,
+      "loss": 0.7817,
+      "step": 3081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.607225219629366e-05,
+      "loss": 0.8065,
+      "step": 3082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.605675167035722e-05,
+      "loss": 0.7585,
+      "step": 3083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.604124770576539e-05,
+      "loss": 0.8412,
+      "step": 3084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.602574030456422e-05,
+      "loss": 0.8278,
+      "step": 3085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.601022946880019e-05,
+      "loss": 0.8536,
+      "step": 3086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.599471520052022e-05,
+      "loss": 0.9133,
+      "step": 3087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.597919750177168e-05,
+      "loss": 0.8426,
+      "step": 3088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.596367637460245e-05,
+      "loss": 0.8603,
+      "step": 3089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.594815182106078e-05,
+      "loss": 0.6734,
+      "step": 3090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.593262384319544e-05,
+      "loss": 0.8372,
+      "step": 3091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.591709244305561e-05,
+      "loss": 0.8439,
+      "step": 3092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.590155762269093e-05,
+      "loss": 0.8073,
+      "step": 3093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.588601938415154e-05,
+      "loss": 0.7425,
+      "step": 3094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.587047772948793e-05,
+      "loss": 0.9171,
+      "step": 3095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.585493266075114e-05,
+      "loss": 0.8281,
+      "step": 3096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.583938417999262e-05,
+      "loss": 0.8613,
+      "step": 3097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.582383228926424e-05,
+      "loss": 0.8085,
+      "step": 3098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.580827699061839e-05,
+      "loss": 0.9178,
+      "step": 3099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.579271828610786e-05,
+      "loss": 0.8605,
+      "step": 3100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.577715617778586e-05,
+      "loss": 0.7229,
+      "step": 3101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.576159066770613e-05,
+      "loss": 0.9288,
+      "step": 3102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.574602175792282e-05,
+      "loss": 0.7272,
+      "step": 3103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.57304494504905e-05,
+      "loss": 0.8034,
+      "step": 3104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.571487374746425e-05,
+      "loss": 0.6905,
+      "step": 3105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.569929465089953e-05,
+      "loss": 0.7804,
+      "step": 3106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.568371216285229e-05,
+      "loss": 0.8417,
+      "step": 3107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.566812628537893e-05,
+      "loss": 0.8936,
+      "step": 3108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.56525370205363e-05,
+      "loss": 0.7789,
+      "step": 3109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.563694437038162e-05,
+      "loss": 0.8162,
+      "step": 3110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.562134833697267e-05,
+      "loss": 0.8315,
+      "step": 3111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.560574892236764e-05,
+      "loss": 0.8934,
+      "step": 3112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.559014612862512e-05,
+      "loss": 0.9113,
+      "step": 3113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55745399578042e-05,
+      "loss": 0.8435,
+      "step": 3114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.555893041196436e-05,
+      "loss": 0.7999,
+      "step": 3115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.554331749316559e-05,
+      "loss": 0.8585,
+      "step": 3116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.552770120346828e-05,
+      "loss": 0.7932,
+      "step": 3117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55120815449333e-05,
+      "loss": 0.8293,
+      "step": 3118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.54964585196219e-05,
+      "loss": 0.7698,
+      "step": 3119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.548083212959588e-05,
+      "loss": 0.8614,
+      "step": 3120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.546520237691737e-05,
+      "loss": 0.7927,
+      "step": 3121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.544956926364901e-05,
+      "loss": 0.8203,
+      "step": 3122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.54339327918539e-05,
+      "loss": 0.7824,
+      "step": 3123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.541829296359551e-05,
+      "loss": 0.8113,
+      "step": 3124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.540264978093783e-05,
+      "loss": 0.7946,
+      "step": 3125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.538700324594521e-05,
+      "loss": 0.8508,
+      "step": 3126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.537135336068255e-05,
+      "loss": 0.7553,
+      "step": 3127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.535570012721509e-05,
+      "loss": 0.8517,
+      "step": 3128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.534004354760857e-05,
+      "loss": 0.7772,
+      "step": 3129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.532438362392917e-05,
+      "loss": 0.8319,
+      "step": 3130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.530872035824345e-05,
+      "loss": 0.8712,
+      "step": 3131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.529305375261853e-05,
+      "loss": 0.8512,
+      "step": 3132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.527738380912186e-05,
+      "loss": 0.7643,
+      "step": 3133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.526171052982136e-05,
+      "loss": 0.6438,
+      "step": 3134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.524603391678541e-05,
+      "loss": 0.8449,
+      "step": 3135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.523035397208282e-05,
+      "loss": 0.8346,
+      "step": 3136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.521467069778284e-05,
+      "loss": 0.8409,
+      "step": 3137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.519898409595516e-05,
+      "loss": 0.9298,
+      "step": 3138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.518329416866992e-05,
+      "loss": 0.8696,
+      "step": 3139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.516760091799766e-05,
+      "loss": 0.92,
+      "step": 3140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.51519043460094e-05,
+      "loss": 0.671,
+      "step": 3141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.51362044547766e-05,
+      "loss": 0.8663,
+      "step": 3142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.512050124637114e-05,
+      "loss": 0.8267,
+      "step": 3143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.51047947228653e-05,
+      "loss": 0.8009,
+      "step": 3144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.508908488633186e-05,
+      "loss": 0.7389,
+      "step": 3145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.507337173884404e-05,
+      "loss": 0.8555,
+      "step": 3146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.505765528247544e-05,
+      "loss": 0.8031,
+      "step": 3147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.504193551930014e-05,
+      "loss": 0.8337,
+      "step": 3148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.502621245139266e-05,
+      "loss": 0.7333,
+      "step": 3149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.501048608082791e-05,
+      "loss": 0.7768,
+      "step": 3150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.499475640968126e-05,
+      "loss": 0.7915,
+      "step": 3151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.497902344002858e-05,
+      "loss": 0.9091,
+      "step": 3152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.496328717394605e-05,
+      "loss": 0.7946,
+      "step": 3153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.494754761351039e-05,
+      "loss": 0.771,
+      "step": 3154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.493180476079871e-05,
+      "loss": 0.8236,
+      "step": 3155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.491605861788855e-05,
+      "loss": 0.842,
+      "step": 3156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.490030918685792e-05,
+      "loss": 0.8611,
+      "step": 3157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.488455646978523e-05,
+      "loss": 0.8384,
+      "step": 3158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.486880046874934e-05,
+      "loss": 0.7283,
+      "step": 3159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.485304118582948e-05,
+      "loss": 0.7324,
+      "step": 3160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.483727862310545e-05,
+      "loss": 0.7928,
+      "step": 3161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.482151278265737e-05,
+      "loss": 0.8134,
+      "step": 3162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.480574366656581e-05,
+      "loss": 0.7777,
+      "step": 3163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.478997127691181e-05,
+      "loss": 0.8451,
+      "step": 3164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.47741956157768e-05,
+      "loss": 0.754,
+      "step": 3165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.475841668524268e-05,
+      "loss": 0.7604,
+      "step": 3166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.474263448739175e-05,
+      "loss": 0.8845,
+      "step": 3167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.472684902430678e-05,
+      "loss": 0.8583,
+      "step": 3168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.47110602980709e-05,
+      "loss": 0.8264,
+      "step": 3169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.469526831076774e-05,
+      "loss": 0.8253,
+      "step": 3170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.467947306448133e-05,
+      "loss": 0.8839,
+      "step": 3171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.466367456129616e-05,
+      "loss": 0.8646,
+      "step": 3172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.464787280329708e-05,
+      "loss": 0.7503,
+      "step": 3173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.463206779256946e-05,
+      "loss": 0.8428,
+      "step": 3174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.461625953119901e-05,
+      "loss": 0.7898,
+      "step": 3175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.460044802127197e-05,
+      "loss": 0.709,
+      "step": 3176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.458463326487491e-05,
+      "loss": 0.7258,
+      "step": 3177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.456881526409489e-05,
+      "loss": 0.773,
+      "step": 3178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.455299402101936e-05,
+      "loss": 0.8354,
+      "step": 3179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.453716953773622e-05,
+      "loss": 0.7905,
+      "step": 3180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.452134181633381e-05,
+      "loss": 0.798,
+      "step": 3181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.450551085890087e-05,
+      "loss": 0.8089,
+      "step": 3182
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.448967666752657e-05,
+      "loss": 0.6723,
+      "step": 3183
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.447383924430055e-05,
+      "loss": 0.8954,
+      "step": 3184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.44579985913128e-05,
+      "loss": 0.8073,
+      "step": 3185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.444215471065378e-05,
+      "loss": 0.8191,
+      "step": 3186
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.442630760441443e-05,
+      "loss": 0.7902,
+      "step": 3187
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.441045727468601e-05,
+      "loss": 0.7842,
+      "step": 3188
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.439460372356025e-05,
+      "loss": 0.7282,
+      "step": 3189
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.437874695312933e-05,
+      "loss": 0.8762,
+      "step": 3190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.436288696548582e-05,
+      "loss": 0.6881,
+      "step": 3191
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.434702376272274e-05,
+      "loss": 0.7959,
+      "step": 3192
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.433115734693354e-05,
+      "loss": 0.7944,
+      "step": 3193
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.431528772021206e-05,
+      "loss": 0.8273,
+      "step": 3194
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.429941488465257e-05,
+      "loss": 0.813,
+      "step": 3195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.42835388423498e-05,
+      "loss": 0.8337,
+      "step": 3196
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.426765959539887e-05,
+      "loss": 0.9494,
+      "step": 3197
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.425177714589533e-05,
+      "loss": 0.8519,
+      "step": 3198
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.423589149593513e-05,
+      "loss": 0.7852,
+      "step": 3199
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.42200026476147e-05,
+      "loss": 0.8548,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.420411060303084e-05,
+      "loss": 0.7839,
+      "step": 3201
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.418821536428081e-05,
+      "loss": 0.7429,
+      "step": 3202
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.417231693346224e-05,
+      "loss": 0.7836,
+      "step": 3203
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415641531267325e-05,
+      "loss": 0.7857,
+      "step": 3204
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.414051050401232e-05,
+      "loss": 0.8129,
+      "step": 3205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.412460250957837e-05,
+      "loss": 0.8038,
+      "step": 3206
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.410869133147076e-05,
+      "loss": 0.9237,
+      "step": 3207
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.409277697178926e-05,
+      "loss": 0.7879,
+      "step": 3208
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.407685943263403e-05,
+      "loss": 0.7783,
+      "step": 3209
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.40609387161057e-05,
+      "loss": 0.749,
+      "step": 3210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.404501482430526e-05,
+      "loss": 0.8874,
+      "step": 3211
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.402908775933419e-05,
+      "loss": 0.7852,
+      "step": 3212
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.401315752329435e-05,
+      "loss": 0.71,
+      "step": 3213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.399722411828798e-05,
+      "loss": 0.7885,
+      "step": 3214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.398128754641784e-05,
+      "loss": 0.7858,
+      "step": 3215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3965347809787e-05,
+      "loss": 0.7699,
+      "step": 3216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.394940491049899e-05,
+      "loss": 0.8033,
+      "step": 3217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.393345885065777e-05,
+      "loss": 0.8598,
+      "step": 3218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.391750963236774e-05,
+      "loss": 0.7914,
+      "step": 3219
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.390155725773365e-05,
+      "loss": 0.7593,
+      "step": 3220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388560172886072e-05,
+      "loss": 0.7944,
+      "step": 3221
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.386964304785456e-05,
+      "loss": 0.7767,
+      "step": 3222
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.38536812168212e-05,
+      "loss": 0.7799,
+      "step": 3223
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.383771623786709e-05,
+      "loss": 0.8327,
+      "step": 3224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.382174811309911e-05,
+      "loss": 0.8457,
+      "step": 3225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.380577684462453e-05,
+      "loss": 0.7966,
+      "step": 3226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.378980243455102e-05,
+      "loss": 0.8701,
+      "step": 3227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.377382488498675e-05,
+      "loss": 0.9068,
+      "step": 3228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.375784419804019e-05,
+      "loss": 0.8546,
+      "step": 3229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.37418603758203e-05,
+      "loss": 0.7729,
+      "step": 3230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.372587342043642e-05,
+      "loss": 0.7939,
+      "step": 3231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.370988333399834e-05,
+      "loss": 0.7118,
+      "step": 3232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.369389011861621e-05,
+      "loss": 0.7987,
+      "step": 3233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.367789377640063e-05,
+      "loss": 0.7841,
+      "step": 3234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.366189430946262e-05,
+      "loss": 0.8522,
+      "step": 3235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.364589171991358e-05,
+      "loss": 0.7858,
+      "step": 3236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.362988600986534e-05,
+      "loss": 0.7592,
+      "step": 3237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.361387718143014e-05,
+      "loss": 0.869,
+      "step": 3238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.359786523672063e-05,
+      "loss": 0.7673,
+      "step": 3239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.358185017784989e-05,
+      "loss": 0.8331,
+      "step": 3240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.356583200693137e-05,
+      "loss": 0.8457,
+      "step": 3241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.354981072607899e-05,
+      "loss": 0.8174,
+      "step": 3242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3533786337407e-05,
+      "loss": 0.835,
+      "step": 3243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.351775884303013e-05,
+      "loss": 0.7307,
+      "step": 3244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.35017282450635e-05,
+      "loss": 0.8721,
+      "step": 3245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.348569454562261e-05,
+      "loss": 0.7685,
+      "step": 3246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.346965774682342e-05,
+      "loss": 0.8297,
+      "step": 3247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.345361785078226e-05,
+      "loss": 0.7021,
+      "step": 3248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.343757485961589e-05,
+      "loss": 0.8165,
+      "step": 3249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.342152877544147e-05,
+      "loss": 0.8442,
+      "step": 3250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.340547960037656e-05,
+      "loss": 0.7448,
+      "step": 3251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.338942733653915e-05,
+      "loss": 0.6256,
+      "step": 3252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.33733719860476e-05,
+      "loss": 0.8029,
+      "step": 3253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.335731355102071e-05,
+      "loss": 0.8159,
+      "step": 3254
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.334125203357768e-05,
+      "loss": 0.9278,
+      "step": 3255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.332518743583814e-05,
+      "loss": 0.8183,
+      "step": 3256
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.330911975992208e-05,
+      "loss": 0.748,
+      "step": 3257
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.329304900794991e-05,
+      "loss": 0.7197,
+      "step": 3258
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.327697518204247e-05,
+      "loss": 0.717,
+      "step": 3259
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.326089828432096e-05,
+      "loss": 0.7422,
+      "step": 3260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.324481831690707e-05,
+      "loss": 0.7925,
+      "step": 3261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.322873528192277e-05,
+      "loss": 0.7646,
+      "step": 3262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.321264918149055e-05,
+      "loss": 0.7686,
+      "step": 3263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.319656001773326e-05,
+      "loss": 0.8579,
+      "step": 3264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.318046779277415e-05,
+      "loss": 0.8352,
+      "step": 3265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.316437250873686e-05,
+      "loss": 0.801,
+      "step": 3266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.314827416774547e-05,
+      "loss": 0.8283,
+      "step": 3267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.313217277192442e-05,
+      "loss": 0.8386,
+      "step": 3268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.311606832339861e-05,
+      "loss": 0.9383,
+      "step": 3269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.30999608242933e-05,
+      "loss": 0.8896,
+      "step": 3270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.308385027673414e-05,
+      "loss": 0.791,
+      "step": 3271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.306773668284724e-05,
+      "loss": 0.6857,
+      "step": 3272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.305162004475906e-05,
+      "loss": 0.7914,
+      "step": 3273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.303550036459646e-05,
+      "loss": 0.8018,
+      "step": 3274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.301937764448675e-05,
+      "loss": 0.795,
+      "step": 3275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.300325188655761e-05,
+      "loss": 0.8195,
+      "step": 3276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.298712309293711e-05,
+      "loss": 0.7433,
+      "step": 3277
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.297099126575375e-05,
+      "loss": 0.9434,
+      "step": 3278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.295485640713639e-05,
+      "loss": 0.778,
+      "step": 3279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.293871851921434e-05,
+      "loss": 0.8213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.292257760411726e-05,
+      "loss": 0.8784,
+      "step": 3281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.290643366397525e-05,
+      "loss": 0.6899,
+      "step": 3282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.289028670091877e-05,
+      "loss": 0.734,
+      "step": 3283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.287413671707875e-05,
+      "loss": 0.8703,
+      "step": 3284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.28579837145864e-05,
+      "loss": 0.8846,
+      "step": 3285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.284182769557346e-05,
+      "loss": 0.9235,
+      "step": 3286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.282566866217197e-05,
+      "loss": 0.8515,
+      "step": 3287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.280950661651442e-05,
+      "loss": 0.809,
+      "step": 3288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.279334156073367e-05,
+      "loss": 0.8155,
+      "step": 3289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.277717349696301e-05,
+      "loss": 0.8177,
+      "step": 3290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.276100242733607e-05,
+      "loss": 0.7797,
+      "step": 3291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.274482835398695e-05,
+      "loss": 0.8338,
+      "step": 3292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.272865127905008e-05,
+      "loss": 0.7997,
+      "step": 3293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.271247120466034e-05,
+      "loss": 0.7491,
+      "step": 3294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.269628813295296e-05,
+      "loss": 0.7939,
+      "step": 3295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.268010206606361e-05,
+      "loss": 0.8008,
+      "step": 3296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.26639130061283e-05,
+      "loss": 0.7999,
+      "step": 3297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264772095528348e-05,
+      "loss": 0.8232,
+      "step": 3298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2631525915666e-05,
+      "loss": 0.7931,
+      "step": 3299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.261532788941306e-05,
+      "loss": 0.8349,
+      "step": 3300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.259912687866228e-05,
+      "loss": 0.7897,
+      "step": 3301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.25829228855517e-05,
+      "loss": 0.8416,
+      "step": 3302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.25667159122197e-05,
+      "loss": 0.9401,
+      "step": 3303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.255050596080509e-05,
+      "loss": 0.8571,
+      "step": 3304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.253429303344707e-05,
+      "loss": 0.8157,
+      "step": 3305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.251807713228523e-05,
+      "loss": 0.8685,
+      "step": 3306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.250185825945954e-05,
+      "loss": 0.8259,
+      "step": 3307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.248563641711036e-05,
+      "loss": 0.8186,
+      "step": 3308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.246941160737846e-05,
+      "loss": 0.8803,
+      "step": 3309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2453183832405e-05,
+      "loss": 0.7506,
+      "step": 3310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.243695309433153e-05,
+      "loss": 0.8162,
+      "step": 3311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.242071939529999e-05,
+      "loss": 0.7273,
+      "step": 3312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.240448273745268e-05,
+      "loss": 0.79,
+      "step": 3313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.238824312293235e-05,
+      "loss": 0.8174,
+      "step": 3314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.23720005538821e-05,
+      "loss": 0.86,
+      "step": 3315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.235575503244541e-05,
+      "loss": 0.8734,
+      "step": 3316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.233950656076618e-05,
+      "loss": 0.8275,
+      "step": 3317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.232325514098869e-05,
+      "loss": 0.7234,
+      "step": 3318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.230700077525761e-05,
+      "loss": 0.85,
+      "step": 3319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.229074346571799e-05,
+      "loss": 0.8834,
+      "step": 3320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.227448321451526e-05,
+      "loss": 0.7814,
+      "step": 3321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.225822002379526e-05,
+      "loss": 0.7979,
+      "step": 3322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.224195389570422e-05,
+      "loss": 0.8002,
+      "step": 3323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.222568483238874e-05,
+      "loss": 0.7078,
+      "step": 3324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.220941283599582e-05,
+      "loss": 0.8198,
+      "step": 3325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.219313790867282e-05,
+      "loss": 0.8081,
+      "step": 3326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.217686005256754e-05,
+      "loss": 0.6869,
+      "step": 3327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.216057926982811e-05,
+      "loss": 0.7729,
+      "step": 3328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.214429556260308e-05,
+      "loss": 0.8217,
+      "step": 3329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.21280089330414e-05,
+      "loss": 0.7827,
+      "step": 3330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.211171938329233e-05,
+      "loss": 0.8557,
+      "step": 3331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.20954269155056e-05,
+      "loss": 0.8202,
+      "step": 3332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.207913153183131e-05,
+      "loss": 0.838,
+      "step": 3333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.20628332344199e-05,
+      "loss": 0.8305,
+      "step": 3334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.204653202542222e-05,
+      "loss": 0.8213,
+      "step": 3335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.203022790698953e-05,
+      "loss": 0.7879,
+      "step": 3336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.201392088127343e-05,
+      "loss": 0.733,
+      "step": 3337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.199761095042595e-05,
+      "loss": 0.8159,
+      "step": 3338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.198129811659945e-05,
+      "loss": 0.7297,
+      "step": 3339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.196498238194672e-05,
+      "loss": 0.7195,
+      "step": 3340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.194866374862091e-05,
+      "loss": 0.8048,
+      "step": 3341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.193234221877552e-05,
+      "loss": 0.7706,
+      "step": 3342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.191601779456452e-05,
+      "loss": 0.7455,
+      "step": 3343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189969047814219e-05,
+      "loss": 0.7514,
+      "step": 3344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.18833602716632e-05,
+      "loss": 0.8199,
+      "step": 3345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.186702717728264e-05,
+      "loss": 0.8043,
+      "step": 3346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.185069119715593e-05,
+      "loss": 0.8472,
+      "step": 3347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.183435233343892e-05,
+      "loss": 0.7303,
+      "step": 3348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.181801058828778e-05,
+      "loss": 0.7854,
+      "step": 3349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.180166596385914e-05,
+      "loss": 0.8265,
+      "step": 3350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.178531846230994e-05,
+      "loss": 0.8299,
+      "step": 3351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.176896808579751e-05,
+      "loss": 0.8774,
+      "step": 3352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.175261483647962e-05,
+      "loss": 0.8056,
+      "step": 3353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.173625871651432e-05,
+      "loss": 0.8621,
+      "step": 3354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.171989972806014e-05,
+      "loss": 0.8703,
+      "step": 3355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.170353787327592e-05,
+      "loss": 0.7057,
+      "step": 3356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.16871731543209e-05,
+      "loss": 0.8113,
+      "step": 3357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167080557335471e-05,
+      "loss": 0.8159,
+      "step": 3358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.165443513253733e-05,
+      "loss": 0.8623,
+      "step": 3359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.163806183402916e-05,
+      "loss": 0.7632,
+      "step": 3360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.162168567999092e-05,
+      "loss": 0.7806,
+      "step": 3361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.160530667258373e-05,
+      "loss": 0.8887,
+      "step": 3362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.158892481396912e-05,
+      "loss": 0.8743,
+      "step": 3363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.157254010630896e-05,
+      "loss": 0.756,
+      "step": 3364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.155615255176551e-05,
+      "loss": 0.8554,
+      "step": 3365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.15397621525014e-05,
+      "loss": 0.8092,
+      "step": 3366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.152336891067963e-05,
+      "loss": 0.7502,
+      "step": 3367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.15069728284636e-05,
+      "loss": 0.824,
+      "step": 3368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.149057390801704e-05,
+      "loss": 0.8137,
+      "step": 3369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.14741721515041e-05,
+      "loss": 0.8149,
+      "step": 3370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.145776756108929e-05,
+      "loss": 0.7604,
+      "step": 3371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.144136013893745e-05,
+      "loss": 0.8135,
+      "step": 3372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.14249498872139e-05,
+      "loss": 0.8246,
+      "step": 3373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.140853680808421e-05,
+      "loss": 0.802,
+      "step": 3374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.139212090371442e-05,
+      "loss": 0.7505,
+      "step": 3375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.137570217627088e-05,
+      "loss": 0.8339,
+      "step": 3376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.135928062792032e-05,
+      "loss": 0.8675,
+      "step": 3377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.134285626082991e-05,
+      "loss": 0.7366,
+      "step": 3378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.13264290771671e-05,
+      "loss": 0.7843,
+      "step": 3379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.130999907909971e-05,
+      "loss": 0.7836,
+      "step": 3380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.129356626879606e-05,
+      "loss": 0.6865,
+      "step": 3381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.127713064842471e-05,
+      "loss": 0.7306,
+      "step": 3382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.126069222015464e-05,
+      "loss": 0.7189,
+      "step": 3383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12442509861552e-05,
+      "loss": 0.8553,
+      "step": 3384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.122780694859609e-05,
+      "loss": 0.7922,
+      "step": 3385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12113601096474e-05,
+      "loss": 0.8087,
+      "step": 3386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.119491047147962e-05,
+      "loss": 0.9131,
+      "step": 3387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.117845803626351e-05,
+      "loss": 0.8448,
+      "step": 3388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.116200280617033e-05,
+      "loss": 0.7854,
+      "step": 3389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.114554478337158e-05,
+      "loss": 0.8313,
+      "step": 3390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.112908397003923e-05,
+      "loss": 0.8292,
+      "step": 3391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.111262036834558e-05,
+      "loss": 0.8067,
+      "step": 3392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.109615398046329e-05,
+      "loss": 0.8802,
+      "step": 3393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.107968480856539e-05,
+      "loss": 0.8435,
+      "step": 3394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.106321285482528e-05,
+      "loss": 0.8235,
+      "step": 3395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.104673812141675e-05,
+      "loss": 0.8782,
+      "step": 3396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.103026061051393e-05,
+      "loss": 0.7864,
+      "step": 3397
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.101378032429129e-05,
+      "loss": 0.8183,
+      "step": 3398
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.099729726492373e-05,
+      "loss": 0.828,
+      "step": 3399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.098081143458648e-05,
+      "loss": 0.7833,
+      "step": 3400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.096432283545514e-05,
+      "loss": 0.8219,
+      "step": 3401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.094783146970568e-05,
+      "loss": 0.792,
+      "step": 3402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.093133733951444e-05,
+      "loss": 0.6659,
+      "step": 3403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.091484044705809e-05,
+      "loss": 0.7569,
+      "step": 3404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.08983407945137e-05,
+      "loss": 0.856,
+      "step": 3405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.088183838405871e-05,
+      "loss": 0.7409,
+      "step": 3406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.086533321787089e-05,
+      "loss": 0.79,
+      "step": 3407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.084882529812839e-05,
+      "loss": 0.761,
+      "step": 3408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.083231462700975e-05,
+      "loss": 0.8497,
+      "step": 3409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.081580120669384e-05,
+      "loss": 0.7982,
+      "step": 3410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.079928503935988e-05,
+      "loss": 0.7809,
+      "step": 3411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.078276612718752e-05,
+      "loss": 0.7626,
+      "step": 3412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.076624447235669e-05,
+      "loss": 0.7762,
+      "step": 3413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.07497200770477e-05,
+      "loss": 0.7692,
+      "step": 3414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.07331929434413e-05,
+      "loss": 0.8854,
+      "step": 3415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.071666307371847e-05,
+      "loss": 0.7455,
+      "step": 3416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.070013047006068e-05,
+      "loss": 0.7678,
+      "step": 3417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.068359513464968e-05,
+      "loss": 0.8596,
+      "step": 3418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.066705706966762e-05,
+      "loss": 0.8173,
+      "step": 3419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.065051627729698e-05,
+      "loss": 0.8148,
+      "step": 3420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.063397275972059e-05,
+      "loss": 0.7623,
+      "step": 3421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.061742651912172e-05,
+      "loss": 0.7343,
+      "step": 3422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.060087755768389e-05,
+      "loss": 0.8083,
+      "step": 3423
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.058432587759108e-05,
+      "loss": 0.7075,
+      "step": 3424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.056777148102753e-05,
+      "loss": 0.8444,
+      "step": 3425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.055121437017792e-05,
+      "loss": 0.8557,
+      "step": 3426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053465454722725e-05,
+      "loss": 0.8268,
+      "step": 3427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.05180920143609e-05,
+      "loss": 0.7687,
+      "step": 3428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.050152677376458e-05,
+      "loss": 0.8224,
+      "step": 3429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.048495882762436e-05,
+      "loss": 0.8353,
+      "step": 3430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.04683881781267e-05,
+      "loss": 0.8038,
+      "step": 3431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.045181482745838e-05,
+      "loss": 0.7875,
+      "step": 3432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.043523877780654e-05,
+      "loss": 0.8419,
+      "step": 3433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.041866003135871e-05,
+      "loss": 0.7952,
+      "step": 3434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.040207859030274e-05,
+      "loss": 0.8663,
+      "step": 3435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.038549445682685e-05,
+      "loss": 0.7655,
+      "step": 3436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.036890763311962e-05,
+      "loss": 0.8005,
+      "step": 3437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.035231812136998e-05,
+      "loss": 0.7624,
+      "step": 3438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.03357259237672e-05,
+      "loss": 0.7643,
+      "step": 3439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.031913104250091e-05,
+      "loss": 0.8258,
+      "step": 3440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.030253347976112e-05,
+      "loss": 0.7902,
+      "step": 3441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02859332377382e-05,
+      "loss": 0.813,
+      "step": 3442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02693303186228e-05,
+      "loss": 0.7118,
+      "step": 3443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.025272472460597e-05,
+      "loss": 0.7698,
+      "step": 3444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.023611645787917e-05,
+      "loss": 0.8011,
+      "step": 3445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.021950552063413e-05,
+      "loss": 0.7946,
+      "step": 3446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.020289191506297e-05,
+      "loss": 0.8006,
+      "step": 3447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.018627564335812e-05,
+      "loss": 0.8093,
+      "step": 3448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.016965670771243e-05,
+      "loss": 0.8089,
+      "step": 3449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.015303511031905e-05,
+      "loss": 0.839,
+      "step": 3450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.01364108533715e-05,
+      "loss": 0.8279,
+      "step": 3451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.011978393906366e-05,
+      "loss": 0.8337,
+      "step": 3452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.010315436958972e-05,
+      "loss": 0.84,
+      "step": 3453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.008652214714428e-05,
+      "loss": 0.8025,
+      "step": 3454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.006988727392224e-05,
+      "loss": 0.7288,
+      "step": 3455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.005324975211889e-05,
+      "loss": 0.7432,
+      "step": 3456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.003660958392983e-05,
+      "loss": 0.7011,
+      "step": 3457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.001996677155105e-05,
+      "loss": 0.744,
+      "step": 3458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.000332131717885e-05,
+      "loss": 0.776,
+      "step": 3459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.99866732230099e-05,
+      "loss": 0.7971,
+      "step": 3460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.99700224912412e-05,
+      "loss": 0.7962,
+      "step": 3461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.995336912407014e-05,
+      "loss": 0.824,
+      "step": 3462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.99367131236944e-05,
+      "loss": 0.8529,
+      "step": 3463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.992005449231208e-05,
+      "loss": 0.8227,
+      "step": 3464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.990339323212155e-05,
+      "loss": 0.7556,
+      "step": 3465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.988672934532157e-05,
+      "loss": 0.764,
+      "step": 3466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.987006283411124e-05,
+      "loss": 0.84,
+      "step": 3467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.985339370068999e-05,
+      "loss": 0.7377,
+      "step": 3468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.983672194725766e-05,
+      "loss": 0.8565,
+      "step": 3469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.982004757601433e-05,
+      "loss": 0.7561,
+      "step": 3470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.980337058916048e-05,
+      "loss": 0.8353,
+      "step": 3471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9786690988897e-05,
+      "loss": 0.8635,
+      "step": 3472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.977000877742501e-05,
+      "loss": 0.8355,
+      "step": 3473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.975332395694605e-05,
+      "loss": 0.7678,
+      "step": 3474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.973663652966196e-05,
+      "loss": 0.7342,
+      "step": 3475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.971994649777497e-05,
+      "loss": 0.731,
+      "step": 3476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.970325386348761e-05,
+      "loss": 0.8154,
+      "step": 3477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.968655862900278e-05,
+      "loss": 0.9694,
+      "step": 3478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.96698607965237e-05,
+      "loss": 0.7348,
+      "step": 3479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.965316036825397e-05,
+      "loss": 0.7203,
+      "step": 3480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.96364573463975e-05,
+      "loss": 0.7727,
+      "step": 3481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.961975173315855e-05,
+      "loss": 0.6511,
+      "step": 3482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.960304353074175e-05,
+      "loss": 0.8492,
+      "step": 3483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.958633274135202e-05,
+      "loss": 0.8549,
+      "step": 3484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.956961936719466e-05,
+      "loss": 0.7906,
+      "step": 3485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.95529034104753e-05,
+      "loss": 0.8539,
+      "step": 3486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.953618487339988e-05,
+      "loss": 0.8111,
+      "step": 3487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.951946375817474e-05,
+      "loss": 0.7217,
+      "step": 3488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.950274006700655e-05,
+      "loss": 0.87,
+      "step": 3489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.948601380210225e-05,
+      "loss": 0.7925,
+      "step": 3490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.946928496566921e-05,
+      "loss": 0.7374,
+      "step": 3491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.94525535599151e-05,
+      "loss": 0.8881,
+      "step": 3492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.94358195870479e-05,
+      "loss": 0.8508,
+      "step": 3493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.941908304927597e-05,
+      "loss": 0.7921,
+      "step": 3494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.940234394880802e-05,
+      "loss": 0.842,
+      "step": 3495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.938560228785304e-05,
+      "loss": 0.7978,
+      "step": 3496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93688580686204e-05,
+      "loss": 0.7613,
+      "step": 3497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.935211129331983e-05,
+      "loss": 0.8331,
+      "step": 3498
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.933536196416131e-05,
+      "loss": 0.7481,
+      "step": 3499
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.931861008335527e-05,
+      "loss": 0.7773,
+      "step": 3500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93018556531124e-05,
+      "loss": 0.7258,
+      "step": 3501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.928509867564374e-05,
+      "loss": 0.8178,
+      "step": 3502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.926833915316069e-05,
+      "loss": 0.7429,
+      "step": 3503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.925157708787493e-05,
+      "loss": 0.8026,
+      "step": 3504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.923481248199857e-05,
+      "loss": 0.7363,
+      "step": 3505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.921804533774396e-05,
+      "loss": 0.7885,
+      "step": 3506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.920127565732385e-05,
+      "loss": 0.8159,
+      "step": 3507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.918450344295129e-05,
+      "loss": 0.728,
+      "step": 3508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.916772869683966e-05,
+      "loss": 0.7428,
+      "step": 3509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.915095142120271e-05,
+      "loss": 0.8789,
+      "step": 3510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.91341716182545e-05,
+      "loss": 0.8175,
+      "step": 3511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.911738929020941e-05,
+      "loss": 0.8572,
+      "step": 3512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.91006044392822e-05,
+      "loss": 0.8146,
+      "step": 3513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.908381706768787e-05,
+      "loss": 0.782,
+      "step": 3514
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.906702717764187e-05,
+      "loss": 0.8333,
+      "step": 3515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.905023477135992e-05,
+      "loss": 0.7335,
+      "step": 3516
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.903343985105807e-05,
+      "loss": 0.74,
+      "step": 3517
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.901664241895272e-05,
+      "loss": 0.8315,
+      "step": 3518
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.899984247726057e-05,
+      "loss": 0.8597,
+      "step": 3519
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.898304002819869e-05,
+      "loss": 0.8478,
+      "step": 3520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.896623507398447e-05,
+      "loss": 0.822,
+      "step": 3521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.894942761683561e-05,
+      "loss": 0.836,
+      "step": 3522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.893261765897016e-05,
+      "loss": 0.7977,
+      "step": 3523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.891580520260648e-05,
+      "loss": 0.8164,
+      "step": 3524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.889899024996332e-05,
+      "loss": 0.7609,
+      "step": 3525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.888217280325967e-05,
+      "loss": 0.792,
+      "step": 3526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.886535286471492e-05,
+      "loss": 0.7779,
+      "step": 3527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.884853043654876e-05,
+      "loss": 0.8345,
+      "step": 3528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.883170552098117e-05,
+      "loss": 0.8637,
+      "step": 3529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.881487812023256e-05,
+      "loss": 0.7825,
+      "step": 3530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.879804823652357e-05,
+      "loss": 0.7532,
+      "step": 3531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.878121587207522e-05,
+      "loss": 0.7884,
+      "step": 3532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.876438102910883e-05,
+      "loss": 0.8498,
+      "step": 3533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.874754370984606e-05,
+      "loss": 0.8309,
+      "step": 3534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.87307039165089e-05,
+      "loss": 0.7364,
+      "step": 3535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.871386165131968e-05,
+      "loss": 0.8246,
+      "step": 3536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8697016916501e-05,
+      "loss": 0.8821,
+      "step": 3537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.868016971427588e-05,
+      "loss": 0.717,
+      "step": 3538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.866332004686755e-05,
+      "loss": 0.8048,
+      "step": 3539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.864646791649966e-05,
+      "loss": 0.7266,
+      "step": 3540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.862961332539615e-05,
+      "loss": 0.8242,
+      "step": 3541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.861275627578127e-05,
+      "loss": 0.8231,
+      "step": 3542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.859589676987962e-05,
+      "loss": 0.806,
+      "step": 3543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.857903480991611e-05,
+      "loss": 0.7675,
+      "step": 3544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.856217039811599e-05,
+      "loss": 0.8547,
+      "step": 3545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.85453035367048e-05,
+      "loss": 0.7564,
+      "step": 3546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.852843422790845e-05,
+      "loss": 0.8142,
+      "step": 3547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.851156247395312e-05,
+      "loss": 0.8112,
+      "step": 3548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.849468827706537e-05,
+      "loss": 0.6812,
+      "step": 3549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.847781163947205e-05,
+      "loss": 0.8643,
+      "step": 3550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.846093256340029e-05,
+      "loss": 0.734,
+      "step": 3551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.844405105107763e-05,
+      "loss": 0.7982,
+      "step": 3552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.842716710473187e-05,
+      "loss": 0.8041,
+      "step": 3553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.841028072659117e-05,
+      "loss": 0.7551,
+      "step": 3554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.839339191888398e-05,
+      "loss": 0.7397,
+      "step": 3555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.837650068383907e-05,
+      "loss": 0.8003,
+      "step": 3556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.835960702368555e-05,
+      "loss": 0.7334,
+      "step": 3557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.834271094065283e-05,
+      "loss": 0.7803,
+      "step": 3558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.832581243697067e-05,
+      "loss": 0.8149,
+      "step": 3559
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.830891151486913e-05,
+      "loss": 0.7256,
+      "step": 3560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.829200817657854e-05,
+      "loss": 0.8128,
+      "step": 3561
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.827510242432968e-05,
+      "loss": 0.7471,
+      "step": 3562
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.825819426035351e-05,
+      "loss": 0.7049,
+      "step": 3563
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.824128368688139e-05,
+      "loss": 0.7811,
+      "step": 3564
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.822437070614496e-05,
+      "loss": 0.8287,
+      "step": 3565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82074553203762e-05,
+      "loss": 0.8903,
+      "step": 3566
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.819053753180739e-05,
+      "loss": 0.8059,
+      "step": 3567
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.817361734267114e-05,
+      "loss": 0.7516,
+      "step": 3568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.815669475520037e-05,
+      "loss": 0.8502,
+      "step": 3569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.813976977162831e-05,
+      "loss": 0.8077,
+      "step": 3570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.812284239418854e-05,
+      "loss": 0.7964,
+      "step": 3571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.810591262511492e-05,
+      "loss": 0.754,
+      "step": 3572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.808898046664161e-05,
+      "loss": 0.75,
+      "step": 3573
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.807204592100315e-05,
+      "loss": 0.7361,
+      "step": 3574
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.805510899043435e-05,
+      "loss": 0.8142,
+      "step": 3575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.803816967717032e-05,
+      "loss": 0.7966,
+      "step": 3576
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.802122798344653e-05,
+      "loss": 0.795,
+      "step": 3577
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80042839114987e-05,
+      "loss": 0.8256,
+      "step": 3578
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798733746356294e-05,
+      "loss": 0.8616,
+      "step": 3579
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.797038864187564e-05,
+      "loss": 0.7702,
+      "step": 3580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.79534374486735e-05,
+      "loss": 0.7578,
+      "step": 3581
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.793648388619352e-05,
+      "loss": 0.8032,
+      "step": 3582
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.791952795667301e-05,
+      "loss": 0.7645,
+      "step": 3583
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.790256966234966e-05,
+      "loss": 0.7488,
+      "step": 3584
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.788560900546137e-05,
+      "loss": 0.7732,
+      "step": 3585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.786864598824644e-05,
+      "loss": 0.8593,
+      "step": 3586
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.78516806129434e-05,
+      "loss": 0.765,
+      "step": 3587
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.78347128817912e-05,
+      "loss": 0.8173,
+      "step": 3588
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.781774279702895e-05,
+      "loss": 0.7631,
+      "step": 3589
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.780077036089624e-05,
+      "loss": 0.8223,
+      "step": 3590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.778379557563286e-05,
+      "loss": 0.8285,
+      "step": 3591
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.776681844347892e-05,
+      "loss": 0.723,
+      "step": 3592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.774983896667486e-05,
+      "loss": 0.8257,
+      "step": 3593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.773285714746144e-05,
+      "loss": 0.8097,
+      "step": 3594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.77158729880797e-05,
+      "loss": 0.8121,
+      "step": 3595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.769888649077102e-05,
+      "loss": 0.8392,
+      "step": 3596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.768189765777705e-05,
+      "loss": 0.7218,
+      "step": 3597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.76649064913398e-05,
+      "loss": 0.8414,
+      "step": 3598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.764791299370155e-05,
+      "loss": 0.8513,
+      "step": 3599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.76309171671049e-05,
+      "loss": 0.8417,
+      "step": 3600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.761391901379274e-05,
+      "loss": 0.788,
+      "step": 3601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.759691853600828e-05,
+      "loss": 0.8329,
+      "step": 3602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.757991573599503e-05,
+      "loss": 0.7812,
+      "step": 3603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.756291061599686e-05,
+      "loss": 0.8756,
+      "step": 3604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.754590317825785e-05,
+      "loss": 0.806,
+      "step": 3605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.752889342502245e-05,
+      "loss": 0.6903,
+      "step": 3606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.751188135853541e-05,
+      "loss": 0.7712,
+      "step": 3607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.749486698104178e-05,
+      "loss": 0.7955,
+      "step": 3608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.747785029478691e-05,
+      "loss": 0.869,
+      "step": 3609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.746083130201646e-05,
+      "loss": 0.7106,
+      "step": 3610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.744381000497637e-05,
+      "loss": 0.7628,
+      "step": 3611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.742678640591292e-05,
+      "loss": 0.6701,
+      "step": 3612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.740976050707269e-05,
+      "loss": 0.8065,
+      "step": 3613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.739273231070254e-05,
+      "loss": 0.8235,
+      "step": 3614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.737570181904962e-05,
+      "loss": 0.8251,
+      "step": 3615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.735866903436145e-05,
+      "loss": 0.7815,
+      "step": 3616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.73416339588858e-05,
+      "loss": 0.851,
+      "step": 3617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.732459659487076e-05,
+      "loss": 0.7357,
+      "step": 3618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.73075569445647e-05,
+      "loss": 0.7432,
+      "step": 3619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.72905150102163e-05,
+      "loss": 0.8796,
+      "step": 3620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.727347079407457e-05,
+      "loss": 0.7706,
+      "step": 3621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.725642429838879e-05,
+      "loss": 0.8176,
+      "step": 3622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.723937552540855e-05,
+      "loss": 0.8486,
+      "step": 3623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.722232447738375e-05,
+      "loss": 0.8093,
+      "step": 3624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.720527115656456e-05,
+      "loss": 0.7616,
+      "step": 3625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.718821556520151e-05,
+      "loss": 0.8228,
+      "step": 3626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.717115770554534e-05,
+      "loss": 0.7667,
+      "step": 3627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.715409757984717e-05,
+      "loss": 0.7702,
+      "step": 3628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.713703519035842e-05,
+      "loss": 0.7588,
+      "step": 3629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.71199705393307e-05,
+      "loss": 0.7385,
+      "step": 3630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.710290362901606e-05,
+      "loss": 0.7634,
+      "step": 3631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.708583446166676e-05,
+      "loss": 0.794,
+      "step": 3632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.706876303953538e-05,
+      "loss": 0.7731,
+      "step": 3633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.705168936487483e-05,
+      "loss": 0.788,
+      "step": 3634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.703461343993826e-05,
+      "loss": 0.7588,
+      "step": 3635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.701753526697915e-05,
+      "loss": 0.7722,
+      "step": 3636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.700045484825126e-05,
+      "loss": 0.835,
+      "step": 3637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.698337218600868e-05,
+      "loss": 0.7485,
+      "step": 3638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696628728250576e-05,
+      "loss": 0.8512,
+      "step": 3639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.694920013999714e-05,
+      "loss": 0.7911,
+      "step": 3640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.693211076073782e-05,
+      "loss": 0.7543,
+      "step": 3641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.691501914698301e-05,
+      "loss": 0.7661,
+      "step": 3642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.689792530098829e-05,
+      "loss": 0.8194,
+      "step": 3643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.688082922500946e-05,
+      "loss": 0.8238,
+      "step": 3644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.686373092130268e-05,
+      "loss": 0.9187,
+      "step": 3645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.684663039212438e-05,
+      "loss": 0.7366,
+      "step": 3646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.682952763973124e-05,
+      "loss": 0.7906,
+      "step": 3647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.681242266638035e-05,
+      "loss": 0.8137,
+      "step": 3648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.679531547432895e-05,
+      "loss": 0.6777,
+      "step": 3649
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.677820606583467e-05,
+      "loss": 0.8308,
+      "step": 3650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.67610944431554e-05,
+      "loss": 0.7491,
+      "step": 3651
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.674398060854931e-05,
+      "loss": 0.9232,
+      "step": 3652
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.67268645642749e-05,
+      "loss": 0.7517,
+      "step": 3653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.670974631259094e-05,
+      "loss": 0.8744,
+      "step": 3654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669262585575647e-05,
+      "loss": 0.7435,
+      "step": 3655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.667550319603084e-05,
+      "loss": 0.6553,
+      "step": 3656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.665837833567373e-05,
+      "loss": 0.8405,
+      "step": 3657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.664125127694501e-05,
+      "loss": 0.8268,
+      "step": 3658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.662412202210498e-05,
+      "loss": 0.8121,
+      "step": 3659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.660699057341406e-05,
+      "loss": 0.7332,
+      "step": 3660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.658985693313312e-05,
+      "loss": 0.7316,
+      "step": 3661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.657272110352323e-05,
+      "loss": 0.9168,
+      "step": 3662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.655558308684579e-05,
+      "loss": 0.7246,
+      "step": 3663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.653844288536244e-05,
+      "loss": 0.8178,
+      "step": 3664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.652130050133516e-05,
+      "loss": 0.7315,
+      "step": 3665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.650415593702616e-05,
+      "loss": 0.7972,
+      "step": 3666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6487009194698e-05,
+      "loss": 0.8034,
+      "step": 3667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.64698602766135e-05,
+      "loss": 0.8535,
+      "step": 3668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.645270918503575e-05,
+      "loss": 0.8861,
+      "step": 3669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.643555592222817e-05,
+      "loss": 0.7531,
+      "step": 3670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.64184004904544e-05,
+      "loss": 0.7498,
+      "step": 3671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.640124289197845e-05,
+      "loss": 0.8011,
+      "step": 3672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.638408312906457e-05,
+      "loss": 0.7247,
+      "step": 3673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.636692120397727e-05,
+      "loss": 0.7635,
+      "step": 3674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.634975711898139e-05,
+      "loss": 0.7287,
+      "step": 3675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.633259087634201e-05,
+      "loss": 0.7757,
+      "step": 3676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.631542247832459e-05,
+      "loss": 0.8403,
+      "step": 3677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.629825192719473e-05,
+      "loss": 0.7937,
+      "step": 3678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.628107922521845e-05,
+      "loss": 0.8415,
+      "step": 3679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.626390437466196e-05,
+      "loss": 0.7819,
+      "step": 3680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.62467273777918e-05,
+      "loss": 0.7425,
+      "step": 3681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.622954823687482e-05,
+      "loss": 0.8745,
+      "step": 3682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.621236695417806e-05,
+      "loss": 0.7337,
+      "step": 3683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.619518353196892e-05,
+      "loss": 0.8528,
+      "step": 3684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.617799797251507e-05,
+      "loss": 0.8277,
+      "step": 3685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.616081027808443e-05,
+      "loss": 0.8032,
+      "step": 3686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.614362045094524e-05,
+      "loss": 0.8408,
+      "step": 3687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.612642849336599e-05,
+      "loss": 0.6748,
+      "step": 3688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.610923440761551e-05,
+      "loss": 0.86,
+      "step": 3689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.609203819596281e-05,
+      "loss": 0.8155,
+      "step": 3690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.607483986067728e-05,
+      "loss": 0.8429,
+      "step": 3691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.605763940402854e-05,
+      "loss": 0.8813,
+      "step": 3692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.604043682828649e-05,
+      "loss": 0.8782,
+      "step": 3693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.602323213572132e-05,
+      "loss": 0.7723,
+      "step": 3694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.600602532860349e-05,
+      "loss": 0.8064,
+      "step": 3695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.598881640920376e-05,
+      "loss": 0.8292,
+      "step": 3696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.597160537979314e-05,
+      "loss": 0.8241,
+      "step": 3697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.595439224264297e-05,
+      "loss": 0.8236,
+      "step": 3698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59371770000248e-05,
+      "loss": 0.8712,
+      "step": 3699
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59199596542105e-05,
+      "loss": 0.7726,
+      "step": 3700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.590274020747217e-05,
+      "loss": 0.7529,
+      "step": 3701
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58855186620823e-05,
+      "loss": 0.7035,
+      "step": 3702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.586829502031353e-05,
+      "loss": 0.7757,
+      "step": 3703
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.585106928443883e-05,
+      "loss": 0.723,
+      "step": 3704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.583384145673145e-05,
+      "loss": 0.7864,
+      "step": 3705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.581661153946491e-05,
+      "loss": 0.7744,
+      "step": 3706
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.579937953491302e-05,
+      "loss": 0.7452,
+      "step": 3707
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.578214544534984e-05,
+      "loss": 0.7975,
+      "step": 3708
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.576490927304972e-05,
+      "loss": 0.7663,
+      "step": 3709
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.574767102028729e-05,
+      "loss": 0.723,
+      "step": 3710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.573043068933741e-05,
+      "loss": 0.8216,
+      "step": 3711
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57131882824753e-05,
+      "loss": 0.7996,
+      "step": 3712
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.569594380197638e-05,
+      "loss": 0.8514,
+      "step": 3713
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.567869725011634e-05,
+      "loss": 0.7774,
+      "step": 3714
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.566144862917123e-05,
+      "loss": 0.8379,
+      "step": 3715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.564419794141728e-05,
+      "loss": 0.7861,
+      "step": 3716
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.562694518913105e-05,
+      "loss": 0.7876,
+      "step": 3717
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.560969037458933e-05,
+      "loss": 0.8028,
+      "step": 3718
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.559243350006919e-05,
+      "loss": 0.7222,
+      "step": 3719
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5575174567848e-05,
+      "loss": 0.7901,
+      "step": 3720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.55579135802034e-05,
+      "loss": 0.7628,
+      "step": 3721
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.554065053941326e-05,
+      "loss": 0.8993,
+      "step": 3722
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.552338544775576e-05,
+      "loss": 0.8082,
+      "step": 3723
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.550611830750934e-05,
+      "loss": 0.7821,
+      "step": 3724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.54888491209527e-05,
+      "loss": 0.8776,
+      "step": 3725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.547157789036484e-05,
+      "loss": 0.7965,
+      "step": 3726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545430461802499e-05,
+      "loss": 0.766,
+      "step": 3727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.543702930621267e-05,
+      "loss": 0.8377,
+      "step": 3728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.541975195720767e-05,
+      "loss": 0.8377,
+      "step": 3729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.540247257329004e-05,
+      "loss": 0.7469,
+      "step": 3730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.538519115674011e-05,
+      "loss": 0.915,
+      "step": 3731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.536790770983846e-05,
+      "loss": 0.7242,
+      "step": 3732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.535062223486597e-05,
+      "loss": 0.8113,
+      "step": 3733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.533333473410376e-05,
+      "loss": 0.7893,
+      "step": 3734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.531604520983323e-05,
+      "loss": 0.7941,
+      "step": 3735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.529875366433604e-05,
+      "loss": 0.8467,
+      "step": 3736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.528146009989411e-05,
+      "loss": 0.7982,
+      "step": 3737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.526416451878965e-05,
+      "loss": 0.7005,
+      "step": 3738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.524686692330512e-05,
+      "loss": 0.8094,
+      "step": 3739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.522956731572323e-05,
+      "loss": 0.8012,
+      "step": 3740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.521226569832699e-05,
+      "loss": 0.8058,
+      "step": 3741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.519496207339965e-05,
+      "loss": 0.8395,
+      "step": 3742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.517765644322475e-05,
+      "loss": 0.783,
+      "step": 3743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.516034881008606e-05,
+      "loss": 0.8953,
+      "step": 3744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.514303917626763e-05,
+      "loss": 0.8396,
+      "step": 3745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.51257275440538e-05,
+      "loss": 0.84,
+      "step": 3746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.510841391572913e-05,
+      "loss": 0.8305,
+      "step": 3747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.509109829357847e-05,
+      "loss": 0.7402,
+      "step": 3748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.507378067988691e-05,
+      "loss": 0.8026,
+      "step": 3749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.505646107693985e-05,
+      "loss": 0.8143,
+      "step": 3750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.50391394870229e-05,
+      "loss": 0.7869,
+      "step": 3751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.502181591242196e-05,
+      "loss": 0.8391,
+      "step": 3752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.50044903554232e-05,
+      "loss": 0.7267,
+      "step": 3753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.498716281831302e-05,
+      "loss": 0.7565,
+      "step": 3754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.496983330337809e-05,
+      "loss": 0.8286,
+      "step": 3755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.495250181290538e-05,
+      "loss": 0.8157,
+      "step": 3756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.493516834918208e-05,
+      "loss": 0.911,
+      "step": 3757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.491783291449562e-05,
+      "loss": 0.7763,
+      "step": 3758
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.490049551113376e-05,
+      "loss": 0.8109,
+      "step": 3759
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.488315614138447e-05,
+      "loss": 0.7635,
+      "step": 3760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4865814807536e-05,
+      "loss": 0.8112,
+      "step": 3761
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.484847151187682e-05,
+      "loss": 0.7822,
+      "step": 3762
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.483112625669573e-05,
+      "loss": 0.835,
+      "step": 3763
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.481377904428171e-05,
+      "loss": 0.8209,
+      "step": 3764
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.479642987692405e-05,
+      "loss": 0.8689,
+      "step": 3765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.477907875691228e-05,
+      "loss": 0.6912,
+      "step": 3766
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.476172568653621e-05,
+      "loss": 0.7529,
+      "step": 3767
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.474437066808585e-05,
+      "loss": 0.6231,
+      "step": 3768
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.472701370385155e-05,
+      "loss": 0.8531,
+      "step": 3769
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.470965479612384e-05,
+      "loss": 0.7139,
+      "step": 3770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.469229394719355e-05,
+      "loss": 0.7615,
+      "step": 3771
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.467493115935178e-05,
+      "loss": 0.7537,
+      "step": 3772
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.465756643488982e-05,
+      "loss": 0.7737,
+      "step": 3773
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.464019977609927e-05,
+      "loss": 0.8295,
+      "step": 3774
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4622831185272e-05,
+      "loss": 0.7435,
+      "step": 3775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.460546066470005e-05,
+      "loss": 0.7924,
+      "step": 3776
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.458808821667581e-05,
+      "loss": 0.8363,
+      "step": 3777
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.457071384349189e-05,
+      "loss": 0.7656,
+      "step": 3778
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.455333754744113e-05,
+      "loss": 0.7869,
+      "step": 3779
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.453595933081667e-05,
+      "loss": 0.8253,
+      "step": 3780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.451857919591185e-05,
+      "loss": 0.8344,
+      "step": 3781
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.45011971450203e-05,
+      "loss": 0.7792,
+      "step": 3782
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.448381318043591e-05,
+      "loss": 0.8535,
+      "step": 3783
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.446642730445277e-05,
+      "loss": 0.7451,
+      "step": 3784
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.444903951936528e-05,
+      "loss": 0.7605,
+      "step": 3785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.443164982746807e-05,
+      "loss": 0.9227,
+      "step": 3786
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.441425823105603e-05,
+      "loss": 0.9079,
+      "step": 3787
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.439686473242428e-05,
+      "loss": 0.8628,
+      "step": 3788
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.437946933386818e-05,
+      "loss": 0.7844,
+      "step": 3789
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.436207203768341e-05,
+      "loss": 0.8046,
+      "step": 3790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.434467284616585e-05,
+      "loss": 0.7852,
+      "step": 3791
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.43272717616116e-05,
+      "loss": 0.8598,
+      "step": 3792
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.430986878631707e-05,
+      "loss": 0.7811,
+      "step": 3793
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.429246392257889e-05,
+      "loss": 0.7599,
+      "step": 3794
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.427505717269395e-05,
+      "loss": 0.7752,
+      "step": 3795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.425764853895937e-05,
+      "loss": 0.7748,
+      "step": 3796
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.424023802367254e-05,
+      "loss": 0.7002,
+      "step": 3797
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.422282562913109e-05,
+      "loss": 0.771,
+      "step": 3798
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.420541135763287e-05,
+      "loss": 0.7738,
+      "step": 3799
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.418799521147606e-05,
+      "loss": 0.7979,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4170577192959e-05,
+      "loss": 0.761,
+      "step": 3801
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.41531573043803e-05,
+      "loss": 0.7662,
+      "step": 3802
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.413573554803884e-05,
+      "loss": 0.8462,
+      "step": 3803
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.411831192623374e-05,
+      "loss": 0.8447,
+      "step": 3804
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.410088644126434e-05,
+      "loss": 0.8692,
+      "step": 3805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.408345909543027e-05,
+      "loss": 0.794,
+      "step": 3806
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.406602989103136e-05,
+      "loss": 0.8851,
+      "step": 3807
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.40485988303677e-05,
+      "loss": 0.7549,
+      "step": 3808
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.403116591573967e-05,
+      "loss": 0.8064,
+      "step": 3809
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.401373114944781e-05,
+      "loss": 0.7697,
+      "step": 3810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.399629453379297e-05,
+      "loss": 0.8,
+      "step": 3811
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.397885607107623e-05,
+      "loss": 0.8581,
+      "step": 3812
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.396141576359889e-05,
+      "loss": 0.915,
+      "step": 3813
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.394397361366252e-05,
+      "loss": 0.8268,
+      "step": 3814
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.392652962356893e-05,
+      "loss": 0.803,
+      "step": 3815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.390908379562017e-05,
+      "loss": 0.8618,
+      "step": 3816
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.389163613211853e-05,
+      "loss": 0.8214,
+      "step": 3817
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.387418663536652e-05,
+      "loss": 0.755,
+      "step": 3818
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.385673530766692e-05,
+      "loss": 0.7523,
+      "step": 3819
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.383928215132278e-05,
+      "loss": 0.7526,
+      "step": 3820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.382182716863732e-05,
+      "loss": 0.8299,
+      "step": 3821
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.380437036191402e-05,
+      "loss": 0.8498,
+      "step": 3822
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.378691173345668e-05,
+      "loss": 0.8202,
+      "step": 3823
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.376945128556922e-05,
+      "loss": 0.8931,
+      "step": 3824
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.375198902055592e-05,
+      "loss": 0.7809,
+      "step": 3825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.373452494072119e-05,
+      "loss": 0.7224,
+      "step": 3826
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.371705904836974e-05,
+      "loss": 0.8149,
+      "step": 3827
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.369959134580648e-05,
+      "loss": 0.7393,
+      "step": 3828
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.368212183533665e-05,
+      "loss": 0.9499,
+      "step": 3829
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.366465051926562e-05,
+      "loss": 0.743,
+      "step": 3830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.364717739989905e-05,
+      "loss": 0.8398,
+      "step": 3831
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.362970247954285e-05,
+      "loss": 0.7341,
+      "step": 3832
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.361222576050312e-05,
+      "loss": 0.8345,
+      "step": 3833
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.359474724508624e-05,
+      "loss": 0.7506,
+      "step": 3834
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.357726693559883e-05,
+      "loss": 0.768,
+      "step": 3835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.355978483434772e-05,
+      "loss": 0.69,
+      "step": 3836
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.354230094363996e-05,
+      "loss": 0.766,
+      "step": 3837
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.352481526578287e-05,
+      "loss": 0.7827,
+      "step": 3838
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.350732780308403e-05,
+      "loss": 0.7016,
+      "step": 3839
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.348983855785121e-05,
+      "loss": 0.8428,
+      "step": 3840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.347234753239242e-05,
+      "loss": 0.7429,
+      "step": 3841
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.345485472901594e-05,
+      "loss": 0.8305,
+      "step": 3842
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.343736015003021e-05,
+      "loss": 0.7713,
+      "step": 3843
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3419863797744e-05,
+      "loss": 0.8523,
+      "step": 3844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.340236567446624e-05,
+      "loss": 0.7991,
+      "step": 3845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.338486578250613e-05,
+      "loss": 0.7326,
+      "step": 3846
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.33673641241731e-05,
+      "loss": 0.7371,
+      "step": 3847
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.33498607017768e-05,
+      "loss": 0.7926,
+      "step": 3848
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.333235551762712e-05,
+      "loss": 0.7756,
+      "step": 3849
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.331484857403417e-05,
+      "loss": 0.8067,
+      "step": 3850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.329733987330834e-05,
+      "loss": 0.8221,
+      "step": 3851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.32798294177602e-05,
+      "loss": 0.7939,
+      "step": 3852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.326231720970056e-05,
+      "loss": 0.7823,
+      "step": 3853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.324480325144047e-05,
+      "loss": 0.8487,
+      "step": 3854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.322728754529122e-05,
+      "loss": 0.7893,
+      "step": 3855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.320977009356431e-05,
+      "loss": 0.7778,
+      "step": 3856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.31922508985715e-05,
+      "loss": 0.7559,
+      "step": 3857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.317472996262473e-05,
+      "loss": 0.8584,
+      "step": 3858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.315720728803624e-05,
+      "loss": 0.8643,
+      "step": 3859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.313968287711844e-05,
+      "loss": 0.7211,
+      "step": 3860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.312215673218399e-05,
+      "loss": 0.8193,
+      "step": 3861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.310462885554579e-05,
+      "loss": 0.7411,
+      "step": 3862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.308709924951693e-05,
+      "loss": 0.8418,
+      "step": 3863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.30695679164108e-05,
+      "loss": 0.7853,
+      "step": 3864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.305203485854095e-05,
+      "loss": 0.7283,
+      "step": 3865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.303450007822116e-05,
+      "loss": 0.7529,
+      "step": 3866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.301696357776548e-05,
+      "loss": 0.7183,
+      "step": 3867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.299942535948816e-05,
+      "loss": 0.8648,
+      "step": 3868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.29818854257037e-05,
+      "loss": 0.7927,
+      "step": 3869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.29643437787268e-05,
+      "loss": 0.8362,
+      "step": 3870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.294680042087238e-05,
+      "loss": 0.8487,
+      "step": 3871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.29292553544556e-05,
+      "loss": 0.7225,
+      "step": 3872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.291170858179187e-05,
+      "loss": 0.8348,
+      "step": 3873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.289416010519678e-05,
+      "loss": 0.8476,
+      "step": 3874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.287660992698618e-05,
+      "loss": 0.7764,
+      "step": 3875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.285905804947612e-05,
+      "loss": 0.7981,
+      "step": 3876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.28415044749829e-05,
+      "loss": 0.92,
+      "step": 3877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2823949205823e-05,
+      "loss": 0.799,
+      "step": 3878
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.280639224431317e-05,
+      "loss": 0.804,
+      "step": 3879
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.278883359277037e-05,
+      "loss": 0.7727,
+      "step": 3880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.277127325351178e-05,
+      "loss": 0.6891,
+      "step": 3881
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.275371122885478e-05,
+      "loss": 0.8015,
+      "step": 3882
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.273614752111703e-05,
+      "loss": 0.7333,
+      "step": 3883
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.271858213261634e-05,
+      "loss": 0.7833,
+      "step": 3884
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.270101506567079e-05,
+      "loss": 0.748,
+      "step": 3885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26834463225987e-05,
+      "loss": 0.8891,
+      "step": 3886
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.266587590571852e-05,
+      "loss": 0.8233,
+      "step": 3887
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.264830381734903e-05,
+      "loss": 0.8413,
+      "step": 3888
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.263073005980917e-05,
+      "loss": 0.8706,
+      "step": 3889
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.261315463541811e-05,
+      "loss": 0.7778,
+      "step": 3890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.259557754649523e-05,
+      "loss": 0.8207,
+      "step": 3891
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.257799879536016e-05,
+      "loss": 0.9198,
+      "step": 3892
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.256041838433274e-05,
+      "loss": 0.8668,
+      "step": 3893
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.254283631573297e-05,
+      "loss": 0.7536,
+      "step": 3894
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.252525259188117e-05,
+      "loss": 0.7262,
+      "step": 3895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.25076672150978e-05,
+      "loss": 0.8785,
+      "step": 3896
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.24900801877036e-05,
+      "loss": 0.8775,
+      "step": 3897
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.247249151201946e-05,
+      "loss": 0.7889,
+      "step": 3898
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.245490119036653e-05,
+      "loss": 0.7801,
+      "step": 3899
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.243730922506618e-05,
+      "loss": 0.9385,
+      "step": 3900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.241971561843995e-05,
+      "loss": 0.8446,
+      "step": 3901
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.240212037280966e-05,
+      "loss": 0.8774,
+      "step": 3902
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.238452349049733e-05,
+      "loss": 0.787,
+      "step": 3903
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.236692497382517e-05,
+      "loss": 0.8172,
+      "step": 3904
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.234932482511561e-05,
+      "loss": 0.8403,
+      "step": 3905
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.233172304669131e-05,
+      "loss": 0.7563,
+      "step": 3906
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.231411964087514e-05,
+      "loss": 0.7038,
+      "step": 3907
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.22965146099902e-05,
+      "loss": 0.7675,
+      "step": 3908
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.227890795635977e-05,
+      "loss": 0.8228,
+      "step": 3909
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.226129968230738e-05,
+      "loss": 0.7926,
+      "step": 3910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.224368979015675e-05,
+      "loss": 0.8397,
+      "step": 3911
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.22260782822318e-05,
+      "loss": 0.7509,
+      "step": 3912
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.22084651608567e-05,
+      "loss": 0.8488,
+      "step": 3913
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.219085042835584e-05,
+      "loss": 0.8519,
+      "step": 3914
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.217323408705378e-05,
+      "loss": 0.7663,
+      "step": 3915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.215561613927529e-05,
+      "loss": 0.8672,
+      "step": 3916
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.213799658734542e-05,
+      "loss": 0.7779,
+      "step": 3917
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.212037543358935e-05,
+      "loss": 0.7997,
+      "step": 3918
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.210275268033253e-05,
+      "loss": 0.9041,
+      "step": 3919
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.208512832990057e-05,
+      "loss": 0.8257,
+      "step": 3920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.206750238461934e-05,
+      "loss": 0.8154,
+      "step": 3921
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.204987484681488e-05,
+      "loss": 0.7053,
+      "step": 3922
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.203224571881349e-05,
+      "loss": 0.8109,
+      "step": 3923
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.201461500294165e-05,
+      "loss": 0.7348,
+      "step": 3924
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.199698270152602e-05,
+      "loss": 0.8082,
+      "step": 3925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.197934881689351e-05,
+      "loss": 0.8292,
+      "step": 3926
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.196171335137122e-05,
+      "loss": 0.7841,
+      "step": 3927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.194407630728649e-05,
+      "loss": 0.8244,
+      "step": 3928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.19264376869668e-05,
+      "loss": 0.7188,
+      "step": 3929
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.190879749273993e-05,
+      "loss": 0.8167,
+      "step": 3930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.18911557269338e-05,
+      "loss": 0.8108,
+      "step": 3931
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.187351239187654e-05,
+      "loss": 0.7491,
+      "step": 3932
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.185586748989655e-05,
+      "loss": 0.9026,
+      "step": 3933
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.183822102332234e-05,
+      "loss": 0.8768,
+      "step": 3934
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.182057299448271e-05,
+      "loss": 0.7917,
+      "step": 3935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.180292340570662e-05,
+      "loss": 0.8552,
+      "step": 3936
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.178527225932325e-05,
+      "loss": 0.801,
+      "step": 3937
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.176761955766199e-05,
+      "loss": 0.854,
+      "step": 3938
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.174996530305242e-05,
+      "loss": 0.8117,
+      "step": 3939
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.173230949782435e-05,
+      "loss": 0.8063,
+      "step": 3940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.171465214430778e-05,
+      "loss": 0.8037,
+      "step": 3941
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.169699324483291e-05,
+      "loss": 0.8489,
+      "step": 3942
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.167933280173013e-05,
+      "loss": 0.8082,
+      "step": 3943
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.16616708173301e-05,
+      "loss": 0.9001,
+      "step": 3944
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.164400729396358e-05,
+      "loss": 0.7583,
+      "step": 3945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.162634223396161e-05,
+      "loss": 0.7793,
+      "step": 3946
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.160867563965543e-05,
+      "loss": 0.8436,
+      "step": 3947
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.159100751337642e-05,
+      "loss": 0.7813,
+      "step": 3948
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.157333785745625e-05,
+      "loss": 0.7132,
+      "step": 3949
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.155566667422672e-05,
+      "loss": 0.8303,
+      "step": 3950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.153799396601987e-05,
+      "loss": 0.8921,
+      "step": 3951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.152031973516794e-05,
+      "loss": 0.8144,
+      "step": 3952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.150264398400337e-05,
+      "loss": 0.8958,
+      "step": 3953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.148496671485875e-05,
+      "loss": 0.7338,
+      "step": 3954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.146728793006694e-05,
+      "loss": 0.7188,
+      "step": 3955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.144960763196097e-05,
+      "loss": 0.8383,
+      "step": 3956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.143192582287407e-05,
+      "loss": 0.7688,
+      "step": 3957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.141424250513966e-05,
+      "loss": 0.8121,
+      "step": 3958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.13965576810914e-05,
+      "loss": 0.7462,
+      "step": 3959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.137887135306312e-05,
+      "loss": 0.7969,
+      "step": 3960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.13611835233888e-05,
+      "loss": 0.8547,
+      "step": 3961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.134349419440271e-05,
+      "loss": 0.829,
+      "step": 3962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.132580336843928e-05,
+      "loss": 0.7415,
+      "step": 3963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.130811104783308e-05,
+      "loss": 0.8786,
+      "step": 3964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.129041723491898e-05,
+      "loss": 0.804,
+      "step": 3965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.127272193203197e-05,
+      "loss": 0.77,
+      "step": 3966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.125502514150727e-05,
+      "loss": 0.7271,
+      "step": 3967
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.123732686568028e-05,
+      "loss": 0.7914,
+      "step": 3968
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.121962710688664e-05,
+      "loss": 0.857,
+      "step": 3969
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.120192586746211e-05,
+      "loss": 0.7369,
+      "step": 3970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.118422314974269e-05,
+      "loss": 0.762,
+      "step": 3971
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.11665189560646e-05,
+      "loss": 0.7556,
+      "step": 3972
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.114881328876418e-05,
+      "loss": 0.8751,
+      "step": 3973
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.113110615017807e-05,
+      "loss": 0.7812,
+      "step": 3974
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.111339754264297e-05,
+      "loss": 0.7639,
+      "step": 3975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.10956874684959e-05,
+      "loss": 0.8087,
+      "step": 3976
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.107797593007403e-05,
+      "loss": 0.7233,
+      "step": 3977
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.106026292971468e-05,
+      "loss": 0.9012,
+      "step": 3978
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.104254846975541e-05,
+      "loss": 0.8088,
+      "step": 3979
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.102483255253397e-05,
+      "loss": 0.7931,
+      "step": 3980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1007115180388285e-05,
+      "loss": 0.8019,
+      "step": 3981
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.098939635565648e-05,
+      "loss": 0.6803,
+      "step": 3982
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.097167608067688e-05,
+      "loss": 0.8147,
+      "step": 3983
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.095395435778797e-05,
+      "loss": 0.6985,
+      "step": 3984
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.093623118932847e-05,
+      "loss": 0.798,
+      "step": 3985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0918506577637255e-05,
+      "loss": 0.7672,
+      "step": 3986
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0900780525053434e-05,
+      "loss": 0.7689,
+      "step": 3987
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.088305303391625e-05,
+      "loss": 0.8193,
+      "step": 3988
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.086532410656517e-05,
+      "loss": 0.7478,
+      "step": 3989
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.084759374533985e-05,
+      "loss": 0.7596,
+      "step": 3990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0829861952580115e-05,
+      "loss": 0.7997,
+      "step": 3991
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.081212873062599e-05,
+      "loss": 0.8647,
+      "step": 3992
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0794394081817717e-05,
+      "loss": 0.7639,
+      "step": 3993
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.077665800849568e-05,
+      "loss": 0.75,
+      "step": 3994
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.075892051300047e-05,
+      "loss": 0.8528,
+      "step": 3995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.07411815976729e-05,
+      "loss": 0.7426,
+      "step": 3996
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0723441264853895e-05,
+      "loss": 0.8656,
+      "step": 3997
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.070569951688464e-05,
+      "loss": 0.7825,
+      "step": 3998
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0687956356106466e-05,
+      "loss": 0.8166,
+      "step": 3999
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0670211784860875e-05,
+      "loss": 0.7163,
+      "step": 4000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0652465805489634e-05,
+      "loss": 0.8039,
+      "step": 4001
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0634718420334604e-05,
+      "loss": 0.7903,
+      "step": 4002
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0616969631737875e-05,
+      "loss": 0.7291,
+      "step": 4003
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0599219442041734e-05,
+      "loss": 0.7825,
+      "step": 4004
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0581467853588615e-05,
+      "loss": 0.731,
+      "step": 4005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0563714868721174e-05,
+      "loss": 0.8311,
+      "step": 4006
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0545960489782236e-05,
+      "loss": 0.8948,
+      "step": 4007
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.052820471911481e-05,
+      "loss": 0.8496,
+      "step": 4008
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.051044755906209e-05,
+      "loss": 0.8382,
+      "step": 4009
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.049268901196742e-05,
+      "loss": 0.7193,
+      "step": 4010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0474929080174405e-05,
+      "loss": 0.7837,
+      "step": 4011
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0457167766026746e-05,
+      "loss": 0.831,
+      "step": 4012
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.043940507186839e-05,
+      "loss": 0.8328,
+      "step": 4013
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.042164100004345e-05,
+      "loss": 0.666,
+      "step": 4014
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0403875552896194e-05,
+      "loss": 0.8216,
+      "step": 4015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.03861087327711e-05,
+      "loss": 0.7685,
+      "step": 4016
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0368340542012824e-05,
+      "loss": 0.7302,
+      "step": 4017
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.035057098296618e-05,
+      "loss": 0.7318,
+      "step": 4018
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.03328000579762e-05,
+      "loss": 0.8653,
+      "step": 4019
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0315027769388054e-05,
+      "loss": 0.7857,
+      "step": 4020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.029725411954712e-05,
+      "loss": 0.8937,
+      "step": 4021
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.027947911079898e-05,
+      "loss": 0.8147,
+      "step": 4022
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0261702745489335e-05,
+      "loss": 0.7848,
+      "step": 4023
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.02439250259641e-05,
+      "loss": 0.6328,
+      "step": 4024
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.022614595456935e-05,
+      "loss": 0.88,
+      "step": 4025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.020836553365138e-05,
+      "loss": 0.7296,
+      "step": 4026
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.019058376555662e-05,
+      "loss": 0.7019,
+      "step": 4027
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.01728006526317e-05,
+      "loss": 0.7231,
+      "step": 4028
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.01550161972234e-05,
+      "loss": 0.8075,
+      "step": 4029
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0137230401678724e-05,
+      "loss": 0.7195,
+      "step": 4030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.011944326834481e-05,
+      "loss": 0.7631,
+      "step": 4031
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0101654799569004e-05,
+      "loss": 0.8165,
+      "step": 4032
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.00838649976988e-05,
+      "loss": 0.785,
+      "step": 4033
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0066073865081895e-05,
+      "loss": 0.6349,
+      "step": 4034
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.004828140406612e-05,
+      "loss": 0.8102,
+      "step": 4035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.003048761699953e-05,
+      "loss": 0.7917,
+      "step": 4036
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.001269250623034e-05,
+      "loss": 0.7465,
+      "step": 4037
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.999489607410692e-05,
+      "loss": 0.8193,
+      "step": 4038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.997709832297783e-05,
+      "loss": 0.7502,
+      "step": 4039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.99592992551918e-05,
+      "loss": 0.7352,
+      "step": 4040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9941498873097753e-05,
+      "loss": 0.8154,
+      "step": 4041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9923697179044756e-05,
+      "loss": 0.8796,
+      "step": 4042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.990589417538207e-05,
+      "loss": 0.7887,
+      "step": 4043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.988808986445911e-05,
+      "loss": 0.7576,
+      "step": 4044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.987028424862545e-05,
+      "loss": 0.8018,
+      "step": 4045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.985247733023091e-05,
+      "loss": 0.7622,
+      "step": 4046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98346691116254e-05,
+      "loss": 0.8419,
+      "step": 4047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.981685959515905e-05,
+      "loss": 0.7827,
+      "step": 4048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9799048783182146e-05,
+      "loss": 0.7659,
+      "step": 4049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9781236678045105e-05,
+      "loss": 0.7565,
+      "step": 4050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9763423282098605e-05,
+      "loss": 0.7876,
+      "step": 4051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9745608597693425e-05,
+      "loss": 0.7533,
+      "step": 4052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9727792627180524e-05,
+      "loss": 0.7728,
+      "step": 4053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.970997537291103e-05,
+      "loss": 0.7693,
+      "step": 4054
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9692156837236276e-05,
+      "loss": 0.8241,
+      "step": 4055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.967433702250772e-05,
+      "loss": 0.7965,
+      "step": 4056
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.965651593107701e-05,
+      "loss": 0.8207,
+      "step": 4057
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9638693565295955e-05,
+      "loss": 0.788,
+      "step": 4058
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.962086992751653e-05,
+      "loss": 0.7806,
+      "step": 4059
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.960304502009091e-05,
+      "loss": 0.8818,
+      "step": 4060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.958521884537138e-05,
+      "loss": 0.8231,
+      "step": 4061
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9567391405710435e-05,
+      "loss": 0.8528,
+      "step": 4062
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.954956270346074e-05,
+      "loss": 0.8039,
+      "step": 4063
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9531732740975077e-05,
+      "loss": 0.8325,
+      "step": 4064
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.951390152060646e-05,
+      "loss": 0.7557,
+      "step": 4065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.949606904470802e-05,
+      "loss": 0.8122,
+      "step": 4066
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.947823531563309e-05,
+      "loss": 0.7728,
+      "step": 4067
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.946040033573515e-05,
+      "loss": 0.7854,
+      "step": 4068
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.94425641073678e-05,
+      "loss": 0.7295,
+      "step": 4069
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.942472663288492e-05,
+      "loss": 0.8586,
+      "step": 4070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.940688791464043e-05,
+      "loss": 0.7073,
+      "step": 4071
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9389047954988494e-05,
+      "loss": 0.8209,
+      "step": 4072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9371206756283424e-05,
+      "loss": 0.7567,
+      "step": 4073
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.935336432087964e-05,
+      "loss": 0.7805,
+      "step": 4074
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9335520651131814e-05,
+      "loss": 0.7281,
+      "step": 4075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.931767574939473e-05,
+      "loss": 0.8071,
+      "step": 4076
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9299829618023336e-05,
+      "loss": 0.7333,
+      "step": 4077
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.928198225937276e-05,
+      "loss": 0.7277,
+      "step": 4078
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.926413367579825e-05,
+      "loss": 0.7627,
+      "step": 4079
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.924628386965528e-05,
+      "loss": 0.8455,
+      "step": 4080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9228432843299454e-05,
+      "loss": 0.83,
+      "step": 4081
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9210580599086506e-05,
+      "loss": 0.8144,
+      "step": 4082
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9192727139372374e-05,
+      "loss": 0.8045,
+      "step": 4083
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9174872466513145e-05,
+      "loss": 0.833,
+      "step": 4084
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9157016582865055e-05,
+      "loss": 0.7879,
+      "step": 4085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.913915949078452e-05,
+      "loss": 0.7702,
+      "step": 4086
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.912130119262811e-05,
+      "loss": 0.833,
+      "step": 4087
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9103441690752515e-05,
+      "loss": 0.8094,
+      "step": 4088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.908558098751463e-05,
+      "loss": 0.7842,
+      "step": 4089
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.906771908527152e-05,
+      "loss": 0.8951,
+      "step": 4090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9049855986380366e-05,
+      "loss": 0.6856,
+      "step": 4091
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.90319916931985e-05,
+      "loss": 0.8208,
+      "step": 4092
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9014126208083475e-05,
+      "loss": 0.6708,
+      "step": 4093
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.899625953339293e-05,
+      "loss": 0.7415,
+      "step": 4094
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.897839167148471e-05,
+      "loss": 0.8276,
+      "step": 4095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.896052262471681e-05,
+      "loss": 0.7238,
+      "step": 4096
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.894265239544736e-05,
+      "loss": 0.7913,
+      "step": 4097
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8924780986034646e-05,
+      "loss": 0.9134,
+      "step": 4098
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8906908398837124e-05,
+      "loss": 0.7739,
+      "step": 4099
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.888903463621341e-05,
+      "loss": 0.8068,
+      "step": 4100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.887115970052226e-05,
+      "loss": 0.828,
+      "step": 4101
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.885328359412261e-05,
+      "loss": 0.7133,
+      "step": 4102
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.883540631937352e-05,
+      "loss": 0.7334,
+      "step": 4103
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8817527878634206e-05,
+      "loss": 0.8236,
+      "step": 4104
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.879964827426406e-05,
+      "loss": 0.8087,
+      "step": 4105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8781767508622634e-05,
+      "loss": 0.8534,
+      "step": 4106
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.876388558406958e-05,
+      "loss": 0.908,
+      "step": 4107
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.874600250296477e-05,
+      "loss": 0.8137,
+      "step": 4108
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8728118267668165e-05,
+      "loss": 0.7507,
+      "step": 4109
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.871023288053994e-05,
+      "loss": 0.8334,
+      "step": 4110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8692346343940364e-05,
+      "loss": 0.6751,
+      "step": 4111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8674458660229914e-05,
+      "loss": 0.8876,
+      "step": 4112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.865656983176917e-05,
+      "loss": 0.7758,
+      "step": 4113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.863867986091891e-05,
+      "loss": 0.7625,
+      "step": 4114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.862078875004001e-05,
+      "loss": 0.7997,
+      "step": 4115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.860289650149353e-05,
+      "loss": 0.7458,
+      "step": 4116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.858500311764068e-05,
+      "loss": 0.7463,
+      "step": 4117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.85671086008428e-05,
+      "loss": 0.6488,
+      "step": 4118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.854921295346141e-05,
+      "loss": 0.7763,
+      "step": 4119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.853131617785813e-05,
+      "loss": 0.8715,
+      "step": 4120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8513418276394796e-05,
+      "loss": 0.7936,
+      "step": 4121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.849551925143334e-05,
+      "loss": 0.7316,
+      "step": 4122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8477619105335854e-05,
+      "loss": 0.8048,
+      "step": 4123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.84597178404646e-05,
+      "loss": 0.8205,
+      "step": 4124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8441815459181956e-05,
+      "loss": 0.8707,
+      "step": 4125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.842391196385048e-05,
+      "loss": 0.7791,
+      "step": 4126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8406007356832824e-05,
+      "loss": 0.8287,
+      "step": 4127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8388101640491844e-05,
+      "loss": 0.7598,
+      "step": 4128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.837019481719053e-05,
+      "loss": 0.6884,
+      "step": 4129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8352286889292e-05,
+      "loss": 0.8367,
+      "step": 4130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.833437785915952e-05,
+      "loss": 0.8174,
+      "step": 4131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.831646772915651e-05,
+      "loss": 0.7974,
+      "step": 4132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8298556501646526e-05,
+      "loss": 0.8716,
+      "step": 4133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.828064417899331e-05,
+      "loss": 0.7319,
+      "step": 4134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.826273076356066e-05,
+      "loss": 0.7779,
+      "step": 4135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.824481625771262e-05,
+      "loss": 0.7572,
+      "step": 4136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8226900663813287e-05,
+      "loss": 0.762,
+      "step": 4137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.820898398422698e-05,
+      "loss": 0.8166,
+      "step": 4138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8191066221318115e-05,
+      "loss": 0.8166,
+      "step": 4139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.817314737745127e-05,
+      "loss": 0.7914,
+      "step": 4140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8155227454991136e-05,
+      "loss": 0.6932,
+      "step": 4141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8137306456302584e-05,
+      "loss": 0.7188,
+      "step": 4142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.811938438375061e-05,
+      "loss": 0.7836,
+      "step": 4143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810146123970034e-05,
+      "loss": 0.6678,
+      "step": 4144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808353702651708e-05,
+      "loss": 0.8013,
+      "step": 4145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.806561174656622e-05,
+      "loss": 0.8095,
+      "step": 4146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.804768540221334e-05,
+      "loss": 0.7031,
+      "step": 4147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802975799582413e-05,
+      "loss": 0.731,
+      "step": 4148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801182952976446e-05,
+      "loss": 0.8353,
+      "step": 4149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.799390000640029e-05,
+      "loss": 0.7373,
+      "step": 4150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797596942809774e-05,
+      "loss": 0.8057,
+      "step": 4151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7958037797223084e-05,
+      "loss": 0.6721,
+      "step": 4152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7940105116142716e-05,
+      "loss": 0.8428,
+      "step": 4153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7922171387223165e-05,
+      "loss": 0.7569,
+      "step": 4154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.790423661283112e-05,
+      "loss": 0.7593,
+      "step": 4155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7886300795333405e-05,
+      "loss": 0.8592,
+      "step": 4156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7868363937096946e-05,
+      "loss": 0.8589,
+      "step": 4157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.785042604048885e-05,
+      "loss": 0.8545,
+      "step": 4158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.783248710787635e-05,
+      "loss": 0.7221,
+      "step": 4159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.78145471416268e-05,
+      "loss": 0.7922,
+      "step": 4160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.77966061441077e-05,
+      "loss": 0.789,
+      "step": 4161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.777866411768669e-05,
+      "loss": 0.807,
+      "step": 4162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7760721064731527e-05,
+      "loss": 0.8481,
+      "step": 4163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.774277698761013e-05,
+      "loss": 0.8185,
+      "step": 4164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7724831888690554e-05,
+      "loss": 0.8157,
+      "step": 4165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.770688577034096e-05,
+      "loss": 0.7692,
+      "step": 4166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7688938634929656e-05,
+      "loss": 0.807,
+      "step": 4167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7670990484825115e-05,
+      "loss": 0.6817,
+      "step": 4168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7653041322395895e-05,
+      "loss": 0.8084,
+      "step": 4169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.763509115001071e-05,
+      "loss": 0.8808,
+      "step": 4170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7617139970038416e-05,
+      "loss": 0.7411,
+      "step": 4171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.759918778484796e-05,
+      "loss": 0.8719,
+      "step": 4172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.758123459680851e-05,
+      "loss": 0.7716,
+      "step": 4173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7563280408289264e-05,
+      "loss": 0.8451,
+      "step": 4174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.754532522165964e-05,
+      "loss": 0.8614,
+      "step": 4175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.752736903928911e-05,
+      "loss": 0.7551,
+      "step": 4176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.750941186354734e-05,
+      "loss": 0.74,
+      "step": 4177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.749145369680407e-05,
+      "loss": 0.8802,
+      "step": 4178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.747349454142925e-05,
+      "loss": 0.7747,
+      "step": 4179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.745553439979287e-05,
+      "loss": 0.7723,
+      "step": 4180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.743757327426511e-05,
+      "loss": 0.8026,
+      "step": 4181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.741961116721625e-05,
+      "loss": 0.8141,
+      "step": 4182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.740164808101671e-05,
+      "loss": 0.9123,
+      "step": 4183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.738368401803708e-05,
+      "loss": 0.7625,
+      "step": 4184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7365718980648e-05,
+      "loss": 0.8627,
+      "step": 4185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.734775297122029e-05,
+      "loss": 0.8312,
+      "step": 4186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.732978599212489e-05,
+      "loss": 0.8297,
+      "step": 4187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.731181804573286e-05,
+      "loss": 0.85,
+      "step": 4188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.72938491344154e-05,
+      "loss": 0.8076,
+      "step": 4189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.727587926054383e-05,
+      "loss": 0.8333,
+      "step": 4190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.725790842648957e-05,
+      "loss": 0.8064,
+      "step": 4191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.723993663462422e-05,
+      "loss": 0.7875,
+      "step": 4192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7221963887319495e-05,
+      "loss": 0.8446,
+      "step": 4193
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.72039901869472e-05,
+      "loss": 0.7815,
+      "step": 4194
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.718601553587929e-05,
+      "loss": 0.7965,
+      "step": 4195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.716803993648784e-05,
+      "loss": 0.6792,
+      "step": 4196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7150063391145035e-05,
+      "loss": 0.8198,
+      "step": 4197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.713208590222323e-05,
+      "loss": 0.82,
+      "step": 4198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.711410747209488e-05,
+      "loss": 0.8663,
+      "step": 4199
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.709612810313253e-05,
+      "loss": 0.7558,
+      "step": 4200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.707814779770891e-05,
+      "loss": 0.7413,
+      "step": 4201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7060166558196826e-05,
+      "loss": 0.8288,
+      "step": 4202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.704218438696924e-05,
+      "loss": 0.7645,
+      "step": 4203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.702420128639921e-05,
+      "loss": 0.8567,
+      "step": 4204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.700621725885994e-05,
+      "loss": 0.7814,
+      "step": 4205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.698823230672473e-05,
+      "loss": 0.7924,
+      "step": 4206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.697024643236701e-05,
+      "loss": 0.7959,
+      "step": 4207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.695225963816038e-05,
+      "loss": 0.8128,
+      "step": 4208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6934271926478467e-05,
+      "loss": 0.7487,
+      "step": 4209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.691628329969509e-05,
+      "loss": 0.8654,
+      "step": 4210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6898293760184204e-05,
+      "loss": 0.7382,
+      "step": 4211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6880303310319803e-05,
+      "loss": 0.7734,
+      "step": 4212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.686231195247608e-05,
+      "loss": 0.7534,
+      "step": 4213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6844319689027305e-05,
+      "loss": 0.7641,
+      "step": 4214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.68263265223479e-05,
+      "loss": 0.7978,
+      "step": 4215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.680833245481234e-05,
+      "loss": 0.7535,
+      "step": 4216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6790337488795296e-05,
+      "loss": 0.7393,
+      "step": 4217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.677234162667153e-05,
+      "loss": 0.7385,
+      "step": 4218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.675434487081591e-05,
+      "loss": 0.8546,
+      "step": 4219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.673634722360344e-05,
+      "loss": 0.8027,
+      "step": 4220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6718348687409216e-05,
+      "loss": 0.7889,
+      "step": 4221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.670034926460847e-05,
+      "loss": 0.8264,
+      "step": 4222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6682348957576557e-05,
+      "loss": 0.7458,
+      "step": 4223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.666434776868895e-05,
+      "loss": 0.846,
+      "step": 4224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.664634570032121e-05,
+      "loss": 0.7967,
+      "step": 4225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6628342754849026e-05,
+      "loss": 0.8082,
+      "step": 4226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.661033893464823e-05,
+      "loss": 0.8409,
+      "step": 4227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.659233424209474e-05,
+      "loss": 0.8374,
+      "step": 4228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6574328679564594e-05,
+      "loss": 0.7965,
+      "step": 4229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.655632224943397e-05,
+      "loss": 0.7934,
+      "step": 4230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.65383149540791e-05,
+      "loss": 0.8338,
+      "step": 4231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.65203067958764e-05,
+      "loss": 0.6873,
+      "step": 4232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.650229777720236e-05,
+      "loss": 0.741,
+      "step": 4233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.64842879004336e-05,
+      "loss": 0.782,
+      "step": 4234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6466277167946835e-05,
+      "loss": 0.7292,
+      "step": 4235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6448265582118884e-05,
+      "loss": 0.7474,
+      "step": 4236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.643025314532675e-05,
+      "loss": 0.6927,
+      "step": 4237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.641223985994746e-05,
+      "loss": 0.742,
+      "step": 4238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.639422572835821e-05,
+      "loss": 0.7892,
+      "step": 4239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6376210752936265e-05,
+      "loss": 0.7795,
+      "step": 4240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.635819493605904e-05,
+      "loss": 0.7375,
+      "step": 4241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.634017828010404e-05,
+      "loss": 0.815,
+      "step": 4242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6322160787448894e-05,
+      "loss": 0.7672,
+      "step": 4243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.630414246047132e-05,
+      "loss": 0.758,
+      "step": 4244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6286123301549156e-05,
+      "loss": 0.8038,
+      "step": 4245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.626810331306036e-05,
+      "loss": 0.7482,
+      "step": 4246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.625008249738301e-05,
+      "loss": 0.8264,
+      "step": 4247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6232060856895254e-05,
+      "loss": 0.8372,
+      "step": 4248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.621403839397538e-05,
+      "loss": 0.9321,
+      "step": 4249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.619601511100177e-05,
+      "loss": 0.724,
+      "step": 4250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.61779910103529e-05,
+      "loss": 0.8471,
+      "step": 4251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6159966094407424e-05,
+      "loss": 0.8362,
+      "step": 4252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.614194036554401e-05,
+      "loss": 0.7487,
+      "step": 4253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.612391382614147e-05,
+      "loss": 0.7649,
+      "step": 4254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6105886478578775e-05,
+      "loss": 0.8365,
+      "step": 4255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6087858325234896e-05,
+      "loss": 0.8166,
+      "step": 4256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6069829368489035e-05,
+      "loss": 0.763,
+      "step": 4257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.605179961072039e-05,
+      "loss": 0.8411,
+      "step": 4258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.603376905430834e-05,
+      "loss": 0.8454,
+      "step": 4259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.601573770163231e-05,
+      "loss": 0.8648,
+      "step": 4260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5997705555071866e-05,
+      "loss": 0.723,
+      "step": 4261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.597967261700671e-05,
+      "loss": 0.8116,
+      "step": 4262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.596163888981656e-05,
+      "loss": 0.7911,
+      "step": 4263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.594360437588132e-05,
+      "loss": 0.7243,
+      "step": 4264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.592556907758096e-05,
+      "loss": 0.8161,
+      "step": 4265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.590753299729556e-05,
+      "loss": 0.8256,
+      "step": 4266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.588949613740531e-05,
+      "loss": 0.8622,
+      "step": 4267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.58714585002905e-05,
+      "loss": 0.7535,
+      "step": 4268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.58534200883315e-05,
+      "loss": 0.8879,
+      "step": 4269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.583538090390882e-05,
+      "loss": 0.858,
+      "step": 4270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.581734094940303e-05,
+      "loss": 0.7159,
+      "step": 4271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.579930022719486e-05,
+      "loss": 0.7996,
+      "step": 4272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5781258739665076e-05,
+      "loss": 0.6972,
+      "step": 4273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.57632164891946e-05,
+      "loss": 0.7373,
+      "step": 4274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5745173478164426e-05,
+      "loss": 0.787,
+      "step": 4275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.572712970895563e-05,
+      "loss": 0.7706,
+      "step": 4276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.570908518394944e-05,
+      "loss": 0.7584,
+      "step": 4277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.569103990552713e-05,
+      "loss": 0.7879,
+      "step": 4278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.567299387607011e-05,
+      "loss": 0.7905,
+      "step": 4279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.565494709795987e-05,
+      "loss": 0.8772,
+      "step": 4280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.563689957357803e-05,
+      "loss": 0.7634,
+      "step": 4281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.561885130530625e-05,
+      "loss": 0.778,
+      "step": 4282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.560080229552634e-05,
+      "loss": 0.8485,
+      "step": 4283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.55827525466202e-05,
+      "loss": 0.7755,
+      "step": 4284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5564702060969796e-05,
+      "loss": 0.8277,
+      "step": 4285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5546650840957225e-05,
+      "loss": 0.7734,
+      "step": 4286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.552859888896468e-05,
+      "loss": 0.8099,
+      "step": 4287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.551054620737442e-05,
+      "loss": 0.813,
+      "step": 4288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.549249279856883e-05,
+      "loss": 0.798,
+      "step": 4289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.547443866493036e-05,
+      "loss": 0.6996,
+      "step": 4290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5456383808841594e-05,
+      "loss": 0.7344,
+      "step": 4291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.543832823268521e-05,
+      "loss": 0.7385,
+      "step": 4292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.542027193884395e-05,
+      "loss": 0.8675,
+      "step": 4293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.540221492970067e-05,
+      "loss": 0.7971,
+      "step": 4294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.538415720763829e-05,
+      "loss": 0.8444,
+      "step": 4295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.536609877503989e-05,
+      "loss": 0.7804,
+      "step": 4296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5348039634288574e-05,
+      "loss": 0.6807,
+      "step": 4297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.532997978776759e-05,
+      "loss": 0.8259,
+      "step": 4298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5311919237860235e-05,
+      "loss": 0.744,
+      "step": 4299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.529385798694995e-05,
+      "loss": 0.8484,
+      "step": 4300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.52757960374202e-05,
+      "loss": 0.7351,
+      "step": 4301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.525773339165463e-05,
+      "loss": 0.7636,
+      "step": 4302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.523967005203692e-05,
+      "loss": 0.8162,
+      "step": 4303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.522160602095082e-05,
+      "loss": 0.7749,
+      "step": 4304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.520354130078023e-05,
+      "loss": 0.7357,
+      "step": 4305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.518547589390911e-05,
+      "loss": 0.752,
+      "step": 4306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5167409802721504e-05,
+      "loss": 0.711,
+      "step": 4307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.514934302960156e-05,
+      "loss": 0.8244,
+      "step": 4308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.513127557693353e-05,
+      "loss": 0.8006,
+      "step": 4309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.511320744710171e-05,
+      "loss": 0.7456,
+      "step": 4310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.509513864249054e-05,
+      "loss": 0.8065,
+      "step": 4311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5077069165484505e-05,
+      "loss": 0.7869,
+      "step": 4312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5058999018468205e-05,
+      "loss": 0.7973,
+      "step": 4313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5040928203826315e-05,
+      "loss": 0.8114,
+      "step": 4314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5022856723943594e-05,
+      "loss": 0.7556,
+      "step": 4315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5004784581204927e-05,
+      "loss": 0.7654,
+      "step": 4316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4986711777995225e-05,
+      "loss": 0.7135,
+      "step": 4317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4968638316699525e-05,
+      "loss": 0.7975,
+      "step": 4318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.495056419970297e-05,
+      "loss": 0.8483,
+      "step": 4319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.493248942939072e-05,
+      "loss": 0.7912,
+      "step": 4320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.491441400814811e-05,
+      "loss": 0.7246,
+      "step": 4321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4896337938360486e-05,
+      "loss": 0.8867,
+      "step": 4322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.487826122241333e-05,
+      "loss": 0.8028,
+      "step": 4323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.486018386269216e-05,
+      "loss": 0.7512,
+      "step": 4324
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4842105861582624e-05,
+      "loss": 0.8754,
+      "step": 4325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4824027221470455e-05,
+      "loss": 0.8202,
+      "step": 4326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.480594794474142e-05,
+      "loss": 0.6978,
+      "step": 4327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.478786803378143e-05,
+      "loss": 0.6995,
+      "step": 4328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4769787490976444e-05,
+      "loss": 0.7605,
+      "step": 4329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.47517063187125e-05,
+      "loss": 0.7569,
+      "step": 4330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.473362451937577e-05,
+      "loss": 0.8019,
+      "step": 4331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.471554209535243e-05,
+      "loss": 0.831,
+      "step": 4332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.469745904902881e-05,
+      "loss": 0.7987,
+      "step": 4333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.467937538279129e-05,
+      "loss": 0.77,
+      "step": 4334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.466129109902629e-05,
+      "loss": 0.7096,
+      "step": 4335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.46432062001204e-05,
+      "loss": 0.8264,
+      "step": 4336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.462512068846025e-05,
+      "loss": 0.7868,
+      "step": 4337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.460703456643252e-05,
+      "loss": 0.7699,
+      "step": 4338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.458894783642402e-05,
+      "loss": 0.7372,
+      "step": 4339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4570860500821607e-05,
+      "loss": 0.9048,
+      "step": 4340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4552772562012236e-05,
+      "loss": 0.7111,
+      "step": 4341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.453468402238293e-05,
+      "loss": 0.8131,
+      "step": 4342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.451659488432079e-05,
+      "loss": 0.8227,
+      "step": 4343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4498505150213e-05,
+      "loss": 0.7583,
+      "step": 4344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.448041482244683e-05,
+      "loss": 0.8055,
+      "step": 4345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.446232390340964e-05,
+      "loss": 0.7965,
+      "step": 4346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4444232395488835e-05,
+      "loss": 0.6808,
+      "step": 4347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.442614030107192e-05,
+      "loss": 0.7891,
+      "step": 4348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4408047622546455e-05,
+      "loss": 0.7699,
+      "step": 4349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.438995436230011e-05,
+      "loss": 0.7723,
+      "step": 4350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.437186052272061e-05,
+      "loss": 0.6668,
+      "step": 4351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.435376610619577e-05,
+      "loss": 0.8347,
+      "step": 4352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.433567111511344e-05,
+      "loss": 0.7823,
+      "step": 4353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4317575551861625e-05,
+      "loss": 0.7721,
+      "step": 4354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4299479418828316e-05,
+      "loss": 0.7872,
+      "step": 4355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.428138271840165e-05,
+      "loss": 0.764,
+      "step": 4356
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4263285452969806e-05,
+      "loss": 0.6907,
+      "step": 4357
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4245187624921045e-05,
+      "loss": 0.8483,
+      "step": 4358
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.422708923664368e-05,
+      "loss": 0.7501,
+      "step": 4359
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.420899029052614e-05,
+      "loss": 0.8129,
+      "step": 4360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4190890788956895e-05,
+      "loss": 0.7526,
+      "step": 4361
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.41727907343245e-05,
+      "loss": 0.731,
+      "step": 4362
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.415469012901758e-05,
+      "loss": 0.7227,
+      "step": 4363
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4136588975424826e-05,
+      "loss": 0.7787,
+      "step": 4364
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4118487275935035e-05,
+      "loss": 0.8268,
+      "step": 4365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.410038503293704e-05,
+      "loss": 0.7916,
+      "step": 4366
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.408228224881975e-05,
+      "loss": 0.7177,
+      "step": 4367
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.406417892597215e-05,
+      "loss": 0.7817,
+      "step": 4368
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.40460750667833e-05,
+      "loss": 0.792,
+      "step": 4369
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.402797067364234e-05,
+      "loss": 0.7884,
+      "step": 4370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.400986574893846e-05,
+      "loss": 0.7821,
+      "step": 4371
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.399176029506091e-05,
+      "loss": 0.7887,
+      "step": 4372
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3973654314399056e-05,
+      "loss": 0.7377,
+      "step": 4373
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.39555478093423e-05,
+      "loss": 0.7199,
+      "step": 4374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3937440782280124e-05,
+      "loss": 0.7561,
+      "step": 4375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.391933323560208e-05,
+      "loss": 0.7539,
+      "step": 4376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.390122517169775e-05,
+      "loss": 0.8071,
+      "step": 4377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.388311659295686e-05,
+      "loss": 0.825,
+      "step": 4378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3865007501769124e-05,
+      "loss": 0.7465,
+      "step": 4379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3846897900524385e-05,
+      "loss": 0.8347,
+      "step": 4380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.382878779161252e-05,
+      "loss": 0.7579,
+      "step": 4381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.38106771774235e-05,
+      "loss": 0.7948,
+      "step": 4382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3792566060347314e-05,
+      "loss": 0.8232,
+      "step": 4383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.377445444277407e-05,
+      "loss": 0.7282,
+      "step": 4384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.375634232709392e-05,
+      "loss": 0.8435,
+      "step": 4385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3738229715697086e-05,
+      "loss": 0.797,
+      "step": 4386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3720116610973835e-05,
+      "loss": 0.7763,
+      "step": 4387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.370200301531453e-05,
+      "loss": 0.7695,
+      "step": 4388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.368388893110956e-05,
+      "loss": 0.8016,
+      "step": 4389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3665774360749434e-05,
+      "loss": 0.7654,
+      "step": 4390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3647659306624686e-05,
+      "loss": 0.7881,
+      "step": 4391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.362954377112592e-05,
+      "loss": 0.774,
+      "step": 4392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.361142775664379e-05,
+      "loss": 0.8558,
+      "step": 4393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3593311265569065e-05,
+      "loss": 0.8121,
+      "step": 4394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.357519430029251e-05,
+      "loss": 0.7597,
+      "step": 4395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.355707686320499e-05,
+      "loss": 0.8568,
+      "step": 4396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.353895895669744e-05,
+      "loss": 0.7934,
+      "step": 4397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3520840583160805e-05,
+      "loss": 0.8348,
+      "step": 4398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.350272174498616e-05,
+      "loss": 0.7445,
+      "step": 4399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.348460244456461e-05,
+      "loss": 0.7262,
+      "step": 4400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.346648268428732e-05,
+      "loss": 0.7775,
+      "step": 4401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.344836246654552e-05,
+      "loss": 0.824,
+      "step": 4402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.343024179373047e-05,
+      "loss": 0.7786,
+      "step": 4403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.341212066823355e-05,
+      "loss": 0.8653,
+      "step": 4404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.339399909244616e-05,
+      "loss": 0.7184,
+      "step": 4405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.337587706875976e-05,
+      "loss": 0.7538,
+      "step": 4406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.335775459956587e-05,
+      "loss": 0.7457,
+      "step": 4407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3339631687256084e-05,
+      "loss": 0.7315,
+      "step": 4408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.332150833422204e-05,
+      "loss": 0.9062,
+      "step": 4409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.330338454285545e-05,
+      "loss": 0.8173,
+      "step": 4410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.328526031554807e-05,
+      "loss": 0.7737,
+      "step": 4411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.32671356546917e-05,
+      "loss": 0.7301,
+      "step": 4412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.324901056267824e-05,
+      "loss": 0.7533,
+      "step": 4413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3230885041899594e-05,
+      "loss": 0.8135,
+      "step": 4414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.321275909474778e-05,
+      "loss": 0.7297,
+      "step": 4415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.31946327236148e-05,
+      "loss": 0.8031,
+      "step": 4416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3176505930892796e-05,
+      "loss": 0.7479,
+      "step": 4417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.315837871897389e-05,
+      "loss": 0.7177,
+      "step": 4418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.314025109025031e-05,
+      "loss": 0.7203,
+      "step": 4419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.312212304711434e-05,
+      "loss": 0.7621,
+      "step": 4420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.310399459195826e-05,
+      "loss": 0.8292,
+      "step": 4421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.308586572717447e-05,
+      "loss": 0.7668,
+      "step": 4422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.306773645515538e-05,
+      "loss": 0.7574,
+      "step": 4423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.304960677829349e-05,
+      "loss": 0.7478,
+      "step": 4424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.303147669898134e-05,
+      "loss": 0.7976,
+      "step": 4425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3013346219611495e-05,
+      "loss": 0.6855,
+      "step": 4426
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.299521534257662e-05,
+      "loss": 0.7896,
+      "step": 4427
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.297708407026939e-05,
+      "loss": 0.8713,
+      "step": 4428
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.295895240508257e-05,
+      "loss": 0.7339,
+      "step": 4429
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2940820349408946e-05,
+      "loss": 0.8779,
+      "step": 4430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.292268790564138e-05,
+      "loss": 0.6596,
+      "step": 4431
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.290455507617276e-05,
+      "loss": 0.7621,
+      "step": 4432
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.288642186339603e-05,
+      "loss": 0.7877,
+      "step": 4433
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.286828826970421e-05,
+      "loss": 0.8693,
+      "step": 4434
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.285015429749035e-05,
+      "loss": 0.7819,
+      "step": 4435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.283201994914755e-05,
+      "loss": 0.7716,
+      "step": 4436
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.281388522706896e-05,
+      "loss": 0.8523,
+      "step": 4437
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.279575013364778e-05,
+      "loss": 0.8048,
+      "step": 4438
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2777614671277276e-05,
+      "loss": 0.8141,
+      "step": 4439
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.275947884235073e-05,
+      "loss": 0.7297,
+      "step": 4440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.274134264926149e-05,
+      "loss": 0.8828,
+      "step": 4441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.272320609440297e-05,
+      "loss": 0.7946,
+      "step": 4442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2705069180168577e-05,
+      "loss": 0.8975,
+      "step": 4443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.268693190895183e-05,
+      "loss": 0.7821,
+      "step": 4444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.266879428314627e-05,
+      "loss": 0.7183,
+      "step": 4445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.265065630514548e-05,
+      "loss": 0.7282,
+      "step": 4446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2632517977343086e-05,
+      "loss": 0.7837,
+      "step": 4447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.261437930213274e-05,
+      "loss": 0.7954,
+      "step": 4448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.259624028190822e-05,
+      "loss": 0.6474,
+      "step": 4449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.257810091906326e-05,
+      "loss": 0.8939,
+      "step": 4450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.255996121599167e-05,
+      "loss": 0.6984,
+      "step": 4451
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2541821175087304e-05,
+      "loss": 0.8569,
+      "step": 4452
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.252368079874409e-05,
+      "loss": 0.8276,
+      "step": 4453
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.250554008935596e-05,
+      "loss": 0.6639,
+      "step": 4454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2487399049316934e-05,
+      "loss": 0.8425,
+      "step": 4455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2469257681021013e-05,
+      "loss": 0.7093,
+      "step": 4456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.245111598686226e-05,
+      "loss": 0.8722,
+      "step": 4457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2432973969234855e-05,
+      "loss": 0.8457,
+      "step": 4458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.241483163053292e-05,
+      "loss": 0.7554,
+      "step": 4459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.239668897315068e-05,
+      "loss": 0.8113,
+      "step": 4460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.237854599948237e-05,
+      "loss": 0.7773,
+      "step": 4461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.236040271192229e-05,
+      "loss": 0.789,
+      "step": 4462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2342259112864745e-05,
+      "loss": 0.8105,
+      "step": 4463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2324115204704164e-05,
+      "loss": 0.6685,
+      "step": 4464
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.230597098983492e-05,
+      "loss": 0.6414,
+      "step": 4465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.228782647065147e-05,
+      "loss": 0.8616,
+      "step": 4466
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2269681649548307e-05,
+      "loss": 0.7814,
+      "step": 4467
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2251536528919984e-05,
+      "loss": 0.7753,
+      "step": 4468
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.223339111116107e-05,
+      "loss": 0.7995,
+      "step": 4469
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.221524539866616e-05,
+      "loss": 0.7116,
+      "step": 4470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.219709939382992e-05,
+      "loss": 0.7706,
+      "step": 4471
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2178953099047026e-05,
+      "loss": 0.7245,
+      "step": 4472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2160806516712236e-05,
+      "loss": 0.7026,
+      "step": 4473
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.214265964922031e-05,
+      "loss": 0.7737,
+      "step": 4474
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2124512498966026e-05,
+      "loss": 0.7298,
+      "step": 4475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.210636506834424e-05,
+      "loss": 0.8113,
+      "step": 4476
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2088217359749835e-05,
+      "loss": 0.8284,
+      "step": 4477
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2070069375577726e-05,
+      "loss": 0.7571,
+      "step": 4478
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.205192111822287e-05,
+      "loss": 0.7712,
+      "step": 4479
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.203377259008023e-05,
+      "loss": 0.821,
+      "step": 4480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.201562379354486e-05,
+      "loss": 0.7878,
+      "step": 4481
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1997474731011795e-05,
+      "loss": 0.7734,
+      "step": 4482
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.197932540487616e-05,
+      "loss": 0.7388,
+      "step": 4483
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1961175817533044e-05,
+      "loss": 0.7927,
+      "step": 4484
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1943025971377644e-05,
+      "loss": 0.7016,
+      "step": 4485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.192487586880514e-05,
+      "loss": 0.6912,
+      "step": 4486
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.190672551221076e-05,
+      "loss": 0.6903,
+      "step": 4487
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.188857490398979e-05,
+      "loss": 0.8198,
+      "step": 4488
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1870424046537504e-05,
+      "loss": 0.7366,
+      "step": 4489
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1852272942249244e-05,
+      "loss": 0.8169,
+      "step": 4490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.183412159352038e-05,
+      "loss": 0.8355,
+      "step": 4491
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.181597000274628e-05,
+      "loss": 0.7655,
+      "step": 4492
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.179781817232242e-05,
+      "loss": 0.8064,
+      "step": 4493
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.177966610464422e-05,
+      "loss": 0.8196,
+      "step": 4494
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1761513802107185e-05,
+      "loss": 0.8459,
+      "step": 4495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.174336126710683e-05,
+      "loss": 0.7951,
+      "step": 4496
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1725208502038704e-05,
+      "loss": 0.8416,
+      "step": 4497
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1707055509298396e-05,
+      "loss": 0.7503,
+      "step": 4498
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.168890229128153e-05,
+      "loss": 0.7578,
+      "step": 4499
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.167074885038373e-05,
+      "loss": 0.8081,
+      "step": 4500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1652595189000685e-05,
+      "loss": 0.8455,
+      "step": 4501
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.163444130952808e-05,
+      "loss": 0.7285,
+      "step": 4502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.161628721436166e-05,
+      "loss": 0.8502,
+      "step": 4503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.159813290589717e-05,
+      "loss": 0.8006,
+      "step": 4504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.157997838653039e-05,
+      "loss": 0.7351,
+      "step": 4505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.156182365865715e-05,
+      "loss": 0.8348,
+      "step": 4506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.154366872467329e-05,
+      "loss": 0.8298,
+      "step": 4507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.152551358697467e-05,
+      "loss": 0.7401,
+      "step": 4508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1507358247957206e-05,
+      "loss": 0.7928,
+      "step": 4509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.14892027100168e-05,
+      "loss": 0.7579,
+      "step": 4510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.147104697554941e-05,
+      "loss": 0.8653,
+      "step": 4511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1452891046951e-05,
+      "loss": 0.7587,
+      "step": 4512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.143473492661758e-05,
+      "loss": 0.671,
+      "step": 4513
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.141657861694518e-05,
+      "loss": 0.7891,
+      "step": 4514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1398422120329825e-05,
+      "loss": 0.7414,
+      "step": 4515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.138026543916763e-05,
+      "loss": 0.7846,
+      "step": 4516
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1362108575854654e-05,
+      "loss": 0.6995,
+      "step": 4517
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.134395153278705e-05,
+      "loss": 0.9096,
+      "step": 4518
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1325794312360956e-05,
+      "loss": 0.7401,
+      "step": 4519
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.130763691697255e-05,
+      "loss": 0.7919,
+      "step": 4520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.128947934901799e-05,
+      "loss": 0.8346,
+      "step": 4521
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.127132161089354e-05,
+      "loss": 0.7526,
+      "step": 4522
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.125316370499542e-05,
+      "loss": 0.764,
+      "step": 4523
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.123500563371988e-05,
+      "loss": 0.8481,
+      "step": 4524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.121684739946321e-05,
+      "loss": 0.7286,
+      "step": 4525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1198689004621715e-05,
+      "loss": 0.7755,
+      "step": 4526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.118053045159174e-05,
+      "loss": 0.7398,
+      "step": 4527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.116237174276961e-05,
+      "loss": 0.8109,
+      "step": 4528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.114421288055169e-05,
+      "loss": 0.7942,
+      "step": 4529
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.112605386733438e-05,
+      "loss": 0.8255,
+      "step": 4530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.110789470551407e-05,
+      "loss": 0.7589,
+      "step": 4531
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.10897353974872e-05,
+      "loss": 0.7687,
+      "step": 4532
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.107157594565023e-05,
+      "loss": 0.8676,
+      "step": 4533
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1053416352399586e-05,
+      "loss": 0.7718,
+      "step": 4534
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.103525662013179e-05,
+      "loss": 0.6585,
+      "step": 4535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.101709675124332e-05,
+      "loss": 0.736,
+      "step": 4536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.099893674813071e-05,
+      "loss": 0.7416,
+      "step": 4537
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.098077661319051e-05,
+      "loss": 0.7633,
+      "step": 4538
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0962616348819236e-05,
+      "loss": 0.7866,
+      "step": 4539
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0944455957413494e-05,
+      "loss": 0.8353,
+      "step": 4540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.092629544136984e-05,
+      "loss": 0.745,
+      "step": 4541
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.090813480308493e-05,
+      "loss": 0.8769,
+      "step": 4542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0889974044955336e-05,
+      "loss": 0.7488,
+      "step": 4543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.087181316937772e-05,
+      "loss": 0.8913,
+      "step": 4544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0853652178748746e-05,
+      "loss": 0.7575,
+      "step": 4545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0835491075465045e-05,
+      "loss": 0.7419,
+      "step": 4546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0817329861923344e-05,
+      "loss": 0.75,
+      "step": 4547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.079916854052031e-05,
+      "loss": 0.8881,
+      "step": 4548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0781007113652665e-05,
+      "loss": 0.7994,
+      "step": 4549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.076284558371715e-05,
+      "loss": 0.8098,
+      "step": 4550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0744683953110464e-05,
+      "loss": 0.807,
+      "step": 4551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.072652222422938e-05,
+      "loss": 0.7945,
+      "step": 4552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.070836039947069e-05,
+      "loss": 0.8682,
+      "step": 4553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.069019848123114e-05,
+      "loss": 0.8749,
+      "step": 4554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.067203647190754e-05,
+      "loss": 0.8007,
+      "step": 4555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.065387437389666e-05,
+      "loss": 0.7719,
+      "step": 4556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0635712189595355e-05,
+      "loss": 0.7369,
+      "step": 4557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.061754992140043e-05,
+      "loss": 0.7019,
+      "step": 4558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0599387571708737e-05,
+      "loss": 0.8959,
+      "step": 4559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.058122514291709e-05,
+      "loss": 0.8047,
+      "step": 4560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.056306263742238e-05,
+      "loss": 0.801,
+      "step": 4561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.054490005762145e-05,
+      "loss": 0.8152,
+      "step": 4562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0526737405911205e-05,
+      "loss": 0.7793,
+      "step": 4563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.050857468468853e-05,
+      "loss": 0.7334,
+      "step": 4564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0490411896350286e-05,
+      "loss": 0.7592,
+      "step": 4565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.047224904329342e-05,
+      "loss": 0.7907,
+      "step": 4566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0454086127914827e-05,
+      "loss": 0.792,
+      "step": 4567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.043592315261143e-05,
+      "loss": 0.741,
+      "step": 4568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0417760119780156e-05,
+      "loss": 0.6984,
+      "step": 4569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.039959703181795e-05,
+      "loss": 0.8475,
+      "step": 4570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0381433891121746e-05,
+      "loss": 0.8909,
+      "step": 4571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.036327070008852e-05,
+      "loss": 0.7588,
+      "step": 4572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.03451074611152e-05,
+      "loss": 0.7614,
+      "step": 4573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.032694417659878e-05,
+      "loss": 0.7363,
+      "step": 4574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.03087808489362e-05,
+      "loss": 0.8191,
+      "step": 4575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.029061748052446e-05,
+      "loss": 0.8192,
+      "step": 4576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.027245407376054e-05,
+      "loss": 0.7051,
+      "step": 4577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.025429063104142e-05,
+      "loss": 0.7801,
+      "step": 4578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.023612715476409e-05,
+      "loss": 0.8848,
+      "step": 4579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.021796364732554e-05,
+      "loss": 0.7811,
+      "step": 4580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0199800111122794e-05,
+      "loss": 0.7829,
+      "step": 4581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.018163654855285e-05,
+      "loss": 0.7436,
+      "step": 4582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.01634729620127e-05,
+      "loss": 0.7849,
+      "step": 4583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.014530935389937e-05,
+      "loss": 0.801,
+      "step": 4584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0127145726609846e-05,
+      "loss": 0.777,
+      "step": 4585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.010898208254118e-05,
+      "loss": 0.788,
+      "step": 4586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.009081842409038e-05,
+      "loss": 0.7813,
+      "step": 4587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.007265475365445e-05,
+      "loss": 0.7356,
+      "step": 4588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.005449107363043e-05,
+      "loss": 0.7433,
+      "step": 4589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0036327386415314e-05,
+      "loss": 0.6869,
+      "step": 4590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0018163694406176e-05,
+      "loss": 0.8896,
+      "step": 4591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-05,
+      "loss": 0.8381,
+      "step": 4592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9981836305593836e-05,
+      "loss": 0.8741,
+      "step": 4593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.996367261358469e-05,
+      "loss": 0.7185,
+      "step": 4594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.99455089263696e-05,
+      "loss": 0.7981,
+      "step": 4595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.992734524634557e-05,
+      "loss": 0.7234,
+      "step": 4596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.990918157590963e-05,
+      "loss": 0.7789,
+      "step": 4597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9891017917458824e-05,
+      "loss": 0.8401,
+      "step": 4598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.987285427339015e-05,
+      "loss": 0.7128,
+      "step": 4599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.985469064610065e-05,
+      "loss": 0.8021,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.983652703798731e-05,
+      "loss": 0.7151,
+      "step": 4601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.981836345144716e-05,
+      "loss": 0.7898,
+      "step": 4602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.980019988887721e-05,
+      "loss": 0.8016,
+      "step": 4603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9782036352674464e-05,
+      "loss": 0.7645,
+      "step": 4604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.976387284523592e-05,
+      "loss": 0.7997,
+      "step": 4605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.97457093689586e-05,
+      "loss": 0.7161,
+      "step": 4606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.972754592623946e-05,
+      "loss": 0.7643,
+      "step": 4607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9709382519475535e-05,
+      "loss": 0.8814,
+      "step": 4608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.96912191510638e-05,
+      "loss": 0.7941,
+      "step": 4609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.967305582340123e-05,
+      "loss": 0.8193,
+      "step": 4610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.96548925388848e-05,
+      "loss": 0.8941,
+      "step": 4611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9636729299911495e-05,
+      "loss": 0.8591,
+      "step": 4612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9618566108878265e-05,
+      "loss": 0.8736,
+      "step": 4613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.960040296818206e-05,
+      "loss": 0.7528,
+      "step": 4614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.958223988021986e-05,
+      "loss": 0.7679,
+      "step": 4615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.956407684738859e-05,
+      "loss": 0.7855,
+      "step": 4616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.954591387208518e-05,
+      "loss": 0.715,
+      "step": 4617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9527750956706584e-05,
+      "loss": 0.762,
+      "step": 4618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.950958810364971e-05,
+      "loss": 0.8334,
+      "step": 4619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9491425315311484e-05,
+      "loss": 0.841,
+      "step": 4620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9473262594088806e-05,
+      "loss": 0.8325,
+      "step": 4621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.945509994237856e-05,
+      "loss": 0.8911,
+      "step": 4622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.943693736257763e-05,
+      "loss": 0.8038,
+      "step": 4623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.941877485708292e-05,
+      "loss": 0.732,
+      "step": 4624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9400612428291295e-05,
+      "loss": 0.699,
+      "step": 4625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.938245007859959e-05,
+      "loss": 0.8657,
+      "step": 4626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.936428781040465e-05,
+      "loss": 0.8134,
+      "step": 4627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.934612562610334e-05,
+      "loss": 0.7599,
+      "step": 4628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.932796352809248e-05,
+      "loss": 0.7849,
+      "step": 4629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9309801518768875e-05,
+      "loss": 0.8023,
+      "step": 4630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9291639600529325e-05,
+      "loss": 0.7449,
+      "step": 4631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9273477775770634e-05,
+      "loss": 0.7337,
+      "step": 4632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.925531604688955e-05,
+      "loss": 0.7586,
+      "step": 4633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.923715441628288e-05,
+      "loss": 0.7211,
+      "step": 4634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.921899288634735e-05,
+      "loss": 0.7356,
+      "step": 4635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.920083145947971e-05,
+      "loss": 0.8102,
+      "step": 4636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.918267013807666e-05,
+      "loss": 0.8262,
+      "step": 4637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.916450892453495e-05,
+      "loss": 0.7618,
+      "step": 4638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9146347821251266e-05,
+      "loss": 0.6856,
+      "step": 4639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9128186830622284e-05,
+      "loss": 0.7051,
+      "step": 4640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.911002595504468e-05,
+      "loss": 0.8266,
+      "step": 4641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.909186519691509e-05,
+      "loss": 0.7963,
+      "step": 4642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.907370455863017e-05,
+      "loss": 0.7656,
+      "step": 4643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.905554404258653e-05,
+      "loss": 0.7927,
+      "step": 4644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.903738365118078e-05,
+      "loss": 0.776,
+      "step": 4645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.901922338680952e-05,
+      "loss": 0.8085,
+      "step": 4646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9001063251869285e-05,
+      "loss": 0.7982,
+      "step": 4647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.898290324875668e-05,
+      "loss": 0.801,
+      "step": 4648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8964743379868215e-05,
+      "loss": 0.7792,
+      "step": 4649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.894658364760042e-05,
+      "loss": 0.7864,
+      "step": 4650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8928424054349784e-05,
+      "loss": 0.6611,
+      "step": 4651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8910264602512804e-05,
+      "loss": 0.7798,
+      "step": 4652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.889210529448594e-05,
+      "loss": 0.6881,
+      "step": 4653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.887394613266564e-05,
+      "loss": 0.7972,
+      "step": 4654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.885578711944833e-05,
+      "loss": 0.8064,
+      "step": 4655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.883762825723039e-05,
+      "loss": 0.8403,
+      "step": 4656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.881946954840826e-05,
+      "loss": 0.8,
+      "step": 4657
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.880131099537828e-05,
+      "loss": 0.7522,
+      "step": 4658
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.87831526005368e-05,
+      "loss": 0.8222,
+      "step": 4659
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.876499436628013e-05,
+      "loss": 0.7198,
+      "step": 4660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.874683629500459e-05,
+      "loss": 0.6549,
+      "step": 4661
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.872867838910647e-05,
+      "loss": 0.7206,
+      "step": 4662
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8710520650982024e-05,
+      "loss": 0.7669,
+      "step": 4663
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8692363083027484e-05,
+      "loss": 0.7392,
+      "step": 4664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.867420568763906e-05,
+      "loss": 0.7168,
+      "step": 4665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.865604846721295e-05,
+      "loss": 0.7419,
+      "step": 4666
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.863789142414535e-05,
+      "loss": 0.7986,
+      "step": 4667
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8619734560832384e-05,
+      "loss": 0.7448,
+      "step": 4668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8601577879670187e-05,
+      "loss": 0.812,
+      "step": 4669
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.858342138305483e-05,
+      "loss": 0.8535,
+      "step": 4670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8565265073382427e-05,
+      "loss": 0.8459,
+      "step": 4671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.854710895304902e-05,
+      "loss": 0.7965,
+      "step": 4672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.852895302445061e-05,
+      "loss": 0.7636,
+      "step": 4673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8510797289983215e-05,
+      "loss": 0.7225,
+      "step": 4674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.849264175204281e-05,
+      "loss": 0.8159,
+      "step": 4675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.847448641302532e-05,
+      "loss": 0.7845,
+      "step": 4676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8456331275326715e-05,
+      "loss": 0.7158,
+      "step": 4677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8438176341342854e-05,
+      "loss": 0.8255,
+      "step": 4678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8420021613469614e-05,
+      "loss": 0.8365,
+      "step": 4679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.840186709410285e-05,
+      "loss": 0.7991,
+      "step": 4680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8383712785638354e-05,
+      "loss": 0.8807,
+      "step": 4681
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.836555869047193e-05,
+      "loss": 0.8163,
+      "step": 4682
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8347404810999334e-05,
+      "loss": 0.7917,
+      "step": 4683
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.832925114961629e-05,
+      "loss": 0.8292,
+      "step": 4684
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.831109770871848e-05,
+      "loss": 0.8056,
+      "step": 4685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.829294449070161e-05,
+      "loss": 0.8571,
+      "step": 4686
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.827479149796131e-05,
+      "loss": 0.777,
+      "step": 4687
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8256638732893176e-05,
+      "loss": 0.8096,
+      "step": 4688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8238486197892827e-05,
+      "loss": 0.8608,
+      "step": 4689
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.822033389535579e-05,
+      "loss": 0.8354,
+      "step": 4690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.820218182767759e-05,
+      "loss": 0.7384,
+      "step": 4691
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.818402999725372e-05,
+      "loss": 0.8082,
+      "step": 4692
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8165878406479645e-05,
+      "loss": 0.701,
+      "step": 4693
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8147727057750774e-05,
+      "loss": 0.8053,
+      "step": 4694
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.812957595346251e-05,
+      "loss": 0.8433,
+      "step": 4695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.811142509601021e-05,
+      "loss": 0.7408,
+      "step": 4696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.809327448778923e-05,
+      "loss": 0.7925,
+      "step": 4697
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.807512413119486e-05,
+      "loss": 0.7786,
+      "step": 4698
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.805697402862236e-05,
+      "loss": 0.7119,
+      "step": 4699
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.803882418246696e-05,
+      "loss": 0.8255,
+      "step": 4700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.802067459512385e-05,
+      "loss": 0.7993,
+      "step": 4701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8002525268988216e-05,
+      "loss": 0.7815,
+      "step": 4702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.798437620645516e-05,
+      "loss": 0.9403,
+      "step": 4703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.796622740991978e-05,
+      "loss": 0.8053,
+      "step": 4704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.794807888177715e-05,
+      "loss": 0.7089,
+      "step": 4705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.792993062442228e-05,
+      "loss": 0.7185,
+      "step": 4706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.791178264025017e-05,
+      "loss": 0.7606,
+      "step": 4707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7893634931655765e-05,
+      "loss": 0.852,
+      "step": 4708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.787548750103399e-05,
+      "loss": 0.8329,
+      "step": 4709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.785734035077971e-05,
+      "loss": 0.8774,
+      "step": 4710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7839193483287776e-05,
+      "loss": 0.8126,
+      "step": 4711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.782104690095298e-05,
+      "loss": 0.8543,
+      "step": 4712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.78029006061701e-05,
+      "loss": 0.6442,
+      "step": 4713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.778475460133386e-05,
+      "loss": 0.8285,
+      "step": 4714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.776660888883894e-05,
+      "loss": 0.8061,
+      "step": 4715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7748463471080014e-05,
+      "loss": 0.7771,
+      "step": 4716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.773031835045169e-05,
+      "loss": 0.6947,
+      "step": 4717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7712173529348544e-05,
+      "loss": 0.7428,
+      "step": 4718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.76940290101651e-05,
+      "loss": 0.8377,
+      "step": 4719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7675884795295855e-05,
+      "loss": 0.7524,
+      "step": 4720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.765774088713526e-05,
+      "loss": 0.8291,
+      "step": 4721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.763959728807773e-05,
+      "loss": 0.7595,
+      "step": 4722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.762145400051765e-05,
+      "loss": 0.7828,
+      "step": 4723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.760331102684934e-05,
+      "loss": 0.7311,
+      "step": 4724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7585168369467076e-05,
+      "loss": 0.7551,
+      "step": 4725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.756702603076514e-05,
+      "loss": 0.794,
+      "step": 4726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7548884013137736e-05,
+      "loss": 0.7104,
+      "step": 4727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7530742318979005e-05,
+      "loss": 0.6905,
+      "step": 4728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7512600950683085e-05,
+      "loss": 0.8839,
+      "step": 4729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.749445991064404e-05,
+      "loss": 0.7822,
+      "step": 4730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7476319201255926e-05,
+      "loss": 0.7312,
+      "step": 4731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74581788249127e-05,
+      "loss": 0.8525,
+      "step": 4732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.744003878400835e-05,
+      "loss": 0.8023,
+      "step": 4733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7421899080936774e-05,
+      "loss": 0.7867,
+      "step": 4734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740375971809178e-05,
+      "loss": 0.8549,
+      "step": 4735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738562069786725e-05,
+      "loss": 0.7198,
+      "step": 4736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7367482022656926e-05,
+      "loss": 0.7939,
+      "step": 4737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.734934369485453e-05,
+      "loss": 0.7931,
+      "step": 4738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.733120571685374e-05,
+      "loss": 0.7804,
+      "step": 4739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.731306809104818e-05,
+      "loss": 0.7542,
+      "step": 4740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.729493081983143e-05,
+      "loss": 0.8205,
+      "step": 4741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.727679390559706e-05,
+      "loss": 0.8051,
+      "step": 4742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7258657350738526e-05,
+      "loss": 0.7509,
+      "step": 4743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.724052115764929e-05,
+      "loss": 0.8202,
+      "step": 4744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.722238532872273e-05,
+      "loss": 0.7082,
+      "step": 4745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7204249866352215e-05,
+      "loss": 0.6723,
+      "step": 4746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7186114772931046e-05,
+      "loss": 0.7738,
+      "step": 4747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.716798005085246e-05,
+      "loss": 0.734,
+      "step": 4748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.714984570250966e-05,
+      "loss": 0.8726,
+      "step": 4749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.71317117302958e-05,
+      "loss": 0.7077,
+      "step": 4750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.711357813660398e-05,
+      "loss": 0.7967,
+      "step": 4751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.709544492382726e-05,
+      "loss": 0.784,
+      "step": 4752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.707731209435865e-05,
+      "loss": 0.7931,
+      "step": 4753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.705917965059107e-05,
+      "loss": 0.7741,
+      "step": 4754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7041047594917434e-05,
+      "loss": 0.8328,
+      "step": 4755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.702291592973062e-05,
+      "loss": 0.8117,
+      "step": 4756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.700478465742339e-05,
+      "loss": 0.7666,
+      "step": 4757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.698665378038852e-05,
+      "loss": 0.784,
+      "step": 4758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6968523301018676e-05,
+      "loss": 0.8277,
+      "step": 4759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6950393221706515e-05,
+      "loss": 0.8568,
+      "step": 4760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6932263544844636e-05,
+      "loss": 0.8593,
+      "step": 4761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6914134272825557e-05,
+      "loss": 0.789,
+      "step": 4762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6896005408041764e-05,
+      "loss": 0.7497,
+      "step": 4763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.687787695288567e-05,
+      "loss": 0.7231,
+      "step": 4764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6859748909749685e-05,
+      "loss": 0.8193,
+      "step": 4765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.684162128102612e-05,
+      "loss": 0.6908,
+      "step": 4766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.682349406910722e-05,
+      "loss": 0.7711,
+      "step": 4767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.680536727638521e-05,
+      "loss": 0.7953,
+      "step": 4768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.678724090525224e-05,
+      "loss": 0.8167,
+      "step": 4769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.676911495810041e-05,
+      "loss": 0.7582,
+      "step": 4770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.675098943732178e-05,
+      "loss": 0.7364,
+      "step": 4771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.673286434530831e-05,
+      "loss": 0.8031,
+      "step": 4772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6714739684451954e-05,
+      "loss": 0.8158,
+      "step": 4773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6696615457144546e-05,
+      "loss": 0.7778,
+      "step": 4774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6678491665777965e-05,
+      "loss": 0.778,
+      "step": 4775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.666036831274392e-05,
+      "loss": 0.8764,
+      "step": 4776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.664224540043415e-05,
+      "loss": 0.7763,
+      "step": 4777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6624122931240246e-05,
+      "loss": 0.802,
+      "step": 4778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6606000907553845e-05,
+      "loss": 0.7755,
+      "step": 4779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.658787933176646e-05,
+      "loss": 0.7525,
+      "step": 4780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6569758206269546e-05,
+      "loss": 0.7446,
+      "step": 4781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6551637533454504e-05,
+      "loss": 0.7758,
+      "step": 4782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.65335173157127e-05,
+      "loss": 0.8025,
+      "step": 4783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.651539755543539e-05,
+      "loss": 0.8478,
+      "step": 4784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6497278255013844e-05,
+      "loss": 0.7217,
+      "step": 4785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.64791594168392e-05,
+      "loss": 0.811,
+      "step": 4786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.646104104330258e-05,
+      "loss": 0.7533,
+      "step": 4787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.644292313679501e-05,
+      "loss": 0.7205,
+      "step": 4788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.64248056997075e-05,
+      "loss": 0.8402,
+      "step": 4789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6406688734430947e-05,
+      "loss": 0.8451,
+      "step": 4790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.638857224335621e-05,
+      "loss": 0.7562,
+      "step": 4791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.63704562288741e-05,
+      "loss": 0.7624,
+      "step": 4792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6352340693375326e-05,
+      "loss": 0.7818,
+      "step": 4793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6334225639250564e-05,
+      "loss": 0.7594,
+      "step": 4794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.631611106889044e-05,
+      "loss": 0.7831,
+      "step": 4795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.629799698468548e-05,
+      "loss": 0.7627,
+      "step": 4796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.627988338902618e-05,
+      "loss": 0.7455,
+      "step": 4797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.626177028430293e-05,
+      "loss": 0.7376,
+      "step": 4798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.624365767290609e-05,
+      "loss": 0.7228,
+      "step": 4799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.622554555722594e-05,
+      "loss": 0.8205,
+      "step": 4800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6207433939652704e-05,
+      "loss": 0.8134,
+      "step": 4801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.618932282257653e-05,
+      "loss": 0.7147,
+      "step": 4802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6171212208387495e-05,
+      "loss": 0.7537,
+      "step": 4803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.615310209947562e-05,
+      "loss": 0.8267,
+      "step": 4804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.613499249823088e-05,
+      "loss": 0.7634,
+      "step": 4805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.611688340704315e-05,
+      "loss": 0.8396,
+      "step": 4806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.609877482830225e-05,
+      "loss": 0.7744,
+      "step": 4807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.608066676439794e-05,
+      "loss": 0.8105,
+      "step": 4808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.606255921771989e-05,
+      "loss": 0.7923,
+      "step": 4809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.604445219065771e-05,
+      "loss": 0.7457,
+      "step": 4810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6026345685600956e-05,
+      "loss": 0.6636,
+      "step": 4811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.60082397049391e-05,
+      "loss": 0.7317,
+      "step": 4812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.599013425106157e-05,
+      "loss": 0.7646,
+      "step": 4813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5972029326357674e-05,
+      "loss": 0.7196,
+      "step": 4814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.59539249332167e-05,
+      "loss": 0.7331,
+      "step": 4815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.593582107402785e-05,
+      "loss": 0.6955,
+      "step": 4816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5917717751180255e-05,
+      "loss": 0.8217,
+      "step": 4817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5899614967062966e-05,
+      "loss": 0.738,
+      "step": 4818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.588151272406497e-05,
+      "loss": 0.7969,
+      "step": 4819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.586341102457518e-05,
+      "loss": 0.8593,
+      "step": 4820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5845309870982435e-05,
+      "loss": 0.7475,
+      "step": 4821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.582720926567552e-05,
+      "loss": 0.8104,
+      "step": 4822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.580910921104312e-05,
+      "loss": 0.8286,
+      "step": 4823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.579100970947386e-05,
+      "loss": 0.8474,
+      "step": 4824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.577291076335632e-05,
+      "loss": 0.7472,
+      "step": 4825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.575481237507897e-05,
+      "loss": 0.8557,
+      "step": 4826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.57367145470302e-05,
+      "loss": 0.7049,
+      "step": 4827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5718617281598366e-05,
+      "loss": 0.8292,
+      "step": 4828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.57005205811717e-05,
+      "loss": 0.7519,
+      "step": 4829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5682424448138394e-05,
+      "loss": 0.7541,
+      "step": 4830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.566432888488657e-05,
+      "loss": 0.8157,
+      "step": 4831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.564623389380426e-05,
+      "loss": 0.6845,
+      "step": 4832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.562813947727939e-05,
+      "loss": 0.819,
+      "step": 4833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.561004563769989e-05,
+      "loss": 0.7498,
+      "step": 4834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.559195237745354e-05,
+      "loss": 0.827,
+      "step": 4835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5573859698928087e-05,
+      "loss": 0.7815,
+      "step": 4836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.555576760451118e-05,
+      "loss": 0.7605,
+      "step": 4837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.553767609659036e-05,
+      "loss": 0.8092,
+      "step": 4838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.551958517755318e-05,
+      "loss": 0.7875,
+      "step": 4839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.550149484978701e-05,
+      "loss": 0.7552,
+      "step": 4840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5483405115679235e-05,
+      "loss": 0.7566,
+      "step": 4841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.54653159776171e-05,
+      "loss": 0.669,
+      "step": 4842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.544722743798777e-05,
+      "loss": 0.8018,
+      "step": 4843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.542913949917839e-05,
+      "loss": 0.7416,
+      "step": 4844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5411052163575986e-05,
+      "loss": 0.747,
+      "step": 4845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.539296543356748e-05,
+      "loss": 0.6752,
+      "step": 4846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.537487931153976e-05,
+      "loss": 0.8315,
+      "step": 4847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.535679379987961e-05,
+      "loss": 0.8245,
+      "step": 4848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.533870890097372e-05,
+      "loss": 0.7736,
+      "step": 4849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.532062461720874e-05,
+      "loss": 0.7811,
+      "step": 4850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5302540950971206e-05,
+      "loss": 0.8301,
+      "step": 4851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.528445790464758e-05,
+      "loss": 0.8466,
+      "step": 4852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.526637548062424e-05,
+      "loss": 0.7919,
+      "step": 4853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.52482936812875e-05,
+      "loss": 0.8292,
+      "step": 4854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.523021250902357e-05,
+      "loss": 0.7983,
+      "step": 4855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.521213196621858e-05,
+      "loss": 0.8441,
+      "step": 4856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.51940520552586e-05,
+      "loss": 0.7513,
+      "step": 4857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.517597277852956e-05,
+      "loss": 0.8241,
+      "step": 4858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.515789413841738e-05,
+      "loss": 0.663,
+      "step": 4859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.513981613730786e-05,
+      "loss": 0.803,
+      "step": 4860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.51217387775867e-05,
+      "loss": 0.7967,
+      "step": 4861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.510366206163953e-05,
+      "loss": 0.7748,
+      "step": 4862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.508558599185189e-05,
+      "loss": 0.8078,
+      "step": 4863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.506751057060928e-05,
+      "loss": 0.8274,
+      "step": 4864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.504943580029705e-05,
+      "loss": 0.7233,
+      "step": 4865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.503136168330048e-05,
+      "loss": 0.7442,
+      "step": 4866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.501328822200478e-05,
+      "loss": 0.7911,
+      "step": 4867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4995215418795085e-05,
+      "loss": 0.6814,
+      "step": 4868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.497714327605641e-05,
+      "loss": 0.7473,
+      "step": 4869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.49590717961737e-05,
+      "loss": 0.7237,
+      "step": 4870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4941000981531814e-05,
+      "loss": 0.8139,
+      "step": 4871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.492293083451551e-05,
+      "loss": 0.8197,
+      "step": 4872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4904861357509464e-05,
+      "loss": 0.7873,
+      "step": 4873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4886792552898286e-05,
+      "loss": 0.812,
+      "step": 4874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.486872442306648e-05,
+      "loss": 0.7858,
+      "step": 4875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4850656970398445e-05,
+      "loss": 0.8418,
+      "step": 4876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.48325901972785e-05,
+      "loss": 0.7827,
+      "step": 4877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4814524106090905e-05,
+      "loss": 0.7735,
+      "step": 4878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4796458699219785e-05,
+      "loss": 0.756,
+      "step": 4879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.47783939790492e-05,
+      "loss": 0.8553,
+      "step": 4880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.476032994796311e-05,
+      "loss": 0.7979,
+      "step": 4881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4742266608345365e-05,
+      "loss": 0.759,
+      "step": 4882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.47242039625798e-05,
+      "loss": 0.776,
+      "step": 4883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4706142013050064e-05,
+      "loss": 0.7971,
+      "step": 4884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.468808076213978e-05,
+      "loss": 0.8032,
+      "step": 4885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4670020212232425e-05,
+      "loss": 0.7061,
+      "step": 4886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.465196036571143e-05,
+      "loss": 0.6235,
+      "step": 4887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.463390122496012e-05,
+      "loss": 0.7724,
+      "step": 4888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4615842792361716e-05,
+      "loss": 0.7489,
+      "step": 4889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459778507029935e-05,
+      "loss": 0.809,
+      "step": 4890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.457972806115607e-05,
+      "loss": 0.8127,
+      "step": 4891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4561671767314785e-05,
+      "loss": 0.7454,
+      "step": 4892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4543616191158404e-05,
+      "loss": 0.6927,
+      "step": 4893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.452556133506965e-05,
+      "loss": 0.6835,
+      "step": 4894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4507507201431183e-05,
+      "loss": 0.7239,
+      "step": 4895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4489453792625595e-05,
+      "loss": 0.8281,
+      "step": 4896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.447140111103533e-05,
+      "loss": 0.6872,
+      "step": 4897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4453349159042786e-05,
+      "loss": 0.7881,
+      "step": 4898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4435297939030215e-05,
+      "loss": 0.7349,
+      "step": 4899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.441724745337982e-05,
+      "loss": 0.7707,
+      "step": 4900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4399197704473675e-05,
+      "loss": 0.7352,
+      "step": 4901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.438114869469375e-05,
+      "loss": 0.7386,
+      "step": 4902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.436310042642198e-05,
+      "loss": 0.7896,
+      "step": 4903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4345052902040126e-05,
+      "loss": 0.8526,
+      "step": 4904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.432700612392989e-05,
+      "loss": 0.7942,
+      "step": 4905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.430896009447288e-05,
+      "loss": 0.7302,
+      "step": 4906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4290914816050574e-05,
+      "loss": 0.8566,
+      "step": 4907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.427287029104438e-05,
+      "loss": 0.7,
+      "step": 4908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.42548265218356e-05,
+      "loss": 0.7181,
+      "step": 4909
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.423678351080542e-05,
+      "loss": 0.8002,
+      "step": 4910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.421874126033493e-05,
+      "loss": 0.8448,
+      "step": 4911
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.420069977280515e-05,
+      "loss": 0.794,
+      "step": 4912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.418265905059697e-05,
+      "loss": 0.7704,
+      "step": 4913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.416461909609119e-05,
+      "loss": 0.8584,
+      "step": 4914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.414657991166851e-05,
+      "loss": 0.8063,
+      "step": 4915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.412854149970952e-05,
+      "loss": 0.8318,
+      "step": 4916
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4110503862594696e-05,
+      "loss": 0.7651,
+      "step": 4917
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.409246700270445e-05,
+      "loss": 0.6556,
+      "step": 4918
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.407443092241906e-05,
+      "loss": 0.8275,
+      "step": 4919
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.405639562411869e-05,
+      "loss": 0.9069,
+      "step": 4920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.403836111018346e-05,
+      "loss": 0.7905,
+      "step": 4921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.402032738299331e-05,
+      "loss": 0.8442,
+      "step": 4922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4002294444928126e-05,
+      "loss": 0.7444,
+      "step": 4923
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.39842622983677e-05,
+      "loss": 0.7911,
+      "step": 4924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.396623094569168e-05,
+      "loss": 0.7611,
+      "step": 4925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.394820038927962e-05,
+      "loss": 0.7601,
+      "step": 4926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3930170631510983e-05,
+      "loss": 0.8223,
+      "step": 4927
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.391214167476511e-05,
+      "loss": 0.7579,
+      "step": 4928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3894113521421244e-05,
+      "loss": 0.8027,
+      "step": 4929
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.387608617385854e-05,
+      "loss": 0.7362,
+      "step": 4930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3858059634456016e-05,
+      "loss": 0.7724,
+      "step": 4931
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.384003390559258e-05,
+      "loss": 0.6844,
+      "step": 4932
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3822008989647095e-05,
+      "loss": 0.7858,
+      "step": 4933
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.380398488899824e-05,
+      "loss": 0.7539,
+      "step": 4934
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3785961606024634e-05,
+      "loss": 0.7994,
+      "step": 4935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.376793914310475e-05,
+      "loss": 0.7949,
+      "step": 4936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3749917502617e-05,
+      "loss": 0.7985,
+      "step": 4937
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.373189668693965e-05,
+      "loss": 0.8094,
+      "step": 4938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.371387669845085e-05,
+      "loss": 0.8828,
+      "step": 4939
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.36958575395287e-05,
+      "loss": 0.6769,
+      "step": 4940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.367783921255111e-05,
+      "loss": 0.732,
+      "step": 4941
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3659821719895954e-05,
+      "loss": 0.7912,
+      "step": 4942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.364180506394096e-05,
+      "loss": 0.6953,
+      "step": 4943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.362378924706374e-05,
+      "loss": 0.7532,
+      "step": 4944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.36057742716418e-05,
+      "loss": 0.8007,
+      "step": 4945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.358776014005255e-05,
+      "loss": 0.6872,
+      "step": 4946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.356974685467327e-05,
+      "loss": 0.8632,
+      "step": 4947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.355173441788112e-05,
+      "loss": 0.7606,
+      "step": 4948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.353372283205319e-05,
+      "loss": 0.7693,
+      "step": 4949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.351571209956643e-05,
+      "loss": 0.8517,
+      "step": 4950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3497702222797645e-05,
+      "loss": 0.7946,
+      "step": 4951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.347969320412361e-05,
+      "loss": 0.7537,
+      "step": 4952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3461685045920905e-05,
+      "loss": 0.7498,
+      "step": 4953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.344367775056605e-05,
+      "loss": 0.7942,
+      "step": 4954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.342567132043541e-05,
+      "loss": 0.7672,
+      "step": 4955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.340766575790527e-05,
+      "loss": 0.7122,
+      "step": 4956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.338966106535178e-05,
+      "loss": 0.7444,
+      "step": 4957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3371657245150985e-05,
+      "loss": 0.7277,
+      "step": 4958
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.335365429967881e-05,
+      "loss": 0.8567,
+      "step": 4959
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.333565223131107e-05,
+      "loss": 0.7402,
+      "step": 4960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3317651042423435e-05,
+      "loss": 0.8266,
+      "step": 4961
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.329965073539153e-05,
+      "loss": 0.7655,
+      "step": 4962
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3281651312590796e-05,
+      "loss": 0.6469,
+      "step": 4963
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.326365277639657e-05,
+      "loss": 0.804,
+      "step": 4964
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.324565512918411e-05,
+      "loss": 0.7133,
+      "step": 4965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3227658373328475e-05,
+      "loss": 0.6563,
+      "step": 4966
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3209662511204716e-05,
+      "loss": 0.7584,
+      "step": 4967
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.319166754518768e-05,
+      "loss": 0.6945,
+      "step": 4968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3173673477652134e-05,
+      "loss": 0.7169,
+      "step": 4969
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.315568031097271e-05,
+      "loss": 0.727,
+      "step": 4970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3137688047523914e-05,
+      "loss": 0.8126,
+      "step": 4971
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.31196966896802e-05,
+      "loss": 0.8044,
+      "step": 4972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.310170623981581e-05,
+      "loss": 0.7641,
+      "step": 4973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.308371670030491e-05,
+      "loss": 0.8486,
+      "step": 4974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.306572807352154e-05,
+      "loss": 0.7921,
+      "step": 4975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.304774036183964e-05,
+      "loss": 0.7823,
+      "step": 4976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3029753567632994e-05,
+      "loss": 0.7844,
+      "step": 4977
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.301176769327528e-05,
+      "loss": 0.7196,
+      "step": 4978
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.299378274114008e-05,
+      "loss": 0.8114,
+      "step": 4979
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.29757987136008e-05,
+      "loss": 0.7783,
+      "step": 4980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2957815613030756e-05,
+      "loss": 0.7538,
+      "step": 4981
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2939833441803165e-05,
+      "loss": 0.7378,
+      "step": 4982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2921852202291094e-05,
+      "loss": 0.7431,
+      "step": 4983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2903871896867475e-05,
+      "loss": 0.6794,
+      "step": 4984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.288589252790513e-05,
+      "loss": 0.7533,
+      "step": 4985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2867914097776776e-05,
+      "loss": 0.7141,
+      "step": 4986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2849936608854977e-05,
+      "loss": 0.6401,
+      "step": 4987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.283196006351219e-05,
+      "loss": 0.6877,
+      "step": 4988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2813984464120736e-05,
+      "loss": 0.7778,
+      "step": 4989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2796009813052824e-05,
+      "loss": 0.8521,
+      "step": 4990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.27780361126805e-05,
+      "loss": 0.8204,
+      "step": 4991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.276006336537577e-05,
+      "loss": 0.7763,
+      "step": 4992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.274209157351044e-05,
+      "loss": 0.7395,
+      "step": 4993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2724120739456186e-05,
+      "loss": 0.833,
+      "step": 4994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.270615086558461e-05,
+      "loss": 0.7764,
+      "step": 4995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2688181954267146e-05,
+      "loss": 0.8125,
+      "step": 4996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2670214007875125e-05,
+      "loss": 0.7503,
+      "step": 4997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2652247028779726e-05,
+      "loss": 0.753,
+      "step": 4998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.263428101935203e-05,
+      "loss": 0.6926,
+      "step": 4999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2616315981962926e-05,
+      "loss": 0.7643,
+      "step": 5000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.259835191898329e-05,
+      "loss": 0.8147,
+      "step": 5001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.258038883278377e-05,
+      "loss": 0.7517,
+      "step": 5002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2562426725734904e-05,
+      "loss": 0.7444,
+      "step": 5003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2544465600207136e-05,
+      "loss": 0.7312,
+      "step": 5004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.252650545857076e-05,
+      "loss": 0.7187,
+      "step": 5005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.250854630319593e-05,
+      "loss": 0.7936,
+      "step": 5006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.249058813645268e-05,
+      "loss": 0.7218,
+      "step": 5007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.24726309607109e-05,
+      "loss": 0.7037,
+      "step": 5008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.245467477834038e-05,
+      "loss": 0.7189,
+      "step": 5009
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.243671959171073e-05,
+      "loss": 0.7113,
+      "step": 5010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.24187654031915e-05,
+      "loss": 0.8145,
+      "step": 5011
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.240081221515203e-05,
+      "loss": 0.7245,
+      "step": 5012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.23828600299616e-05,
+      "loss": 0.815,
+      "step": 5013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.23649088499893e-05,
+      "loss": 0.7209,
+      "step": 5014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.234695867760412e-05,
+      "loss": 0.8125,
+      "step": 5015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.23290095151749e-05,
+      "loss": 0.7605,
+      "step": 5016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2311061365070355e-05,
+      "loss": 0.7192,
+      "step": 5017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.229311422965906e-05,
+      "loss": 0.7205,
+      "step": 5018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.227516811130946e-05,
+      "loss": 0.7381,
+      "step": 5019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.225722301238987e-05,
+      "loss": 0.7846,
+      "step": 5020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.223927893526849e-05,
+      "loss": 0.741,
+      "step": 5021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2221335882313324e-05,
+      "loss": 0.7342,
+      "step": 5022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2203393855892306e-05,
+      "loss": 0.7489,
+      "step": 5023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.218545285837321e-05,
+      "loss": 0.6913,
+      "step": 5024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.216751289212366e-05,
+      "loss": 0.7682,
+      "step": 5025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.214957395951116e-05,
+      "loss": 0.7757,
+      "step": 5026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2131636062903066e-05,
+      "loss": 0.8526,
+      "step": 5027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.211369920466661e-05,
+      "loss": 0.7105,
+      "step": 5028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.209576338716889e-05,
+      "loss": 0.7894,
+      "step": 5029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.207782861277684e-05,
+      "loss": 0.8503,
+      "step": 5030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.205989488385729e-05,
+      "loss": 0.8995,
+      "step": 5031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.204196220277692e-05,
+      "loss": 0.7855,
+      "step": 5032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2024030571902265e-05,
+      "loss": 0.8086,
+      "step": 5033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.200609999359972e-05,
+      "loss": 0.7337,
+      "step": 5034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1988170470235555e-05,
+      "loss": 0.7697,
+      "step": 5035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.197024200417588e-05,
+      "loss": 0.8263,
+      "step": 5036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.195231459778667e-05,
+      "loss": 0.7423,
+      "step": 5037
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.19343882534338e-05,
+      "loss": 0.817,
+      "step": 5038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1916462973482945e-05,
+      "loss": 0.7862,
+      "step": 5039
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1898538760299656e-05,
+      "loss": 0.8387,
+      "step": 5040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1880615616249396e-05,
+      "loss": 0.7314,
+      "step": 5041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.186269354369742e-05,
+      "loss": 0.7477,
+      "step": 5042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.184477254500887e-05,
+      "loss": 0.8551,
+      "step": 5043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1826852622548745e-05,
+      "loss": 0.7896,
+      "step": 5044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.180893377868189e-05,
+      "loss": 0.7782,
+      "step": 5045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.179101601577303e-05,
+      "loss": 0.7983,
+      "step": 5046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.177309933618672e-05,
+      "loss": 0.8554,
+      "step": 5047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1755183742287406e-05,
+      "loss": 0.7133,
+      "step": 5048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.173726923643933e-05,
+      "loss": 0.7079,
+      "step": 5049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.17193558210067e-05,
+      "loss": 0.8782,
+      "step": 5050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.170144349835347e-05,
+      "loss": 0.7892,
+      "step": 5051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1683532270843504e-05,
+      "loss": 0.7448,
+      "step": 5052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1665622140840495e-05,
+      "loss": 0.7731,
+      "step": 5053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.164771311070802e-05,
+      "loss": 0.7704,
+      "step": 5054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.162980518280949e-05,
+      "loss": 0.7249,
+      "step": 5055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.161189835950816e-05,
+      "loss": 0.8446,
+      "step": 5056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1593992643167194e-05,
+      "loss": 0.8026,
+      "step": 5057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1576088036149554e-05,
+      "loss": 0.7834,
+      "step": 5058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.155818454081805e-05,
+      "loss": 0.7717,
+      "step": 5059
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1540282159535407e-05,
+      "loss": 0.7126,
+      "step": 5060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.152238089466415e-05,
+      "loss": 0.6942,
+      "step": 5061
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.150448074856667e-05,
+      "loss": 0.6586,
+      "step": 5062
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1486581723605215e-05,
+      "loss": 0.7678,
+      "step": 5063
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1468683822141886e-05,
+      "loss": 0.8327,
+      "step": 5064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.145078704653861e-05,
+      "loss": 0.7258,
+      "step": 5065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.143289139915721e-05,
+      "loss": 0.7545,
+      "step": 5066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1414996882359335e-05,
+      "loss": 0.7099,
+      "step": 5067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.139710349850648e-05,
+      "loss": 0.7009,
+      "step": 5068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1379211249959994e-05,
+      "loss": 0.8128,
+      "step": 5069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.13613201390811e-05,
+      "loss": 0.8183,
+      "step": 5070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.134343016823082e-05,
+      "loss": 0.7607,
+      "step": 5071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.13255413397701e-05,
+      "loss": 0.8134,
+      "step": 5072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.130765365605965e-05,
+      "loss": 0.7696,
+      "step": 5073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1289767119460074e-05,
+      "loss": 0.7698,
+      "step": 5074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1271881732331854e-05,
+      "loss": 0.8468,
+      "step": 5075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1253997497035254e-05,
+      "loss": 0.7138,
+      "step": 5076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.123611441593044e-05,
+      "loss": 0.702,
+      "step": 5077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.121823249137739e-05,
+      "loss": 0.7651,
+      "step": 5078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.120035172573593e-05,
+      "loss": 0.8001,
+      "step": 5079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.118247212136579e-05,
+      "loss": 0.641,
+      "step": 5080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.116459368062649e-05,
+      "loss": 0.7417,
+      "step": 5081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.11467164058774e-05,
+      "loss": 0.7601,
+      "step": 5082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.112884029947775e-05,
+      "loss": 0.8586,
+      "step": 5083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.11109653637866e-05,
+      "loss": 0.7815,
+      "step": 5084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1093091601162895e-05,
+      "loss": 0.7266,
+      "step": 5085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.107521901396538e-05,
+      "loss": 0.7995,
+      "step": 5086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.105734760455266e-05,
+      "loss": 0.7832,
+      "step": 5087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.103947737528321e-05,
+      "loss": 0.7518,
+      "step": 5088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.102160832851528e-05,
+      "loss": 0.8609,
+      "step": 5089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.100374046660707e-05,
+      "loss": 0.8126,
+      "step": 5090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.098587379191654e-05,
+      "loss": 0.7774,
+      "step": 5091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.096800830680151e-05,
+      "loss": 0.8317,
+      "step": 5092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.095014401361965e-05,
+      "loss": 0.7028,
+      "step": 5093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.093228091472849e-05,
+      "loss": 0.7782,
+      "step": 5094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0914419012485375e-05,
+      "loss": 0.8009,
+      "step": 5095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.08965583092475e-05,
+      "loss": 0.833,
+      "step": 5096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.087869880737192e-05,
+      "loss": 0.7699,
+      "step": 5097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0860840509215496e-05,
+      "loss": 0.699,
+      "step": 5098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.084298341713494e-05,
+      "loss": 0.7213,
+      "step": 5099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.082512753348686e-05,
+      "loss": 0.7804,
+      "step": 5100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.080727286062764e-05,
+      "loss": 0.8055,
+      "step": 5101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0789419400913506e-05,
+      "loss": 0.8492,
+      "step": 5102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.077156715670056e-05,
+      "loss": 0.7624,
+      "step": 5103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.075371613034472e-05,
+      "loss": 0.7487,
+      "step": 5104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.073586632420176e-05,
+      "loss": 0.7632,
+      "step": 5105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.071801774062727e-05,
+      "loss": 0.8304,
+      "step": 5106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.070017038197668e-05,
+      "loss": 0.6865,
+      "step": 5107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.068232425060527e-05,
+      "loss": 0.7586,
+      "step": 5108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.066447934886819e-05,
+      "loss": 0.7673,
+      "step": 5109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.064663567912037e-05,
+      "loss": 0.7783,
+      "step": 5110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062879324371659e-05,
+      "loss": 0.7464,
+      "step": 5111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.061095204501151e-05,
+      "loss": 0.7873,
+      "step": 5112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059311208535958e-05,
+      "loss": 0.7386,
+      "step": 5113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0575273367115095e-05,
+      "loss": 0.7409,
+      "step": 5114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055743589263221e-05,
+      "loss": 0.8181,
+      "step": 5115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0539599664264885e-05,
+      "loss": 0.6826,
+      "step": 5116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052176468436693e-05,
+      "loss": 0.7407,
+      "step": 5117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050393095529198e-05,
+      "loss": 0.7083,
+      "step": 5118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048609847939355e-05,
+      "loss": 0.7561,
+      "step": 5119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0468267259024935e-05,
+      "loss": 0.7376,
+      "step": 5120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045043729653927e-05,
+      "loss": 0.7253,
+      "step": 5121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.043260859428957e-05,
+      "loss": 0.8258,
+      "step": 5122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0414781154628635e-05,
+      "loss": 0.7651,
+      "step": 5123
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.039695497990911e-05,
+      "loss": 0.7885,
+      "step": 5124
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0379130072483486e-05,
+      "loss": 0.8053,
+      "step": 5125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.036130643470407e-05,
+      "loss": 0.9255,
+      "step": 5126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034348406892301e-05,
+      "loss": 0.8497,
+      "step": 5127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0325662977492286e-05,
+      "loss": 0.8721,
+      "step": 5128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0307843162763735e-05,
+      "loss": 0.767,
+      "step": 5129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0290024627088965e-05,
+      "loss": 0.836,
+      "step": 5130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.027220737281949e-05,
+      "loss": 0.7918,
+      "step": 5131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.025439140230659e-05,
+      "loss": 0.6861,
+      "step": 5132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.02365767179014e-05,
+      "loss": 0.6977,
+      "step": 5133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0218763321954906e-05,
+      "loss": 0.7288,
+      "step": 5134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0200951216817886e-05,
+      "loss": 0.7526,
+      "step": 5135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.018314040484096e-05,
+      "loss": 0.7849,
+      "step": 5136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.016533088837461e-05,
+      "loss": 0.8214,
+      "step": 5137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.01475226697691e-05,
+      "loss": 0.8099,
+      "step": 5138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0129715751374546e-05,
+      "loss": 0.8538,
+      "step": 5139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.011191013554091e-05,
+      "loss": 0.665,
+      "step": 5140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.009410582461794e-05,
+      "loss": 0.8013,
+      "step": 5141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.007630282095525e-05,
+      "loss": 0.6955,
+      "step": 5142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.005850112690225e-05,
+      "loss": 0.699,
+      "step": 5143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0040700744808204e-05,
+      "loss": 0.8747,
+      "step": 5144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.002290167702218e-05,
+      "loss": 0.8361,
+      "step": 5145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.000510392589309e-05,
+      "loss": 0.7126,
+      "step": 5146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.998730749376968e-05,
+      "loss": 0.8371,
+      "step": 5147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9969512383000465e-05,
+      "loss": 0.8084,
+      "step": 5148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.995171859593389e-05,
+      "loss": 0.7889,
+      "step": 5149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9933926134918124e-05,
+      "loss": 0.8147,
+      "step": 5150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9916135002301203e-05,
+      "loss": 0.7753,
+      "step": 5151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9898345200431e-05,
+      "loss": 0.7733,
+      "step": 5152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.988055673165519e-05,
+      "loss": 0.7374,
+      "step": 5153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9862769598321295e-05,
+      "loss": 0.78,
+      "step": 5154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.98449838027766e-05,
+      "loss": 0.7618,
+      "step": 5155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.982719934736832e-05,
+      "loss": 0.7565,
+      "step": 5156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9809416234443394e-05,
+      "loss": 0.6838,
+      "step": 5157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.979163446634862e-05,
+      "loss": 0.7675,
+      "step": 5158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.977385404543065e-05,
+      "loss": 0.7855,
+      "step": 5159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9756074974035915e-05,
+      "loss": 0.7213,
+      "step": 5160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9738297254510684e-05,
+      "loss": 0.7215,
+      "step": 5161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.972052088920103e-05,
+      "loss": 0.773,
+      "step": 5162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.970274588045289e-05,
+      "loss": 0.7635,
+      "step": 5163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.968497223061196e-05,
+      "loss": 0.708,
+      "step": 5164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.966719994202382e-05,
+      "loss": 0.6912,
+      "step": 5165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.964942901703384e-05,
+      "loss": 0.732,
+      "step": 5166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.963165945798718e-05,
+      "loss": 0.8202,
+      "step": 5167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.96138912672289e-05,
+      "loss": 0.7071,
+      "step": 5168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.959612444710381e-05,
+      "loss": 0.6088,
+      "step": 5169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.957835899995655e-05,
+      "loss": 0.7854,
+      "step": 5170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9560594928131614e-05,
+      "loss": 0.6722,
+      "step": 5171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.954283223397327e-05,
+      "loss": 0.745,
+      "step": 5172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9525070919825614e-05,
+      "loss": 0.7584,
+      "step": 5173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.950731098803259e-05,
+      "loss": 0.7358,
+      "step": 5174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.948955244093794e-05,
+      "loss": 0.8506,
+      "step": 5175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.94717952808852e-05,
+      "loss": 0.6578,
+      "step": 5176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9454039510217755e-05,
+      "loss": 0.7511,
+      "step": 5177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9436285131278824e-05,
+      "loss": 0.7964,
+      "step": 5178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.941853214641138e-05,
+      "loss": 0.7718,
+      "step": 5179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.940078055795828e-05,
+      "loss": 0.8858,
+      "step": 5180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.938303036826214e-05,
+      "loss": 0.7872,
+      "step": 5181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.936528157966541e-05,
+      "loss": 0.7633,
+      "step": 5182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.934753419451038e-05,
+      "loss": 0.7238,
+      "step": 5183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.932978821513913e-05,
+      "loss": 0.8029,
+      "step": 5184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.931204364389356e-05,
+      "loss": 0.7772,
+      "step": 5185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.929430048311537e-05,
+      "loss": 0.7413,
+      "step": 5186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9276558735146096e-05,
+      "loss": 0.8216,
+      "step": 5187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9258818402327104e-05,
+      "loss": 0.665,
+      "step": 5188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.924107948699952e-05,
+      "loss": 0.7578,
+      "step": 5189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.922334199150432e-05,
+      "loss": 0.797,
+      "step": 5190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9205605918182295e-05,
+      "loss": 0.7711,
+      "step": 5191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.918787126937401e-05,
+      "loss": 0.7173,
+      "step": 5192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.91701380474199e-05,
+      "loss": 0.6977,
+      "step": 5193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9152406254660176e-05,
+      "loss": 0.8512,
+      "step": 5194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9134675893434843e-05,
+      "loss": 0.6316,
+      "step": 5195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9116946966083764e-05,
+      "loss": 0.7931,
+      "step": 5196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.909921947494657e-05,
+      "loss": 0.8453,
+      "step": 5197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9081493422362736e-05,
+      "loss": 0.765,
+      "step": 5198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.906376881067153e-05,
+      "loss": 0.7632,
+      "step": 5199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.904604564221204e-05,
+      "loss": 0.7668,
+      "step": 5200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.902832391932313e-05,
+      "loss": 0.8173,
+      "step": 5201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.901060364434353e-05,
+      "loss": 0.8225,
+      "step": 5202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.899288481961173e-05,
+      "loss": 0.7928,
+      "step": 5203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.897516744746604e-05,
+      "loss": 0.7293,
+      "step": 5204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8957451530244604e-05,
+      "loss": 0.7171,
+      "step": 5205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.893973707028534e-05,
+      "loss": 0.7664,
+      "step": 5206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8922024069925975e-05,
+      "loss": 0.8344,
+      "step": 5207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.89043125315041e-05,
+      "loss": 0.7804,
+      "step": 5208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.888660245735704e-05,
+      "loss": 0.8196,
+      "step": 5209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.886889384982195e-05,
+      "loss": 0.7691,
+      "step": 5210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.885118671123582e-05,
+      "loss": 0.7687,
+      "step": 5211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8833481043935424e-05,
+      "loss": 0.812,
+      "step": 5212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.881577685025733e-05,
+      "loss": 0.7803,
+      "step": 5213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.879807413253791e-05,
+      "loss": 0.7822,
+      "step": 5214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.878037289311338e-05,
+      "loss": 0.7547,
+      "step": 5215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8762673134319734e-05,
+      "loss": 0.7703,
+      "step": 5216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.874497485849273e-05,
+      "loss": 0.7663,
+      "step": 5217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.872727806796804e-05,
+      "loss": 0.6948,
+      "step": 5218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.870958276508103e-05,
+      "loss": 0.7782,
+      "step": 5219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.869188895216693e-05,
+      "loss": 0.7532,
+      "step": 5220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.867419663156074e-05,
+      "loss": 0.7786,
+      "step": 5221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.86565058055973e-05,
+      "loss": 0.7779,
+      "step": 5222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.863881647661121e-05,
+      "loss": 0.6648,
+      "step": 5223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.862112864693691e-05,
+      "loss": 0.6906,
+      "step": 5224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8603442318908615e-05,
+      "loss": 0.7986,
+      "step": 5225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.858575749486033e-05,
+      "loss": 0.7166,
+      "step": 5226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.856807417712594e-05,
+      "loss": 0.8603,
+      "step": 5227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.855039236803905e-05,
+      "loss": 0.7965,
+      "step": 5228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.853271206993307e-05,
+      "loss": 0.7086,
+      "step": 5229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8515033285141265e-05,
+      "loss": 0.8204,
+      "step": 5230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.849735601599665e-05,
+      "loss": 0.7943,
+      "step": 5231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.847968026483206e-05,
+      "loss": 0.7708,
+      "step": 5232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.846200603398014e-05,
+      "loss": 0.7647,
+      "step": 5233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8444333325773294e-05,
+      "loss": 0.7383,
+      "step": 5234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8426662142543765e-05,
+      "loss": 0.8161,
+      "step": 5235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.840899248662358e-05,
+      "loss": 0.716,
+      "step": 5236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.839132436034459e-05,
+      "loss": 0.7795,
+      "step": 5237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8373657766038394e-05,
+      "loss": 0.7705,
+      "step": 5238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8355992706036434e-05,
+      "loss": 0.7341,
+      "step": 5239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.833832918266992e-05,
+      "loss": 0.7867,
+      "step": 5240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8320667198269864e-05,
+      "loss": 0.7659,
+      "step": 5241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.830300675516711e-05,
+      "loss": 0.881,
+      "step": 5242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8285347855692225e-05,
+      "loss": 0.8052,
+      "step": 5243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8267690502175645e-05,
+      "loss": 0.7726,
+      "step": 5244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.825003469694759e-05,
+      "loss": 0.7826,
+      "step": 5245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8232380442338015e-05,
+      "loss": 0.6327,
+      "step": 5246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8214727740676745e-05,
+      "loss": 0.7833,
+      "step": 5247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.819707659429339e-05,
+      "loss": 0.6936,
+      "step": 5248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.81794270055173e-05,
+      "loss": 0.7682,
+      "step": 5249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8161778976677666e-05,
+      "loss": 0.7262,
+      "step": 5250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.814413251010347e-05,
+      "loss": 0.7126,
+      "step": 5251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.812648760812347e-05,
+      "loss": 0.7675,
+      "step": 5252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.810884427306622e-05,
+      "loss": 0.8316,
+      "step": 5253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8091202507260084e-05,
+      "loss": 0.6887,
+      "step": 5254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.807356231303321e-05,
+      "loss": 0.7382,
+      "step": 5255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8055923692713525e-05,
+      "loss": 0.7181,
+      "step": 5256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.803828664862879e-05,
+      "loss": 0.7572,
+      "step": 5257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8020651183106506e-05,
+      "loss": 0.7826,
+      "step": 5258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8003017298473995e-05,
+      "loss": 0.7457,
+      "step": 5259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7985384997058366e-05,
+      "loss": 0.7548,
+      "step": 5260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.796775428118651e-05,
+      "loss": 0.8698,
+      "step": 5261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.795012515318513e-05,
+      "loss": 0.8121,
+      "step": 5262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.793249761538067e-05,
+      "loss": 0.761,
+      "step": 5263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.791487167009945e-05,
+      "loss": 0.7171,
+      "step": 5264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.78972473196675e-05,
+      "loss": 0.6624,
+      "step": 5265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7879624566410655e-05,
+      "loss": 0.8126,
+      "step": 5266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.786200341265458e-05,
+      "loss": 0.8378,
+      "step": 5267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7844383860724706e-05,
+      "loss": 0.769,
+      "step": 5268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.782676591294624e-05,
+      "loss": 0.8834,
+      "step": 5269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.780914957164418e-05,
+      "loss": 0.8052,
+      "step": 5270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.779153483914331e-05,
+      "loss": 0.7382,
+      "step": 5271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.777392171776821e-05,
+      "loss": 0.705,
+      "step": 5272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7756310209843275e-05,
+      "loss": 0.6313,
+      "step": 5273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.773870031769264e-05,
+      "loss": 0.7729,
+      "step": 5274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.772109204364023e-05,
+      "loss": 0.7901,
+      "step": 5275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7703485390009796e-05,
+      "loss": 0.7487,
+      "step": 5276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7685880359124856e-05,
+      "loss": 0.7765,
+      "step": 5277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7668276953308695e-05,
+      "loss": 0.7396,
+      "step": 5278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7650675174884406e-05,
+      "loss": 0.8015,
+      "step": 5279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7633075026174845e-05,
+      "loss": 0.7399,
+      "step": 5280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7615476509502664e-05,
+      "loss": 0.8105,
+      "step": 5281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7597879627190334e-05,
+      "loss": 0.8387,
+      "step": 5282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.758028438156006e-05,
+      "loss": 0.7851,
+      "step": 5283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.756269077493385e-05,
+      "loss": 0.7779,
+      "step": 5284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.754509880963347e-05,
+      "loss": 0.81,
+      "step": 5285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7527508487980544e-05,
+      "loss": 0.8726,
+      "step": 5286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.750991981229641e-05,
+      "loss": 0.8312,
+      "step": 5287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7492332784902194e-05,
+      "loss": 0.6965,
+      "step": 5288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.747474740811884e-05,
+      "loss": 0.7463,
+      "step": 5289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.745716368426705e-05,
+      "loss": 0.7769,
+      "step": 5290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.743958161566729e-05,
+      "loss": 0.7896,
+      "step": 5291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.742200120463986e-05,
+      "loss": 0.7142,
+      "step": 5292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7404422453504785e-05,
+      "loss": 0.7767,
+      "step": 5293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7386845364581916e-05,
+      "loss": 0.7656,
+      "step": 5294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7369269940190834e-05,
+      "loss": 0.8342,
+      "step": 5295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.735169618265097e-05,
+      "loss": 0.7652,
+      "step": 5296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.733412409428148e-05,
+      "loss": 0.766,
+      "step": 5297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7316553677401324e-05,
+      "loss": 0.8716,
+      "step": 5298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.729898493432922e-05,
+      "loss": 0.7726,
+      "step": 5299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7281417867383664e-05,
+      "loss": 0.7481,
+      "step": 5300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.726385247888298e-05,
+      "loss": 0.6852,
+      "step": 5301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.724628877114523e-05,
+      "loss": 0.7061,
+      "step": 5302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.722872674648824e-05,
+      "loss": 0.8213,
+      "step": 5303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.721116640722965e-05,
+      "loss": 0.653,
+      "step": 5304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7193607755686836e-05,
+      "loss": 0.7045,
+      "step": 5305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.717605079417701e-05,
+      "loss": 0.7922,
+      "step": 5306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.715849552501712e-05,
+      "loss": 0.7995,
+      "step": 5307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.714094195052389e-05,
+      "loss": 0.723,
+      "step": 5308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7123390073013824e-05,
+      "loss": 0.6883,
+      "step": 5309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.710583989480323e-05,
+      "loss": 0.7581,
+      "step": 5310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.708829141820814e-05,
+      "loss": 0.8024,
+      "step": 5311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7070744645544407e-05,
+      "loss": 0.7376,
+      "step": 5312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7053199579127644e-05,
+      "loss": 0.7729,
+      "step": 5313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.703565622127323e-05,
+      "loss": 0.7848,
+      "step": 5314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.70181145742963e-05,
+      "loss": 0.7252,
+      "step": 5315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.700057464051184e-05,
+      "loss": 0.8234,
+      "step": 5316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6983036422234526e-05,
+      "loss": 0.7995,
+      "step": 5317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.696549992177885e-05,
+      "loss": 0.7926,
+      "step": 5318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.694796514145907e-05,
+      "loss": 0.7931,
+      "step": 5319
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.693043208358922e-05,
+      "loss": 0.7903,
+      "step": 5320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6912900750483074e-05,
+      "loss": 0.6974,
+      "step": 5321
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.689537114445423e-05,
+      "loss": 0.6202,
+      "step": 5322
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.687784326781603e-05,
+      "loss": 0.6576,
+      "step": 5323
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.686031712288158e-05,
+      "loss": 0.7394,
+      "step": 5324
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.684279271196377e-05,
+      "loss": 0.6834,
+      "step": 5325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6825270037375274e-05,
+      "loss": 0.8047,
+      "step": 5326
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.680774910142851e-05,
+      "loss": 0.6993,
+      "step": 5327
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6790229906435705e-05,
+      "loss": 0.7784,
+      "step": 5328
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6772712454708795e-05,
+      "loss": 0.7856,
+      "step": 5329
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.675519674855955e-05,
+      "loss": 0.7023,
+      "step": 5330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.673768279029946e-05,
+      "loss": 0.7838,
+      "step": 5331
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.672017058223982e-05,
+      "loss": 0.7268,
+      "step": 5332
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.670266012669168e-05,
+      "loss": 0.7208,
+      "step": 5333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.668515142596583e-05,
+      "loss": 0.9201,
+      "step": 5334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.66676444823729e-05,
+      "loss": 0.8025,
+      "step": 5335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.665013929822322e-05,
+      "loss": 0.7213,
+      "step": 5336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.663263587582691e-05,
+      "loss": 0.7995,
+      "step": 5337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.661513421749388e-05,
+      "loss": 0.7755,
+      "step": 5338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.659763432553377e-05,
+      "loss": 0.8201,
+      "step": 5339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.658013620225602e-05,
+      "loss": 0.7486,
+      "step": 5340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.65626398499698e-05,
+      "loss": 0.7396,
+      "step": 5341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.654514527098409e-05,
+      "loss": 0.7643,
+      "step": 5342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.652765246760758e-05,
+      "loss": 0.7521,
+      "step": 5343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.651016144214878e-05,
+      "loss": 0.8171,
+      "step": 5344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6492672196915964e-05,
+      "loss": 0.7799,
+      "step": 5345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.647518473421712e-05,
+      "loss": 0.7331,
+      "step": 5346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.645769905636005e-05,
+      "loss": 0.7805,
+      "step": 5347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.64402151656523e-05,
+      "loss": 0.7508,
+      "step": 5348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.642273306440118e-05,
+      "loss": 0.7324,
+      "step": 5349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.640525275491376e-05,
+      "loss": 0.7314,
+      "step": 5350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.638777423949689e-05,
+      "loss": 0.8149,
+      "step": 5351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6370297520457156e-05,
+      "loss": 0.7421,
+      "step": 5352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.635282260010096e-05,
+      "loss": 0.8384,
+      "step": 5353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.633534948073438e-05,
+      "loss": 0.7689,
+      "step": 5354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.631787816466335e-05,
+      "loss": 0.7225,
+      "step": 5355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6300408654193514e-05,
+      "loss": 0.8691,
+      "step": 5356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.628294095163028e-05,
+      "loss": 0.7284,
+      "step": 5357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6265475059278834e-05,
+      "loss": 0.8264,
+      "step": 5358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6248010979444095e-05,
+      "loss": 0.7867,
+      "step": 5359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6230548714430776e-05,
+      "loss": 0.7441,
+      "step": 5360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.621308826654334e-05,
+      "loss": 0.6709,
+      "step": 5361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.619562963808598e-05,
+      "loss": 0.7343,
+      "step": 5362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.617817283136271e-05,
+      "loss": 0.8342,
+      "step": 5363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.616071784867723e-05,
+      "loss": 0.762,
+      "step": 5364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.614326469233307e-05,
+      "loss": 0.7536,
+      "step": 5365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.612581336463349e-05,
+      "loss": 0.7649,
+      "step": 5366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6108363867881484e-05,
+      "loss": 0.8613,
+      "step": 5367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6090916204379835e-05,
+      "loss": 0.7058,
+      "step": 5368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6073470376431075e-05,
+      "loss": 0.7424,
+      "step": 5369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6056026386337487e-05,
+      "loss": 0.7724,
+      "step": 5370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6038584236401116e-05,
+      "loss": 0.7294,
+      "step": 5371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6021143928923785e-05,
+      "loss": 0.698,
+      "step": 5372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.600370546620705e-05,
+      "loss": 0.6761,
+      "step": 5373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.598626885055219e-05,
+      "loss": 0.7867,
+      "step": 5374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.596883408426034e-05,
+      "loss": 0.7983,
+      "step": 5375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.595140116963229e-05,
+      "loss": 0.6839,
+      "step": 5376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.593397010896865e-05,
+      "loss": 0.7544,
+      "step": 5377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.591654090456974e-05,
+      "loss": 0.7465,
+      "step": 5378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.589911355873567e-05,
+      "loss": 0.8574,
+      "step": 5379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.588168807376627e-05,
+      "loss": 0.7096,
+      "step": 5380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5864264451961164e-05,
+      "loss": 0.7662,
+      "step": 5381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5846842695619715e-05,
+      "loss": 0.7577,
+      "step": 5382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5829422807041025e-05,
+      "loss": 0.6948,
+      "step": 5383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5812004788523936e-05,
+      "loss": 0.8146,
+      "step": 5384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.579458864236712e-05,
+      "loss": 0.7146,
+      "step": 5385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5777174370868926e-05,
+      "loss": 0.7273,
+      "step": 5386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5759761976327474e-05,
+      "loss": 0.7684,
+      "step": 5387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.574235146104065e-05,
+      "loss": 0.8633,
+      "step": 5388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.572494282730606e-05,
+      "loss": 0.7512,
+      "step": 5389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5707536077421124e-05,
+      "loss": 0.7851,
+      "step": 5390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5690131213682944e-05,
+      "loss": 0.7456,
+      "step": 5391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.567272823838842e-05,
+      "loss": 0.6777,
+      "step": 5392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.565532715383416e-05,
+      "loss": 0.8711,
+      "step": 5393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5637927962316585e-05,
+      "loss": 0.7996,
+      "step": 5394
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.562053066613181e-05,
+      "loss": 0.7401,
+      "step": 5395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5603135267575737e-05,
+      "loss": 0.7651,
+      "step": 5396
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.558574176894398e-05,
+      "loss": 0.792,
+      "step": 5397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.556835017253194e-05,
+      "loss": 0.7626,
+      "step": 5398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.555096048063472e-05,
+      "loss": 0.726,
+      "step": 5399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.553357269554724e-05,
+      "loss": 0.74,
+      "step": 5400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5516186819564115e-05,
+      "loss": 0.7618,
+      "step": 5401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5498802854979714e-05,
+      "loss": 0.7374,
+      "step": 5402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.548142080408814e-05,
+      "loss": 0.7705,
+      "step": 5403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.546404066918333e-05,
+      "loss": 0.8145,
+      "step": 5404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5446662452558865e-05,
+      "loss": 0.7936,
+      "step": 5405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.542928615650811e-05,
+      "loss": 0.7222,
+      "step": 5406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.54119117833242e-05,
+      "loss": 0.7413,
+      "step": 5407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.539453933529996e-05,
+      "loss": 0.6941,
+      "step": 5408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.537716881472802e-05,
+      "loss": 0.8871,
+      "step": 5409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.535980022390074e-05,
+      "loss": 0.7205,
+      "step": 5410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.534243356511019e-05,
+      "loss": 0.7319,
+      "step": 5411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.532506884064824e-05,
+      "loss": 0.8009,
+      "step": 5412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.530770605280643e-05,
+      "loss": 0.8059,
+      "step": 5413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5290345203876154e-05,
+      "loss": 0.7517,
+      "step": 5414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5272986296148456e-05,
+      "loss": 0.7562,
+      "step": 5415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.525562933191415e-05,
+      "loss": 0.7705,
+      "step": 5416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.523827431346381e-05,
+      "loss": 0.7633,
+      "step": 5417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.522092124308773e-05,
+      "loss": 0.7718,
+      "step": 5418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.520357012307597e-05,
+      "loss": 0.7162,
+      "step": 5419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5186220955718306e-05,
+      "loss": 0.8491,
+      "step": 5420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.51688737433043e-05,
+      "loss": 0.746,
+      "step": 5421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.51515284881232e-05,
+      "loss": 0.8082,
+      "step": 5422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.513418519246401e-05,
+      "loss": 0.7485,
+      "step": 5423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.511684385861553e-05,
+      "loss": 0.7124,
+      "step": 5424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.509950448886624e-05,
+      "loss": 0.7204,
+      "step": 5425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.508216708550438e-05,
+      "loss": 0.8094,
+      "step": 5426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.506483165081794e-05,
+      "loss": 0.7548,
+      "step": 5427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.504749818709463e-05,
+      "loss": 0.7367,
+      "step": 5428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.503016669662192e-05,
+      "loss": 0.7328,
+      "step": 5429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5012837181687e-05,
+      "loss": 0.8522,
+      "step": 5430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4995509644576825e-05,
+      "loss": 0.7801,
+      "step": 5431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4978184087578056e-05,
+      "loss": 0.8055,
+      "step": 5432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.496086051297711e-05,
+      "loss": 0.7765,
+      "step": 5433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.494353892306016e-05,
+      "loss": 0.8397,
+      "step": 5434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.49262193201131e-05,
+      "loss": 0.8692,
+      "step": 5435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.490890170642155e-05,
+      "loss": 0.7136,
+      "step": 5436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4891586084270886e-05,
+      "loss": 0.7896,
+      "step": 5437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.487427245594622e-05,
+      "loss": 0.7148,
+      "step": 5438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4856960823732385e-05,
+      "loss": 0.7031,
+      "step": 5439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4839651189913966e-05,
+      "loss": 0.7462,
+      "step": 5440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.482234355677528e-05,
+      "loss": 0.7864,
+      "step": 5441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.480503792660036e-05,
+      "loss": 0.8385,
+      "step": 5442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.478773430167302e-05,
+      "loss": 0.7273,
+      "step": 5443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.477043268427679e-05,
+      "loss": 0.7268,
+      "step": 5444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.475313307669489e-05,
+      "loss": 0.6528,
+      "step": 5445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.473583548121036e-05,
+      "loss": 0.6569,
+      "step": 5446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4718539900105894e-05,
+      "loss": 0.6785,
+      "step": 5447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.470124633566397e-05,
+      "loss": 0.7097,
+      "step": 5448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.468395479016678e-05,
+      "loss": 0.7426,
+      "step": 5449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.466666526589625e-05,
+      "loss": 0.7487,
+      "step": 5450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.464937776513404e-05,
+      "loss": 0.7019,
+      "step": 5451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4632092290161546e-05,
+      "loss": 0.7403,
+      "step": 5452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4614808843259905e-05,
+      "loss": 0.7743,
+      "step": 5453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.459752742670997e-05,
+      "loss": 0.7509,
+      "step": 5454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.458024804279234e-05,
+      "loss": 0.7861,
+      "step": 5455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.456297069378734e-05,
+      "loss": 0.7772,
+      "step": 5456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.454569538197502e-05,
+      "loss": 0.7946,
+      "step": 5457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.452842210963517e-05,
+      "loss": 0.7513,
+      "step": 5458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.45111508790473e-05,
+      "loss": 0.8413,
+      "step": 5459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4493881692490675e-05,
+      "loss": 0.879,
+      "step": 5460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4476614552244244e-05,
+      "loss": 0.8251,
+      "step": 5461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.445934946058674e-05,
+      "loss": 0.7353,
+      "step": 5462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4442086419796605e-05,
+      "loss": 0.7741,
+      "step": 5463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4424825432152e-05,
+      "loss": 0.7249,
+      "step": 5464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.440756649993082e-05,
+      "loss": 0.8665,
+      "step": 5465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4390309625410686e-05,
+      "loss": 0.7782,
+      "step": 5466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.437305481086896e-05,
+      "loss": 0.6661,
+      "step": 5467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.435580205858272e-05,
+      "loss": 0.746,
+      "step": 5468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.433855137082878e-05,
+      "loss": 0.7927,
+      "step": 5469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4321302749883654e-05,
+      "loss": 0.801,
+      "step": 5470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.430405619802365e-05,
+      "loss": 0.7362,
+      "step": 5471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4286811717524715e-05,
+      "loss": 0.7939,
+      "step": 5472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4269569310662584e-05,
+      "loss": 0.6395,
+      "step": 5473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.425232897971273e-05,
+      "loss": 0.6903,
+      "step": 5474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.423509072695029e-05,
+      "loss": 0.8427,
+      "step": 5475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.421785455465017e-05,
+      "loss": 0.6345,
+      "step": 5476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.420062046508699e-05,
+      "loss": 0.7741,
+      "step": 5477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.41833884605351e-05,
+      "loss": 0.7394,
+      "step": 5478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.416615854326856e-05,
+      "loss": 0.7148,
+      "step": 5479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.414893071556119e-05,
+      "loss": 0.6403,
+      "step": 5480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4131704979686494e-05,
+      "loss": 0.6968,
+      "step": 5481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4114481337917704e-05,
+      "loss": 0.722,
+      "step": 5482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.409725979252782e-05,
+      "loss": 0.8103,
+      "step": 5483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4080040345789513e-05,
+      "loss": 0.8029,
+      "step": 5484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.406282299997521e-05,
+      "loss": 0.8267,
+      "step": 5485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.404560775735704e-05,
+      "loss": 0.8345,
+      "step": 5486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.402839462020686e-05,
+      "loss": 0.807,
+      "step": 5487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.401118359079625e-05,
+      "loss": 0.6886,
+      "step": 5488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.399397467139652e-05,
+      "loss": 0.7304,
+      "step": 5489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.39767678642787e-05,
+      "loss": 0.6617,
+      "step": 5490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.395956317171354e-05,
+      "loss": 0.7422,
+      "step": 5491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3942360595971466e-05,
+      "loss": 0.7728,
+      "step": 5492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.392516013932272e-05,
+      "loss": 0.8544,
+      "step": 5493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.390796180403719e-05,
+      "loss": 0.6686,
+      "step": 5494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.389076559238451e-05,
+      "loss": 0.7964,
+      "step": 5495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3873571506634015e-05,
+      "loss": 0.7326,
+      "step": 5496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.385637954905477e-05,
+      "loss": 0.7635,
+      "step": 5497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3839189721915586e-05,
+      "loss": 0.7832,
+      "step": 5498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.382200202748495e-05,
+      "loss": 0.8589,
+      "step": 5499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.38048164680311e-05,
+      "loss": 0.7817,
+      "step": 5500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.378763304582197e-05,
+      "loss": 0.7492,
+      "step": 5501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.377045176312519e-05,
+      "loss": 0.785,
+      "step": 5502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.375327262220819e-05,
+      "loss": 0.6976,
+      "step": 5503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.373609562533804e-05,
+      "loss": 0.7418,
+      "step": 5504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.371892077478157e-05,
+      "loss": 0.7874,
+      "step": 5505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.370174807280528e-05,
+      "loss": 0.6804,
+      "step": 5506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.368457752167543e-05,
+      "loss": 0.6946,
+      "step": 5507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.366740912365799e-05,
+      "loss": 0.8144,
+      "step": 5508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.365024288101863e-05,
+      "loss": 0.7443,
+      "step": 5509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3633078796022754e-05,
+      "loss": 0.7939,
+      "step": 5510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.361591687093544e-05,
+      "loss": 0.8001,
+      "step": 5511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3598757108021546e-05,
+      "loss": 0.744,
+      "step": 5512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.35815995095456e-05,
+      "loss": 0.836,
+      "step": 5513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.356444407777185e-05,
+      "loss": 0.8108,
+      "step": 5514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.354729081496426e-05,
+      "loss": 0.7648,
+      "step": 5515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.353013972338651e-05,
+      "loss": 0.7536,
+      "step": 5516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3512990805302014e-05,
+      "loss": 0.6919,
+      "step": 5517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3495844062973856e-05,
+      "loss": 0.8101,
+      "step": 5518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.347869949866487e-05,
+      "loss": 0.7828,
+      "step": 5519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3461557114637575e-05,
+      "loss": 0.752,
+      "step": 5520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.344441691315421e-05,
+      "loss": 0.8046,
+      "step": 5521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.342727889647675e-05,
+      "loss": 0.8016,
+      "step": 5522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.341014306686688e-05,
+      "loss": 0.7404,
+      "step": 5523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3393009426585946e-05,
+      "loss": 0.7718,
+      "step": 5524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.337587797789504e-05,
+      "loss": 0.7259,
+      "step": 5525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.335874872305499e-05,
+      "loss": 0.757,
+      "step": 5526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.33416216643263e-05,
+      "loss": 0.7621,
+      "step": 5527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.332449680396917e-05,
+      "loss": 0.7764,
+      "step": 5528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.330737414424355e-05,
+      "loss": 0.7628,
+      "step": 5529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.329025368740909e-05,
+      "loss": 0.6635,
+      "step": 5530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.32731354357251e-05,
+      "loss": 0.7586,
+      "step": 5531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.325601939145069e-05,
+      "loss": 0.7927,
+      "step": 5532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.323890555684462e-05,
+      "loss": 0.7692,
+      "step": 5533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.322179393416534e-05,
+      "loss": 0.6732,
+      "step": 5534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.320468452567106e-05,
+      "loss": 0.7631,
+      "step": 5535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3187577333619665e-05,
+      "loss": 0.7218,
+      "step": 5536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3170472360268754e-05,
+      "loss": 0.7413,
+      "step": 5537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.315336960787564e-05,
+      "loss": 0.8188,
+      "step": 5538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3136269078697335e-05,
+      "loss": 0.7443,
+      "step": 5539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.311917077499056e-05,
+      "loss": 0.7582,
+      "step": 5540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.310207469901172e-05,
+      "loss": 0.8362,
+      "step": 5541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.308498085301699e-05,
+      "loss": 0.8117,
+      "step": 5542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3067889239262195e-05,
+      "loss": 0.7836,
+      "step": 5543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3050799860002854e-05,
+      "loss": 0.7823,
+      "step": 5544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3033712717494255e-05,
+      "loss": 0.8133,
+      "step": 5545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3016627813991336e-05,
+      "loss": 0.7857,
+      "step": 5546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2999545151748756e-05,
+      "loss": 0.7571,
+      "step": 5547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.298246473302087e-05,
+      "loss": 0.6977,
+      "step": 5548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.296538656006176e-05,
+      "loss": 0.7465,
+      "step": 5549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.294831063512518e-05,
+      "loss": 0.8111,
+      "step": 5550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.293123696046461e-05,
+      "loss": 0.7783,
+      "step": 5551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.291416553833325e-05,
+      "loss": 0.7891,
+      "step": 5552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.289709637098394e-05,
+      "loss": 0.7818,
+      "step": 5553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.288002946066931e-05,
+      "loss": 0.7447,
+      "step": 5554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.286296480964161e-05,
+      "loss": 0.7975,
+      "step": 5555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.284590242015283e-05,
+      "loss": 0.7311,
+      "step": 5556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2828842294454675e-05,
+      "loss": 0.7509,
+      "step": 5557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.281178443479852e-05,
+      "loss": 0.7717,
+      "step": 5558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2794728843435445e-05,
+      "loss": 0.7039,
+      "step": 5559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.277767552261626e-05,
+      "loss": 0.6779,
+      "step": 5560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2760624474591454e-05,
+      "loss": 0.8353,
+      "step": 5561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.274357570161121e-05,
+      "loss": 0.7517,
+      "step": 5562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.272652920592543e-05,
+      "loss": 0.7332,
+      "step": 5563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.270948498978371e-05,
+      "loss": 0.654,
+      "step": 5564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2692443055435315e-05,
+      "loss": 0.7676,
+      "step": 5565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2675403405129256e-05,
+      "loss": 0.7494,
+      "step": 5566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2658366041114206e-05,
+      "loss": 0.7013,
+      "step": 5567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2641330965638564e-05,
+      "loss": 0.7633,
+      "step": 5568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.262429818095039e-05,
+      "loss": 0.8289,
+      "step": 5569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.260726768929748e-05,
+      "loss": 0.8015,
+      "step": 5570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.259023949292731e-05,
+      "loss": 0.6571,
+      "step": 5571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.257321359408708e-05,
+      "loss": 0.6807,
+      "step": 5572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.255618999502363e-05,
+      "loss": 0.7234,
+      "step": 5573
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.253916869798355e-05,
+      "loss": 0.7481,
+      "step": 5574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2522149705213095e-05,
+      "loss": 0.8694,
+      "step": 5575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.250513301895822e-05,
+      "loss": 0.7849,
+      "step": 5576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.24881186414646e-05,
+      "loss": 0.7778,
+      "step": 5577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.247110657497756e-05,
+      "loss": 0.7051,
+      "step": 5578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.245409682174217e-05,
+      "loss": 0.7143,
+      "step": 5579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.243708938400316e-05,
+      "loss": 0.6926,
+      "step": 5580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2420084264004965e-05,
+      "loss": 0.7978,
+      "step": 5581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.240308146399173e-05,
+      "loss": 0.7935,
+      "step": 5582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2386080986207274e-05,
+      "loss": 0.6918,
+      "step": 5583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.236908283289511e-05,
+      "loss": 0.7685,
+      "step": 5584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.235208700629845e-05,
+      "loss": 0.7767,
+      "step": 5585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.23350935086602e-05,
+      "loss": 0.736,
+      "step": 5586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2318102342222944e-05,
+      "loss": 0.8487,
+      "step": 5587
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2301113509228995e-05,
+      "loss": 0.7265,
+      "step": 5588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.228412701192032e-05,
+      "loss": 0.7546,
+      "step": 5589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.226714285253856e-05,
+      "loss": 0.7442,
+      "step": 5590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.225016103332514e-05,
+      "loss": 0.6879,
+      "step": 5591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.223318155652109e-05,
+      "loss": 0.7396,
+      "step": 5592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2216204424367155e-05,
+      "loss": 0.7616,
+      "step": 5593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.219922963910376e-05,
+      "loss": 0.7717,
+      "step": 5594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.218225720297105e-05,
+      "loss": 0.7632,
+      "step": 5595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.216528711820882e-05,
+      "loss": 0.8221,
+      "step": 5596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.21483193870566e-05,
+      "loss": 0.8035,
+      "step": 5597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.213135401175358e-05,
+      "loss": 0.7985,
+      "step": 5598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2114390994538644e-05,
+      "loss": 0.7422,
+      "step": 5599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.209743033765035e-05,
+      "loss": 0.8164,
+      "step": 5600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2080472043326984e-05,
+      "loss": 0.7779,
+      "step": 5601
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.206351611380649e-05,
+      "loss": 0.8188,
+      "step": 5602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.204656255132651e-05,
+      "loss": 0.7722,
+      "step": 5603
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.202961135812437e-05,
+      "loss": 0.7894,
+      "step": 5604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.201266253643707e-05,
+      "loss": 0.6598,
+      "step": 5605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1995716088501316e-05,
+      "loss": 0.7217,
+      "step": 5606
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.197877201655351e-05,
+      "loss": 0.7337,
+      "step": 5607
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1961830322829706e-05,
+      "loss": 0.7571,
+      "step": 5608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.194489100956568e-05,
+      "loss": 0.7969,
+      "step": 5609
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1927954078996845e-05,
+      "loss": 0.6841,
+      "step": 5610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.191101953335839e-05,
+      "loss": 0.8785,
+      "step": 5611
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1894087374885095e-05,
+      "loss": 0.7107,
+      "step": 5612
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.187715760581147e-05,
+      "loss": 0.7733,
+      "step": 5613
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.18602302283717e-05,
+      "loss": 0.7519,
+      "step": 5614
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1843305244799645e-05,
+      "loss": 0.7397,
+      "step": 5615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1826382657328875e-05,
+      "loss": 0.7449,
+      "step": 5616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.180946246819263e-05,
+      "loss": 0.7701,
+      "step": 5617
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.179254467962382e-05,
+      "loss": 0.8089,
+      "step": 5618
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1775629293855035e-05,
+      "loss": 0.7851,
+      "step": 5619
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1758716313118606e-05,
+      "loss": 0.7706,
+      "step": 5620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.174180573964649e-05,
+      "loss": 0.8291,
+      "step": 5621
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.172489757567033e-05,
+      "loss": 0.8157,
+      "step": 5622
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1707991823421454e-05,
+      "loss": 0.7596,
+      "step": 5623
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1691088485130896e-05,
+      "loss": 0.7771,
+      "step": 5624
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1674187563029346e-05,
+      "loss": 0.6499,
+      "step": 5625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.165728905934718e-05,
+      "loss": 0.7361,
+      "step": 5626
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.164039297631447e-05,
+      "loss": 0.8,
+      "step": 5627
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.162349931616095e-05,
+      "loss": 0.7516,
+      "step": 5628
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.160660808111602e-05,
+      "loss": 0.7909,
+      "step": 5629
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1589719273408824e-05,
+      "loss": 0.836,
+      "step": 5630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1572832895268124e-05,
+      "loss": 0.7985,
+      "step": 5631
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1555948948922376e-05,
+      "loss": 0.857,
+      "step": 5632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.153906743659971e-05,
+      "loss": 0.7296,
+      "step": 5633
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.152218836052797e-05,
+      "loss": 0.7525,
+      "step": 5634
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.150531172293463e-05,
+      "loss": 0.803,
+      "step": 5635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1488437526046874e-05,
+      "loss": 0.7551,
+      "step": 5636
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.147156577209156e-05,
+      "loss": 0.7305,
+      "step": 5637
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1454696463295206e-05,
+      "loss": 0.6893,
+      "step": 5638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.143782960188401e-05,
+      "loss": 0.7208,
+      "step": 5639
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.142096519008389e-05,
+      "loss": 0.7386,
+      "step": 5640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.140410323012039e-05,
+      "loss": 0.7625,
+      "step": 5641
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.138724372421875e-05,
+      "loss": 0.7749,
+      "step": 5642
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.137038667460386e-05,
+      "loss": 0.63,
+      "step": 5643
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.135353208350035e-05,
+      "loss": 0.7396,
+      "step": 5644
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.133667995313247e-05,
+      "loss": 0.7724,
+      "step": 5645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.131983028572414e-05,
+      "loss": 0.6323,
+      "step": 5646
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.130298308349901e-05,
+      "loss": 0.6665,
+      "step": 5647
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.128613834868035e-05,
+      "loss": 0.781,
+      "step": 5648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1269296083491094e-05,
+      "loss": 0.7415,
+      "step": 5649
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1252456290153954e-05,
+      "loss": 0.6982,
+      "step": 5650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.123561897089119e-05,
+      "loss": 0.846,
+      "step": 5651
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1218784127924794e-05,
+      "loss": 0.7456,
+      "step": 5652
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1201951763476437e-05,
+      "loss": 0.7619,
+      "step": 5653
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1185121879767455e-05,
+      "loss": 0.7319,
+      "step": 5654
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.116829447901883e-05,
+      "loss": 0.8094,
+      "step": 5655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.115146956345127e-05,
+      "loss": 0.808,
+      "step": 5656
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.11346471352851e-05,
+      "loss": 0.7345,
+      "step": 5657
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.111782719674033e-05,
+      "loss": 0.8312,
+      "step": 5658
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.110100975003669e-05,
+      "loss": 0.7417,
+      "step": 5659
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1084194797393516e-05,
+      "loss": 0.735,
+      "step": 5660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.106738234102985e-05,
+      "loss": 0.8053,
+      "step": 5661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.10505723831644e-05,
+      "loss": 0.7063,
+      "step": 5662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1033764926015547e-05,
+      "loss": 0.7835,
+      "step": 5663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1016959971801326e-05,
+      "loss": 0.7753,
+      "step": 5664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.100015752273945e-05,
+      "loss": 0.8009,
+      "step": 5665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.09833575810473e-05,
+      "loss": 0.7747,
+      "step": 5666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0966560148941934e-05,
+      "loss": 0.8267,
+      "step": 5667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0949765228640095e-05,
+      "loss": 0.808,
+      "step": 5668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.093297282235813e-05,
+      "loss": 0.7459,
+      "step": 5669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0916182932312134e-05,
+      "loss": 0.7282,
+      "step": 5670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0899395560717824e-05,
+      "loss": 0.7669,
+      "step": 5671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0882610709790607e-05,
+      "loss": 0.7182,
+      "step": 5672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.086582838174551e-05,
+      "loss": 0.8799,
+      "step": 5673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0849048578797305e-05,
+      "loss": 0.7856,
+      "step": 5674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0832271303160356e-05,
+      "loss": 0.8461,
+      "step": 5675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.081549655704874e-05,
+      "loss": 0.7305,
+      "step": 5676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0798724342676164e-05,
+      "loss": 0.6961,
+      "step": 5677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.078195466225604e-05,
+      "loss": 0.7885,
+      "step": 5678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.076518751800144e-05,
+      "loss": 0.7268,
+      "step": 5679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.074842291212506e-05,
+      "loss": 0.7713,
+      "step": 5680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.073166084683933e-05,
+      "loss": 0.6844,
+      "step": 5681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0714901324356274e-05,
+      "loss": 0.762,
+      "step": 5682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0698144346887616e-05,
+      "loss": 0.7788,
+      "step": 5683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.068138991664474e-05,
+      "loss": 0.7516,
+      "step": 5684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.06646380358387e-05,
+      "loss": 0.7667,
+      "step": 5685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.064788870668019e-05,
+      "loss": 0.7426,
+      "step": 5686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.063114193137961e-05,
+      "loss": 0.772,
+      "step": 5687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.061439771214698e-05,
+      "loss": 0.8288,
+      "step": 5688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0597656051191994e-05,
+      "loss": 0.7476,
+      "step": 5689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0580916950724025e-05,
+      "loss": 0.805,
+      "step": 5690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0564180412952116e-05,
+      "loss": 0.7254,
+      "step": 5691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0547446440084917e-05,
+      "loss": 0.7575,
+      "step": 5692
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.05307150343308e-05,
+      "loss": 0.8325,
+      "step": 5693
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.051398619789776e-05,
+      "loss": 0.7245,
+      "step": 5694
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.049725993299347e-05,
+      "loss": 0.7924,
+      "step": 5695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0480536241825263e-05,
+      "loss": 0.6919,
+      "step": 5696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.046381512660014e-05,
+      "loss": 0.7661,
+      "step": 5697
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0447096589524714e-05,
+      "loss": 0.8521,
+      "step": 5698
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0430380632805343e-05,
+      "loss": 0.7462,
+      "step": 5699
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.041366725864798e-05,
+      "loss": 0.7205,
+      "step": 5700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0396956469258252e-05,
+      "loss": 0.8367,
+      "step": 5701
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.038024826684145e-05,
+      "loss": 0.7068,
+      "step": 5702
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.036354265360251e-05,
+      "loss": 0.7254,
+      "step": 5703
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0346839631746038e-05,
+      "loss": 0.7391,
+      "step": 5704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0330139203476314e-05,
+      "loss": 0.7251,
+      "step": 5705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.031344137099724e-05,
+      "loss": 0.6987,
+      "step": 5706
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.029674613651241e-05,
+      "loss": 0.7848,
+      "step": 5707
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0280053502225038e-05,
+      "loss": 0.8037,
+      "step": 5708
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.026336347033804e-05,
+      "loss": 0.7514,
+      "step": 5709
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0246676043053957e-05,
+      "loss": 0.7497,
+      "step": 5710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0229991222574992e-05,
+      "loss": 0.8119,
+      "step": 5711
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.021330901110301e-05,
+      "loss": 0.6943,
+      "step": 5712
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0196629410839522e-05,
+      "loss": 0.7479,
+      "step": 5713
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.017995242398569e-05,
+      "loss": 0.7704,
+      "step": 5714
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0163278052742366e-05,
+      "loss": 0.7522,
+      "step": 5715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0146606299310022e-05,
+      "loss": 0.7878,
+      "step": 5716
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.012993716588879e-05,
+      "loss": 0.8875,
+      "step": 5717
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.011327065467844e-05,
+      "loss": 0.7662,
+      "step": 5718
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.009660676787846e-05,
+      "loss": 0.7405,
+      "step": 5719
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.007994550768793e-05,
+      "loss": 0.7654,
+      "step": 5720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0063286876305602e-05,
+      "loss": 0.7521,
+      "step": 5721
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0046630875929876e-05,
+      "loss": 0.8001,
+      "step": 5722
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0029977508758806e-05,
+      "loss": 0.818,
+      "step": 5723
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.001332677699012e-05,
+      "loss": 0.6152,
+      "step": 5724
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9996678682821167e-05,
+      "loss": 0.7862,
+      "step": 5725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9980033228448968e-05,
+      "loss": 0.7222,
+      "step": 5726
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.996339041607018e-05,
+      "loss": 0.7201,
+      "step": 5727
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9946750247881105e-05,
+      "loss": 0.8056,
+      "step": 5728
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9930112726077763e-05,
+      "loss": 0.7279,
+      "step": 5729
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.991347785285573e-05,
+      "loss": 0.7056,
+      "step": 5730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9896845630410296e-05,
+      "loss": 0.7267,
+      "step": 5731
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9880216060936363e-05,
+      "loss": 0.7808,
+      "step": 5732
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.986358914662851e-05,
+      "loss": 0.6767,
+      "step": 5733
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9846964889680967e-05,
+      "loss": 0.7505,
+      "step": 5734
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.983034329228759e-05,
+      "loss": 0.7249,
+      "step": 5735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.98137243566419e-05,
+      "loss": 0.7811,
+      "step": 5736
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9797108084937044e-05,
+      "loss": 0.6825,
+      "step": 5737
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9780494479365866e-05,
+      "loss": 0.6339,
+      "step": 5738
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.976388354212083e-05,
+      "loss": 0.7483,
+      "step": 5739
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9747275275394027e-05,
+      "loss": 0.68,
+      "step": 5740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9730669681377215e-05,
+      "loss": 0.8249,
+      "step": 5741
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9714066762261823e-05,
+      "loss": 0.7473,
+      "step": 5742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.969746652023888e-05,
+      "loss": 0.8349,
+      "step": 5743
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.968086895749911e-05,
+      "loss": 0.7842,
+      "step": 5744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.966427407623283e-05,
+      "loss": 0.7628,
+      "step": 5745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.964768187863005e-05,
+      "loss": 0.7555,
+      "step": 5746
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9631092366880388e-05,
+      "loss": 0.7864,
+      "step": 5747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9614505543173155e-05,
+      "loss": 0.7593,
+      "step": 5748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.959792140969727e-05,
+      "loss": 0.7612,
+      "step": 5749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9581339968641302e-05,
+      "loss": 0.6985,
+      "step": 5750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9564761222193464e-05,
+      "loss": 0.8139,
+      "step": 5751
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9548185172541636e-05,
+      "loss": 0.7411,
+      "step": 5752
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9531611821873318e-05,
+      "loss": 0.8899,
+      "step": 5753
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.951504117237565e-05,
+      "loss": 0.7528,
+      "step": 5754
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.949847322623544e-05,
+      "loss": 0.7723,
+      "step": 5755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9481907985639117e-05,
+      "loss": 0.7786,
+      "step": 5756
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.946534545277274e-05,
+      "loss": 0.8467,
+      "step": 5757
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9448785629822086e-05,
+      "loss": 0.7601,
+      "step": 5758
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.943222851897248e-05,
+      "loss": 0.7973,
+      "step": 5759
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.941567412240893e-05,
+      "loss": 0.7536,
+      "step": 5760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9399122442316107e-05,
+      "loss": 0.7798,
+      "step": 5761
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9382573480878295e-05,
+      "loss": 0.6795,
+      "step": 5762
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9366027240279415e-05,
+      "loss": 0.8049,
+      "step": 5763
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9349483722703047e-05,
+      "loss": 0.7781,
+      "step": 5764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9332942930332403e-05,
+      "loss": 0.7106,
+      "step": 5765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9316404865350326e-05,
+      "loss": 0.7344,
+      "step": 5766
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.929986952993933e-05,
+      "loss": 0.6655,
+      "step": 5767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9283336926281536e-05,
+      "loss": 0.7439,
+      "step": 5768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.926680705655872e-05,
+      "loss": 0.8329,
+      "step": 5769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9250279922952306e-05,
+      "loss": 0.8106,
+      "step": 5770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9233755527643336e-05,
+      "loss": 0.7662,
+      "step": 5771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.92172338728125e-05,
+      "loss": 0.803,
+      "step": 5772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9200714960640118e-05,
+      "loss": 0.7727,
+      "step": 5773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9184198793306168e-05,
+      "loss": 0.7029,
+      "step": 5774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.916768537299025e-05,
+      "loss": 0.727,
+      "step": 5775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9151174701871614e-05,
+      "loss": 0.7827,
+      "step": 5776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9134666782129106e-05,
+      "loss": 0.8123,
+      "step": 5777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9118161615941297e-05,
+      "loss": 0.7934,
+      "step": 5778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9101659205486297e-05,
+      "loss": 0.7868,
+      "step": 5779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.908515955294192e-05,
+      "loss": 0.7684,
+      "step": 5780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9068662660485573e-05,
+      "loss": 0.7205,
+      "step": 5781
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9052168530294323e-05,
+      "loss": 0.7614,
+      "step": 5782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9035677164544865e-05,
+      "loss": 0.7487,
+      "step": 5783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9019188565413536e-05,
+      "loss": 0.8104,
+      "step": 5784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.900270273507629e-05,
+      "loss": 0.7859,
+      "step": 5785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8986219675708735e-05,
+      "loss": 0.7945,
+      "step": 5786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.896973938948609e-05,
+      "loss": 0.7958,
+      "step": 5787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.895326187858326e-05,
+      "loss": 0.7442,
+      "step": 5788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8936787145174727e-05,
+      "loss": 0.7451,
+      "step": 5789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8920315191434622e-05,
+      "loss": 0.742,
+      "step": 5790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.890384601953671e-05,
+      "loss": 0.7615,
+      "step": 5791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.888737963165442e-05,
+      "loss": 0.8141,
+      "step": 5792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.887091602996077e-05,
+      "loss": 0.7283,
+      "step": 5793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.885445521662843e-05,
+      "loss": 0.6909,
+      "step": 5794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8837997193829696e-05,
+      "loss": 0.8222,
+      "step": 5795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.882154196373648e-05,
+      "loss": 0.8007,
+      "step": 5796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.880508952852039e-05,
+      "loss": 0.7322,
+      "step": 5797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8788639890352588e-05,
+      "loss": 0.809,
+      "step": 5798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8772193051403912e-05,
+      "loss": 0.7453,
+      "step": 5799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.875574901384481e-05,
+      "loss": 0.8156,
+      "step": 5800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8739307779845366e-05,
+      "loss": 0.7904,
+      "step": 5801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.87228693515753e-05,
+      "loss": 0.7975,
+      "step": 5802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8706433731203953e-05,
+      "loss": 0.8353,
+      "step": 5803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.86900009209003e-05,
+      "loss": 0.7416,
+      "step": 5804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8673570922832947e-05,
+      "loss": 0.7395,
+      "step": 5805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8657143739170113e-05,
+      "loss": 0.7356,
+      "step": 5806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8640719372079683e-05,
+      "loss": 0.8604,
+      "step": 5807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.862429782372914e-05,
+      "loss": 0.7493,
+      "step": 5808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8607879096285573e-05,
+      "loss": 0.7436,
+      "step": 5809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.859146319191578e-05,
+      "loss": 0.8161,
+      "step": 5810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.85750501127861e-05,
+      "loss": 0.8034,
+      "step": 5811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.855863986106254e-05,
+      "loss": 0.8004,
+      "step": 5812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8542232438910732e-05,
+      "loss": 0.6837,
+      "step": 5813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8525827848495913e-05,
+      "loss": 0.6521,
+      "step": 5814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8509426091982977e-05,
+      "loss": 0.7135,
+      "step": 5815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8493027171536403e-05,
+      "loss": 0.7674,
+      "step": 5816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8476631089320372e-05,
+      "loss": 0.7707,
+      "step": 5817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8460237847498606e-05,
+      "loss": 0.8198,
+      "step": 5818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8443847448234496e-05,
+      "loss": 0.8991,
+      "step": 5819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8427459893691055e-05,
+      "loss": 0.7107,
+      "step": 5820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.84110751860309e-05,
+      "loss": 0.7609,
+      "step": 5821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8394693327416295e-05,
+      "loss": 0.7649,
+      "step": 5822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8378314320009112e-05,
+      "loss": 0.7886,
+      "step": 5823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8361938165970876e-05,
+      "loss": 0.6908,
+      "step": 5824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.834556486746267e-05,
+      "loss": 0.7776,
+      "step": 5825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.83291944266453e-05,
+      "loss": 0.7851,
+      "step": 5826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.831282684567911e-05,
+      "loss": 0.8011,
+      "step": 5827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8296462126724076e-05,
+      "loss": 0.7483,
+      "step": 5828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8280100271939857e-05,
+      "loss": 0.7208,
+      "step": 5829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.826374128348568e-05,
+      "loss": 0.8406,
+      "step": 5830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8247385163520397e-05,
+      "loss": 0.7816,
+      "step": 5831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.82310319142025e-05,
+      "loss": 0.7477,
+      "step": 5832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8214681537690078e-05,
+      "loss": 0.8052,
+      "step": 5833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8198334036140874e-05,
+      "loss": 0.6923,
+      "step": 5834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8181989411712228e-05,
+      "loss": 0.7507,
+      "step": 5835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.816564766656108e-05,
+      "loss": 0.7094,
+      "step": 5836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8149308802844065e-05,
+      "loss": 0.7145,
+      "step": 5837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8132972822717362e-05,
+      "loss": 0.7263,
+      "step": 5838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8116639728336802e-05,
+      "loss": 0.8215,
+      "step": 5839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8100309521857826e-05,
+      "loss": 0.7952,
+      "step": 5840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.80839822054355e-05,
+      "loss": 0.7103,
+      "step": 5841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.80676577812245e-05,
+      "loss": 0.7725,
+      "step": 5842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8051336251379133e-05,
+      "loss": 0.7443,
+      "step": 5843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8035017618053296e-05,
+      "loss": 0.769,
+      "step": 5844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8018701883400557e-05,
+      "loss": 0.7826,
+      "step": 5845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.800238904957404e-05,
+      "loss": 0.7809,
+      "step": 5846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7986079118726554e-05,
+      "loss": 0.7485,
+      "step": 5847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7969772093010467e-05,
+      "loss": 0.7943,
+      "step": 5848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7953467974577775e-05,
+      "loss": 0.6805,
+      "step": 5849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7937166765580115e-05,
+      "loss": 0.8889,
+      "step": 5850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7920868468168705e-05,
+      "loss": 0.7246,
+      "step": 5851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7904573084494406e-05,
+      "loss": 0.7432,
+      "step": 5852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7888280616707684e-05,
+      "loss": 0.7459,
+      "step": 5853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7871991066958634e-05,
+      "loss": 0.6263,
+      "step": 5854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7855704437396914e-05,
+      "loss": 0.8219,
+      "step": 5855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7839420730171895e-05,
+      "loss": 0.7254,
+      "step": 5856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.782313994743247e-05,
+      "loss": 0.8382,
+      "step": 5857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.780686209132719e-05,
+      "loss": 0.8133,
+      "step": 5858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.77905871640042e-05,
+      "loss": 0.793,
+      "step": 5859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7774315167611276e-05,
+      "loss": 0.7321,
+      "step": 5860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7758046104295797e-05,
+      "loss": 0.7094,
+      "step": 5861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7741779976204745e-05,
+      "loss": 0.7035,
+      "step": 5862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7725516785484756e-05,
+      "loss": 0.7402,
+      "step": 5863
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7709256534282036e-05,
+      "loss": 0.8342,
+      "step": 5864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.769299922474239e-05,
+      "loss": 0.8279,
+      "step": 5865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7676744859011307e-05,
+      "loss": 0.7453,
+      "step": 5866
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.766049343923382e-05,
+      "loss": 0.7992,
+      "step": 5867
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7644244967554595e-05,
+      "loss": 0.6988,
+      "step": 5868
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7627999446117912e-05,
+      "loss": 0.8078,
+      "step": 5869
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7611756877067653e-05,
+      "loss": 0.7474,
+      "step": 5870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7595517262547322e-05,
+      "loss": 0.8029,
+      "step": 5871
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7579280604700023e-05,
+      "loss": 0.7198,
+      "step": 5872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7563046905668477e-05,
+      "loss": 0.7847,
+      "step": 5873
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7546816167595012e-05,
+      "loss": 0.6755,
+      "step": 5874
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7530588392621544e-05,
+      "loss": 0.7392,
+      "step": 5875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.751436358288965e-05,
+      "loss": 0.8729,
+      "step": 5876
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7498141740540478e-05,
+      "loss": 0.6917,
+      "step": 5877
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7481922867714783e-05,
+      "loss": 0.7667,
+      "step": 5878
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.746570696655294e-05,
+      "loss": 0.7772,
+      "step": 5879
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.74494940391949e-05,
+      "loss": 0.652,
+      "step": 5880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7433284087780305e-05,
+      "loss": 0.7366,
+      "step": 5881
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.741707711444831e-05,
+      "loss": 0.7762,
+      "step": 5882
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7400873121337727e-05,
+      "loss": 0.8132,
+      "step": 5883
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.738467211058696e-05,
+      "loss": 0.786,
+      "step": 5884
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7368474084334005e-05,
+      "loss": 0.7863,
+      "step": 5885
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.735227904471652e-05,
+      "loss": 0.7958,
+      "step": 5886
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7336086993871712e-05,
+      "loss": 0.7696,
+      "step": 5887
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7319897933936404e-05,
+      "loss": 0.792,
+      "step": 5888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7303711867047042e-05,
+      "loss": 0.7057,
+      "step": 5889
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7287528795339667e-05,
+      "loss": 0.7426,
+      "step": 5890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.727134872094993e-05,
+      "loss": 0.7467,
+      "step": 5891
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7255171646013066e-05,
+      "loss": 0.741,
+      "step": 5892
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.723899757266395e-05,
+      "loss": 0.7853,
+      "step": 5893
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.722282650303702e-05,
+      "loss": 0.8003,
+      "step": 5894
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.720665843926633e-05,
+      "loss": 0.7758,
+      "step": 5895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.719049338348559e-05,
+      "loss": 0.6958,
+      "step": 5896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7174331337828042e-05,
+      "loss": 0.7464,
+      "step": 5897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7158172304426555e-05,
+      "loss": 0.7012,
+      "step": 5898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7142016285413585e-05,
+      "loss": 0.8103,
+      "step": 5899
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.712586328292126e-05,
+      "loss": 0.6813,
+      "step": 5900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.710971329908122e-05,
+      "loss": 0.7144,
+      "step": 5901
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7093566336024754e-05,
+      "loss": 0.7709,
+      "step": 5902
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7077422395882745e-05,
+      "loss": 0.7338,
+      "step": 5903
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7061281480785655e-05,
+      "loss": 0.8413,
+      "step": 5904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7045143592863602e-05,
+      "loss": 0.7206,
+      "step": 5905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.702900873424625e-05,
+      "loss": 0.8636,
+      "step": 5906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.701287690706289e-05,
+      "loss": 0.7865,
+      "step": 5907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6996748113442394e-05,
+      "loss": 0.7569,
+      "step": 5908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6980622355513264e-05,
+      "loss": 0.7517,
+      "step": 5909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.696449963540356e-05,
+      "loss": 0.7404,
+      "step": 5910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6948379955240976e-05,
+      "loss": 0.7875,
+      "step": 5911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6932263317152797e-05,
+      "loss": 0.7479,
+      "step": 5912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6916149723265888e-05,
+      "loss": 0.9236,
+      "step": 5913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.690003917570672e-05,
+      "loss": 0.6944,
+      "step": 5914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6883931676601404e-05,
+      "loss": 0.6426,
+      "step": 5915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6867827228075592e-05,
+      "loss": 0.807,
+      "step": 5916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6851725832254537e-05,
+      "loss": 0.7845,
+      "step": 5917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6835627491263143e-05,
+      "loss": 0.82,
+      "step": 5918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6819532207225855e-05,
+      "loss": 0.6711,
+      "step": 5919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6803439982266743e-05,
+      "loss": 0.803,
+      "step": 5920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.678735081850945e-05,
+      "loss": 0.7487,
+      "step": 5921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6771264718077237e-05,
+      "loss": 0.7002,
+      "step": 5922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6755181683092955e-05,
+      "loss": 0.7351,
+      "step": 5923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6739101715679034e-05,
+      "loss": 0.7466,
+      "step": 5924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6723024817957542e-05,
+      "loss": 0.8531,
+      "step": 5925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6706950992050094e-05,
+      "loss": 0.7696,
+      "step": 5926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.669088024007793e-05,
+      "loss": 0.7874,
+      "step": 5927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6674812564161867e-05,
+      "loss": 0.6931,
+      "step": 5928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.665874796642232e-05,
+      "loss": 0.7668,
+      "step": 5929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6642686448979304e-05,
+      "loss": 0.7198,
+      "step": 5930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6626628013952425e-05,
+      "loss": 0.7259,
+      "step": 5931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6610572663460886e-05,
+      "loss": 0.734,
+      "step": 5932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6594520399623445e-05,
+      "loss": 0.794,
+      "step": 5933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6578471224558545e-05,
+      "loss": 0.7946,
+      "step": 5934
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6562425140384117e-05,
+      "loss": 0.7635,
+      "step": 5935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.654638214921773e-05,
+      "loss": 0.7286,
+      "step": 5936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6530342253176576e-05,
+      "loss": 0.7269,
+      "step": 5937
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6514305454377396e-05,
+      "loss": 0.7022,
+      "step": 5938
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6498271754936523e-05,
+      "loss": 0.7623,
+      "step": 5939
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6482241156969883e-05,
+      "loss": 0.6611,
+      "step": 5940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.646621366259302e-05,
+      "loss": 0.747,
+      "step": 5941
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6450189273921032e-05,
+      "loss": 0.7515,
+      "step": 5942
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6434167993068642e-05,
+      "loss": 0.7583,
+      "step": 5943
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6418149822150112e-05,
+      "loss": 0.7591,
+      "step": 5944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6402134763279375e-05,
+      "loss": 0.7957,
+      "step": 5945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.638612281856987e-05,
+      "loss": 0.7875,
+      "step": 5946
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6370113990134677e-05,
+      "loss": 0.715,
+      "step": 5947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6354108280086444e-05,
+      "loss": 0.8064,
+      "step": 5948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.63381056905374e-05,
+      "loss": 0.7042,
+      "step": 5949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6322106223599384e-05,
+      "loss": 0.7861,
+      "step": 5950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6306109881383816e-05,
+      "loss": 0.8088,
+      "step": 5951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6290116666001676e-05,
+      "loss": 0.7216,
+      "step": 5952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6274126579563586e-05,
+      "loss": 0.7364,
+      "step": 5953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6258139624179702e-05,
+      "loss": 0.8312,
+      "step": 5954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.624215580195981e-05,
+      "loss": 0.6868,
+      "step": 5955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6226175115013252e-05,
+      "loss": 0.7454,
+      "step": 5956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6210197565448967e-05,
+      "loss": 0.7553,
+      "step": 5957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6194223155375486e-05,
+      "loss": 0.8272,
+      "step": 5958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.61782518869009e-05,
+      "loss": 0.8923,
+      "step": 5959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6162283762132923e-05,
+      "loss": 0.6831,
+      "step": 5960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6146318783178813e-05,
+      "loss": 0.8984,
+      "step": 5961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6130356952145463e-05,
+      "loss": 0.7774,
+      "step": 5962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6114398271139285e-05,
+      "loss": 0.7846,
+      "step": 5963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6098442742266348e-05,
+      "loss": 0.8371,
+      "step": 5964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6082490367632268e-05,
+      "loss": 0.7134,
+      "step": 5965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6066541149342233e-05,
+      "loss": 0.6862,
+      "step": 5966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.605059508950103e-05,
+      "loss": 0.7315,
+      "step": 5967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6034652190213038e-05,
+      "loss": 0.7754,
+      "step": 5968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6018712453582194e-05,
+      "loss": 0.8041,
+      "step": 5969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6002775881712017e-05,
+      "loss": 0.8437,
+      "step": 5970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5986842476705665e-05,
+      "loss": 0.8124,
+      "step": 5971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5970912240665813e-05,
+      "loss": 0.7737,
+      "step": 5972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.595498517569473e-05,
+      "loss": 0.7983,
+      "step": 5973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5939061283894307e-05,
+      "loss": 0.8188,
+      "step": 5974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5923140567365977e-05,
+      "loss": 0.7458,
+      "step": 5975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.590722302821075e-05,
+      "loss": 0.7551,
+      "step": 5976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5891308668529247e-05,
+      "loss": 0.8178,
+      "step": 5977
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.587539749042164e-05,
+      "loss": 0.6694,
+      "step": 5978
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5859489495987693e-05,
+      "loss": 0.7593,
+      "step": 5979
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.584358468732676e-05,
+      "loss": 0.7276,
+      "step": 5980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5827683066537765e-05,
+      "loss": 0.8518,
+      "step": 5981
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.581178463571921e-05,
+      "loss": 0.7406,
+      "step": 5982
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5795889396969158e-05,
+      "loss": 0.9344,
+      "step": 5983
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.577999735238531e-05,
+      "loss": 0.7747,
+      "step": 5984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5764108504064883e-05,
+      "loss": 0.7908,
+      "step": 5985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.57482228541047e-05,
+      "loss": 0.7641,
+      "step": 5986
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.573234040460115e-05,
+      "loss": 0.845,
+      "step": 5987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.571646115765022e-05,
+      "loss": 0.8114,
+      "step": 5988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5700585115347424e-05,
+      "loss": 0.7148,
+      "step": 5989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5684712279787947e-05,
+      "loss": 0.7704,
+      "step": 5990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5668842653066467e-05,
+      "loss": 0.7417,
+      "step": 5991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.565297623727726e-05,
+      "loss": 0.8778,
+      "step": 5992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.563711303451417e-05,
+      "loss": 0.7695,
+      "step": 5993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5621253046870676e-05,
+      "loss": 0.672,
+      "step": 5994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5605396276439763e-05,
+      "loss": 0.7501,
+      "step": 5995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.558954272531401e-05,
+      "loss": 0.7302,
+      "step": 5996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.557369239558558e-05,
+      "loss": 0.733,
+      "step": 5997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5557845289346215e-05,
+      "loss": 0.7654,
+      "step": 5998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.554200140868722e-05,
+      "loss": 0.7713,
+      "step": 5999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5526160755699475e-05,
+      "loss": 0.7655,
+      "step": 6000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5510323332473447e-05,
+      "loss": 0.7775,
+      "step": 6001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5494489141099153e-05,
+      "loss": 0.8158,
+      "step": 6002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5478658183666205e-05,
+      "loss": 0.7049,
+      "step": 6003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5462830462263787e-05,
+      "loss": 0.7537,
+      "step": 6004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.544700597898066e-05,
+      "loss": 0.7104,
+      "step": 6005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5431184735905133e-05,
+      "loss": 0.7631,
+      "step": 6006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.541536673512509e-05,
+      "loss": 0.7497,
+      "step": 6007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.539955197872803e-05,
+      "loss": 0.6714,
+      "step": 6008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5383740468800978e-05,
+      "loss": 0.7513,
+      "step": 6009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.536793220743055e-05,
+      "loss": 0.799,
+      "step": 6010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5352127196702925e-05,
+      "loss": 0.7763,
+      "step": 6011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.533632543870386e-05,
+      "loss": 0.7899,
+      "step": 6012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.532052693551866e-05,
+      "loss": 0.7627,
+      "step": 6013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5304731689232264e-05,
+      "loss": 0.7455,
+      "step": 6014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.528893970192911e-05,
+      "loss": 0.7616,
+      "step": 6015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5273150975693237e-05,
+      "loss": 0.6829,
+      "step": 6016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5257365512608257e-05,
+      "loss": 0.7549,
+      "step": 6017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5241583314757327e-05,
+      "loss": 0.7059,
+      "step": 6018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5225804384223207e-05,
+      "loss": 0.6803,
+      "step": 6019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.521002872308821e-05,
+      "loss": 0.7492,
+      "step": 6020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5194256333434208e-05,
+      "loss": 0.7213,
+      "step": 6021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5178487217342638e-05,
+      "loss": 0.794,
+      "step": 6022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5162721376894553e-05,
+      "loss": 0.7728,
+      "step": 6023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.514695881417052e-05,
+      "loss": 0.7807,
+      "step": 6024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5131199531250694e-05,
+      "loss": 0.757,
+      "step": 6025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5115443530214765e-05,
+      "loss": 0.6897,
+      "step": 6026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5099690813142074e-05,
+      "loss": 0.7344,
+      "step": 6027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.508394138211144e-05,
+      "loss": 0.6477,
+      "step": 6028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5068195239201296e-05,
+      "loss": 0.7363,
+      "step": 6029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.505245238648962e-05,
+      "loss": 0.8144,
+      "step": 6030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5036712826053964e-05,
+      "loss": 0.827,
+      "step": 6031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5020976559971432e-05,
+      "loss": 0.6419,
+      "step": 6032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.500524359031873e-05,
+      "loss": 0.7449,
+      "step": 6033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.49895139191721e-05,
+      "loss": 0.7684,
+      "step": 6034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.497378754860736e-05,
+      "loss": 0.6123,
+      "step": 6035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4958064480699867e-05,
+      "loss": 0.7783,
+      "step": 6036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4942344717524568e-05,
+      "loss": 0.7505,
+      "step": 6037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4926628261155978e-05,
+      "loss": 0.7052,
+      "step": 6038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4910915113668156e-05,
+      "loss": 0.7462,
+      "step": 6039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4895205277134732e-05,
+      "loss": 0.6963,
+      "step": 6040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4879498753628882e-05,
+      "loss": 0.8141,
+      "step": 6041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4863795545223406e-05,
+      "loss": 0.6623,
+      "step": 6042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4848095653990605e-05,
+      "loss": 0.8376,
+      "step": 6043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4832399082002337e-05,
+      "loss": 0.7894,
+      "step": 6044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.481670583133009e-05,
+      "loss": 0.7592,
+      "step": 6045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4801015904044844e-05,
+      "loss": 0.7602,
+      "step": 6046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.478532930221717e-05,
+      "loss": 0.7294,
+      "step": 6047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.47696460279172e-05,
+      "loss": 0.7128,
+      "step": 6048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4753966083214615e-05,
+      "loss": 0.7273,
+      "step": 6049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4738289470178665e-05,
+      "loss": 0.7389,
+      "step": 6050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4722616190878166e-05,
+      "loss": 0.7782,
+      "step": 6051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.470694624738147e-05,
+      "loss": 0.7257,
+      "step": 6052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4691279641756536e-05,
+      "loss": 0.7306,
+      "step": 6053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4675616376070843e-05,
+      "loss": 0.6384,
+      "step": 6054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4659956452391434e-05,
+      "loss": 0.6926,
+      "step": 6055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4644299872784925e-05,
+      "loss": 0.7566,
+      "step": 6056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4628646639317472e-05,
+      "loss": 0.7727,
+      "step": 6057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4612996754054807e-05,
+      "loss": 0.6985,
+      "step": 6058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4597350219062205e-05,
+      "loss": 0.7945,
+      "step": 6059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4581707036404495e-05,
+      "loss": 0.7971,
+      "step": 6060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.456606720814611e-05,
+      "loss": 0.6512,
+      "step": 6061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4550430736350972e-05,
+      "loss": 0.7234,
+      "step": 6062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4534797623082623e-05,
+      "loss": 0.7665,
+      "step": 6063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4519167870404125e-05,
+      "loss": 0.8281,
+      "step": 6064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4503541480378096e-05,
+      "loss": 0.8802,
+      "step": 6065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4487918455066715e-05,
+      "loss": 0.7711,
+      "step": 6066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.447229879653173e-05,
+      "loss": 0.7803,
+      "step": 6067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.445668250683443e-05,
+      "loss": 0.6738,
+      "step": 6068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.444106958803566e-05,
+      "loss": 0.777,
+      "step": 6069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.442546004219583e-05,
+      "loss": 0.7504,
+      "step": 6070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.44098538713749e-05,
+      "loss": 0.7899,
+      "step": 6071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4394251077632362e-05,
+      "loss": 0.7052,
+      "step": 6072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4378651663027325e-05,
+      "loss": 0.7812,
+      "step": 6073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4363055629618388e-05,
+      "loss": 0.6825,
+      "step": 6074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4347462979463733e-05,
+      "loss": 0.791,
+      "step": 6075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4331873714621084e-05,
+      "loss": 0.6931,
+      "step": 6076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.431628783714772e-05,
+      "loss": 0.7396,
+      "step": 6077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.430070534910048e-05,
+      "loss": 0.7567,
+      "step": 6078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.428512625253576e-05,
+      "loss": 0.8035,
+      "step": 6079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.42695505495095e-05,
+      "loss": 0.778,
+      "step": 6080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4253978242077173e-05,
+      "loss": 0.8021,
+      "step": 6081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4238409332293864e-05,
+      "loss": 0.6551,
+      "step": 6082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4222843822214142e-05,
+      "loss": 0.8063,
+      "step": 6083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.420728171389216e-05,
+      "loss": 0.732,
+      "step": 6084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.419172300938162e-05,
+      "loss": 0.8574,
+      "step": 6085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4176167710735765e-05,
+      "loss": 0.7222,
+      "step": 6086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4160615820007397e-05,
+      "loss": 0.7711,
+      "step": 6087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4145067339248872e-05,
+      "loss": 0.8025,
+      "step": 6088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4129522270512084e-05,
+      "loss": 0.7852,
+      "step": 6089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4113980615848493e-05,
+      "loss": 0.7151,
+      "step": 6090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4098442377309066e-05,
+      "loss": 0.8076,
+      "step": 6091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4082907556944405e-05,
+      "loss": 0.764,
+      "step": 6092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4067376156804576e-05,
+      "loss": 0.7161,
+      "step": 6093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.405184817893924e-05,
+      "loss": 0.662,
+      "step": 6094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4036323625397576e-05,
+      "loss": 0.7225,
+      "step": 6095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4020802498228335e-05,
+      "loss": 0.8088,
+      "step": 6096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4005284799479787e-05,
+      "loss": 0.7313,
+      "step": 6097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3989770531199816e-05,
+      "loss": 0.6838,
+      "step": 6098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.397425969543578e-05,
+      "loss": 0.7502,
+      "step": 6099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.395875229423461e-05,
+      "loss": 0.7819,
+      "step": 6100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3943248329642776e-05,
+      "loss": 0.714,
+      "step": 6101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3927747803706334e-05,
+      "loss": 0.8476,
+      "step": 6102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.391225071847084e-05,
+      "loss": 0.7457,
+      "step": 6103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3896757075981414e-05,
+      "loss": 0.7614,
+      "step": 6104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3881266878282715e-05,
+      "loss": 0.7805,
+      "step": 6105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3865780127418956e-05,
+      "loss": 0.6828,
+      "step": 6106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3850296825433892e-05,
+      "loss": 0.7547,
+      "step": 6107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.383481697437082e-05,
+      "loss": 0.7038,
+      "step": 6108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3819340576272588e-05,
+      "loss": 0.7639,
+      "step": 6109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3803867633181574e-05,
+      "loss": 0.7017,
+      "step": 6110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3788398147139708e-05,
+      "loss": 0.6911,
+      "step": 6111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3772932120188485e-05,
+      "loss": 0.7661,
+      "step": 6112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3757469554368918e-05,
+      "loss": 0.7871,
+      "step": 6113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3742010451721557e-05,
+      "loss": 0.7463,
+      "step": 6114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3726554814286506e-05,
+      "loss": 0.7152,
+      "step": 6115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3711102644103446e-05,
+      "loss": 0.716,
+      "step": 6116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.369565394321154e-05,
+      "loss": 0.7567,
+      "step": 6117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.368020871364952e-05,
+      "loss": 0.7757,
+      "step": 6118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3664766957455674e-05,
+      "loss": 0.8014,
+      "step": 6119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3649328676667815e-05,
+      "loss": 0.7221,
+      "step": 6120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3633893873323276e-05,
+      "loss": 0.7145,
+      "step": 6121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3618462549458998e-05,
+      "loss": 0.7113,
+      "step": 6122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.36030347071114e-05,
+      "loss": 0.7798,
+      "step": 6123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3587610348316457e-05,
+      "loss": 0.7709,
+      "step": 6124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3572189475109702e-05,
+      "loss": 0.6925,
+      "step": 6125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3556772089526186e-05,
+      "loss": 0.7726,
+      "step": 6126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3541358193600517e-05,
+      "loss": 0.7925,
+      "step": 6127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.352594778936682e-05,
+      "loss": 0.7744,
+      "step": 6128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3510540878858795e-05,
+      "loss": 0.8381,
+      "step": 6129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3495137464109622e-05,
+      "loss": 0.8324,
+      "step": 6130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.347973754715211e-05,
+      "loss": 0.7197,
+      "step": 6131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3464341130018524e-05,
+      "loss": 0.8382,
+      "step": 6132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.34489482147407e-05,
+      "loss": 0.7798,
+      "step": 6133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.343355880335e-05,
+      "loss": 0.775,
+      "step": 6134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3418172897877356e-05,
+      "loss": 0.8135,
+      "step": 6135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.34027905003532e-05,
+      "loss": 0.6494,
+      "step": 6136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.338741161280752e-05,
+      "loss": 0.7075,
+      "step": 6137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.337203623726983e-05,
+      "loss": 0.7763,
+      "step": 6138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3356664375769194e-05,
+      "loss": 0.8063,
+      "step": 6139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3341296030334176e-05,
+      "loss": 0.6919,
+      "step": 6140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3325931202992945e-05,
+      "loss": 0.7952,
+      "step": 6141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3310569895773153e-05,
+      "loss": 0.8391,
+      "step": 6142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.329521211070199e-05,
+      "loss": 0.8186,
+      "step": 6143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3279857849806204e-05,
+      "loss": 0.76,
+      "step": 6144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.326450711511205e-05,
+      "loss": 0.7152,
+      "step": 6145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3249159908645347e-05,
+      "loss": 0.7595,
+      "step": 6146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3233816232431416e-05,
+      "loss": 0.687,
+      "step": 6147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.321847608849515e-05,
+      "loss": 0.7034,
+      "step": 6148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3203139478860924e-05,
+      "loss": 0.7088,
+      "step": 6149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.318780640555272e-05,
+      "loss": 0.6709,
+      "step": 6150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.317247687059399e-05,
+      "loss": 0.8517,
+      "step": 6151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3157150876007728e-05,
+      "loss": 0.7254,
+      "step": 6152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.314182842381651e-05,
+      "loss": 0.7905,
+      "step": 6153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3126509516042383e-05,
+      "loss": 0.7368,
+      "step": 6154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.311119415470696e-05,
+      "loss": 0.8163,
+      "step": 6155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3095882341831372e-05,
+      "loss": 0.7882,
+      "step": 6156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3080574079436284e-05,
+      "loss": 0.7809,
+      "step": 6157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3065269369541907e-05,
+      "loss": 0.8029,
+      "step": 6158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.304996821416796e-05,
+      "loss": 0.7185,
+      "step": 6159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.303467061533369e-05,
+      "loss": 0.6981,
+      "step": 6160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.301937657505793e-05,
+      "loss": 0.7681,
+      "step": 6161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.300408609535899e-05,
+      "loss": 0.8288,
+      "step": 6162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2988799178254706e-05,
+      "loss": 0.7193,
+      "step": 6163
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2973515825762467e-05,
+      "loss": 0.8138,
+      "step": 6164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.295823603989919e-05,
+      "loss": 0.7642,
+      "step": 6165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2942959822681314e-05,
+      "loss": 0.737,
+      "step": 6166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2927687176124813e-05,
+      "loss": 0.7975,
+      "step": 6167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2912418102245158e-05,
+      "loss": 0.7675,
+      "step": 6168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2897152603057426e-05,
+      "loss": 0.7356,
+      "step": 6169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2881890680576145e-05,
+      "loss": 0.7941,
+      "step": 6170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2866632336815387e-05,
+      "loss": 0.8041,
+      "step": 6171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2851377573788795e-05,
+      "loss": 0.8544,
+      "step": 6172
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.28361263935095e-05,
+      "loss": 0.807,
+      "step": 6173
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.282087879799016e-05,
+      "loss": 0.7623,
+      "step": 6174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.280563478924297e-05,
+      "loss": 0.6822,
+      "step": 6175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2790394369279654e-05,
+      "loss": 0.8052,
+      "step": 6176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2775157540111454e-05,
+      "loss": 0.7983,
+      "step": 6177
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2759924303749143e-05,
+      "loss": 0.7444,
+      "step": 6178
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.274469466220302e-05,
+      "loss": 0.8534,
+      "step": 6179
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.272946861748289e-05,
+      "loss": 0.7484,
+      "step": 6180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.271424617159814e-05,
+      "loss": 0.7443,
+      "step": 6181
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.269902732655763e-05,
+      "loss": 0.8623,
+      "step": 6182
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.268381208436976e-05,
+      "loss": 0.7388,
+      "step": 6183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.266860044704245e-05,
+      "loss": 0.712,
+      "step": 6184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.265339241658314e-05,
+      "loss": 0.7499,
+      "step": 6185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.26381879949988e-05,
+      "loss": 0.7734,
+      "step": 6186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2622987184295953e-05,
+      "loss": 0.6729,
+      "step": 6187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.26077899864806e-05,
+      "loss": 0.6377,
+      "step": 6188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2592596403558263e-05,
+      "loss": 0.6767,
+      "step": 6189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2577406437534054e-05,
+      "loss": 0.7401,
+      "step": 6190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2562220090412538e-05,
+      "loss": 0.7844,
+      "step": 6191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2547037364197825e-05,
+      "loss": 0.7736,
+      "step": 6192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2531858260893546e-05,
+      "loss": 0.7332,
+      "step": 6193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.251668278250286e-05,
+      "loss": 0.8385,
+      "step": 6194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2501510931028436e-05,
+      "loss": 0.8066,
+      "step": 6195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.248634270847248e-05,
+      "loss": 0.7491,
+      "step": 6196
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2471178116836705e-05,
+      "loss": 0.7354,
+      "step": 6197
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2456017158122355e-05,
+      "loss": 0.7609,
+      "step": 6198
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.244085983433017e-05,
+      "loss": 0.7776,
+      "step": 6199
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2425706147460468e-05,
+      "loss": 0.7461,
+      "step": 6200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.241055609951303e-05,
+      "loss": 0.8437,
+      "step": 6201
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2395409692487175e-05,
+      "loss": 0.7173,
+      "step": 6202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2380266928381744e-05,
+      "loss": 0.6183,
+      "step": 6203
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2365127809195095e-05,
+      "loss": 0.7394,
+      "step": 6204
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2349992336925098e-05,
+      "loss": 0.7485,
+      "step": 6205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2334860513569174e-05,
+      "loss": 0.7986,
+      "step": 6206
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2319732341124224e-05,
+      "loss": 0.7238,
+      "step": 6207
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2304607821586677e-05,
+      "loss": 0.8463,
+      "step": 6208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2289486956952473e-05,
+      "loss": 0.7935,
+      "step": 6209
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2274369749217118e-05,
+      "loss": 0.8247,
+      "step": 6210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2259256200375573e-05,
+      "loss": 0.7598,
+      "step": 6211
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2244146312422348e-05,
+      "loss": 0.7851,
+      "step": 6212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.222904008735146e-05,
+      "loss": 0.7896,
+      "step": 6213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.221393752715645e-05,
+      "loss": 0.7536,
+      "step": 6214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2198838633830366e-05,
+      "loss": 0.7535,
+      "step": 6215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2183743409365785e-05,
+      "loss": 0.8605,
+      "step": 6216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2168651855754784e-05,
+      "loss": 0.7728,
+      "step": 6217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2153563974988973e-05,
+      "loss": 0.6985,
+      "step": 6218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2138479769059445e-05,
+      "loss": 0.6502,
+      "step": 6219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2123399239956864e-05,
+      "loss": 0.6642,
+      "step": 6220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.210832238967136e-05,
+      "loss": 0.752,
+      "step": 6221
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2093249220192602e-05,
+      "loss": 0.7241,
+      "step": 6222
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2078179733509737e-05,
+      "loss": 0.6883,
+      "step": 6223
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.206311393161149e-05,
+      "loss": 0.7767,
+      "step": 6224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2048051816486055e-05,
+      "loss": 0.7075,
+      "step": 6225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2032993390121137e-05,
+      "loss": 0.8305,
+      "step": 6226
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2017938654503973e-05,
+      "loss": 0.7041,
+      "step": 6227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.20028876116213e-05,
+      "loss": 0.8393,
+      "step": 6228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.198784026345936e-05,
+      "loss": 0.7687,
+      "step": 6229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1972796612003947e-05,
+      "loss": 0.7449,
+      "step": 6230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.195775665924033e-05,
+      "loss": 0.7453,
+      "step": 6231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1942720407153305e-05,
+      "loss": 0.8343,
+      "step": 6232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1927687857727163e-05,
+      "loss": 0.8355,
+      "step": 6233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.191265901294573e-05,
+      "loss": 0.7627,
+      "step": 6234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1897633874792323e-05,
+      "loss": 0.6383,
+      "step": 6235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.188261244524978e-05,
+      "loss": 0.7421,
+      "step": 6236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.186759472630045e-05,
+      "loss": 0.8252,
+      "step": 6237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1852580719926187e-05,
+      "loss": 0.7762,
+      "step": 6238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1837570428108338e-05,
+      "loss": 0.6503,
+      "step": 6239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1822563852827827e-05,
+      "loss": 0.6703,
+      "step": 6240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1807560996065018e-05,
+      "loss": 0.6915,
+      "step": 6241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1792561859799782e-05,
+      "loss": 0.7707,
+      "step": 6242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1777566446011565e-05,
+      "loss": 0.7808,
+      "step": 6243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.176257475667926e-05,
+      "loss": 0.7592,
+      "step": 6244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1747586793781294e-05,
+      "loss": 0.7416,
+      "step": 6245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1732602559295596e-05,
+      "loss": 0.8034,
+      "step": 6246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.17176220551996e-05,
+      "loss": 0.8141,
+      "step": 6247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1702645283470236e-05,
+      "loss": 0.7236,
+      "step": 6248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1687672246083995e-05,
+      "loss": 0.7039,
+      "step": 6249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.167270294501682e-05,
+      "loss": 0.7189,
+      "step": 6250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1657737382244176e-05,
+      "loss": 0.7258,
+      "step": 6251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1642775559741043e-05,
+      "loss": 0.6706,
+      "step": 6252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1627817479481888e-05,
+      "loss": 0.8054,
+      "step": 6253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1612863143440716e-05,
+      "loss": 0.7049,
+      "step": 6254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1597912553591005e-05,
+      "loss": 0.774,
+      "step": 6255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.158296571190576e-05,
+      "loss": 0.7812,
+      "step": 6256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.156802262035747e-05,
+      "loss": 0.7783,
+      "step": 6257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1553083280918167e-05,
+      "loss": 0.8005,
+      "step": 6258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.153814769555936e-05,
+      "loss": 0.667,
+      "step": 6259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1523215866252045e-05,
+      "loss": 0.7396,
+      "step": 6260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.150828779496678e-05,
+      "loss": 0.7531,
+      "step": 6261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.149336348367357e-05,
+      "loss": 0.7683,
+      "step": 6262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1478442934341948e-05,
+      "loss": 0.7455,
+      "step": 6263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.146352614894095e-05,
+      "loss": 0.713,
+      "step": 6264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1448613129439115e-05,
+      "loss": 0.714,
+      "step": 6265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1433703877804474e-05,
+      "loss": 0.727,
+      "step": 6266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.141879839600458e-05,
+      "loss": 0.7357,
+      "step": 6267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1403896686006454e-05,
+      "loss": 0.7403,
+      "step": 6268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1388998749776678e-05,
+      "loss": 0.7896,
+      "step": 6269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1374104589281296e-05,
+      "loss": 0.7688,
+      "step": 6270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1359214206485845e-05,
+      "loss": 0.7555,
+      "step": 6271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1344327603355384e-05,
+      "loss": 0.77,
+      "step": 6272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.132944478185447e-05,
+      "loss": 0.7698,
+      "step": 6273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1314565743947152e-05,
+      "loss": 0.7207,
+      "step": 6274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1299690491596992e-05,
+      "loss": 0.8007,
+      "step": 6275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1284819026767016e-05,
+      "loss": 0.7633,
+      "step": 6276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1269951351419835e-05,
+      "loss": 0.8513,
+      "step": 6277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.125508746751747e-05,
+      "loss": 0.8565,
+      "step": 6278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.124022737702147e-05,
+      "loss": 0.7892,
+      "step": 6279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1225371081892927e-05,
+      "loss": 0.801,
+      "step": 6280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1210518584092364e-05,
+      "loss": 0.8618,
+      "step": 6281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.119566988557985e-05,
+      "loss": 0.8399,
+      "step": 6282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.118082498831493e-05,
+      "loss": 0.8664,
+      "step": 6283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1165983894256647e-05,
+      "loss": 0.774,
+      "step": 6284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1151146605363565e-05,
+      "loss": 0.802,
+      "step": 6285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1136313123593717e-05,
+      "loss": 0.7534,
+      "step": 6286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.112148345090465e-05,
+      "loss": 0.6925,
+      "step": 6287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1106657589253393e-05,
+      "loss": 0.8085,
+      "step": 6288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1091835540596518e-05,
+      "loss": 0.7765,
+      "step": 6289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1077017306890033e-05,
+      "loss": 0.7147,
+      "step": 6290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1062202890089477e-05,
+      "loss": 0.7721,
+      "step": 6291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1047392292149882e-05,
+      "loss": 0.7791,
+      "step": 6292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.103258551502576e-05,
+      "loss": 0.7692,
+      "step": 6293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1017782560671123e-05,
+      "loss": 0.7759,
+      "step": 6294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1002983431039524e-05,
+      "loss": 0.7855,
+      "step": 6295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0988188128083942e-05,
+      "loss": 0.7634,
+      "step": 6296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0973396653756895e-05,
+      "loss": 0.8186,
+      "step": 6297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.095860901001036e-05,
+      "loss": 0.6968,
+      "step": 6298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0943825198795867e-05,
+      "loss": 0.719,
+      "step": 6299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0929045222064387e-05,
+      "loss": 0.707,
+      "step": 6300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0914269081766407e-05,
+      "loss": 0.7782,
+      "step": 6301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0899496779851902e-05,
+      "loss": 0.8065,
+      "step": 6302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0884728318270336e-05,
+      "loss": 0.7777,
+      "step": 6303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0869963698970674e-05,
+      "loss": 0.7299,
+      "step": 6304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.085520292390138e-05,
+      "loss": 0.8103,
+      "step": 6305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0840445995010394e-05,
+      "loss": 0.7792,
+      "step": 6306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0825692914245148e-05,
+      "loss": 0.8224,
+      "step": 6307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0810943683552602e-05,
+      "loss": 0.6774,
+      "step": 6308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0796198304879166e-05,
+      "loss": 0.6616,
+      "step": 6309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.078145678017076e-05,
+      "loss": 0.7036,
+      "step": 6310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0766719111372795e-05,
+      "loss": 0.7253,
+      "step": 6311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.075198530043016e-05,
+      "loss": 0.7782,
+      "step": 6312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.073725534928724e-05,
+      "loss": 0.7083,
+      "step": 6313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.072252925988794e-05,
+      "loss": 0.746,
+      "step": 6314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0707807034175626e-05,
+      "loss": 0.7878,
+      "step": 6315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0693088674093147e-05,
+      "loss": 0.7574,
+      "step": 6316
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0678374181582845e-05,
+      "loss": 0.8464,
+      "step": 6317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0663663558586593e-05,
+      "loss": 0.6683,
+      "step": 6318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0648956807045706e-05,
+      "loss": 0.716,
+      "step": 6319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0634253928900997e-05,
+      "loss": 0.7531,
+      "step": 6320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0619554926092782e-05,
+      "loss": 0.7639,
+      "step": 6321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0604859800560862e-05,
+      "loss": 0.8734,
+      "step": 6322
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0590168554244504e-05,
+      "loss": 0.7878,
+      "step": 6323
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0575481189082502e-05,
+      "loss": 0.8147,
+      "step": 6324
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0560797707013102e-05,
+      "loss": 0.8127,
+      "step": 6325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.054611810997406e-05,
+      "loss": 0.7854,
+      "step": 6326
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.053144239990259e-05,
+      "loss": 0.6563,
+      "step": 6327
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0516770578735455e-05,
+      "loss": 0.7079,
+      "step": 6328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0502102648408844e-05,
+      "loss": 0.7664,
+      "step": 6329
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0487438610858458e-05,
+      "loss": 0.7064,
+      "step": 6330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0472778468019454e-05,
+      "loss": 0.8164,
+      "step": 6331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.045812222182654e-05,
+      "loss": 0.7634,
+      "step": 6332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0443469874213855e-05,
+      "loss": 0.9165,
+      "step": 6333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0428821427115036e-05,
+      "loss": 0.7055,
+      "step": 6334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.041417688246321e-05,
+      "loss": 0.7473,
+      "step": 6335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.039953624219098e-05,
+      "loss": 0.7782,
+      "step": 6336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0384899508230437e-05,
+      "loss": 0.6988,
+      "step": 6337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0370266682513183e-05,
+      "loss": 0.7164,
+      "step": 6338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.035563776697027e-05,
+      "loss": 0.7553,
+      "step": 6339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0341012763532243e-05,
+      "loss": 0.6669,
+      "step": 6340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0326391674129136e-05,
+      "loss": 0.7257,
+      "step": 6341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0311774500690462e-05,
+      "loss": 0.7292,
+      "step": 6342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.029716124514522e-05,
+      "loss": 0.6854,
+      "step": 6343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0282551909421888e-05,
+      "loss": 0.6476,
+      "step": 6344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.026794649544843e-05,
+      "loss": 0.7508,
+      "step": 6345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0253345005152296e-05,
+      "loss": 0.7937,
+      "step": 6346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0238747440460397e-05,
+      "loss": 0.7312,
+      "step": 6347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.022415380329918e-05,
+      "loss": 0.8064,
+      "step": 6348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0209564095594507e-05,
+      "loss": 0.7186,
+      "step": 6349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0194978319271745e-05,
+      "loss": 0.6887,
+      "step": 6350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0180396476255787e-05,
+      "loss": 0.7202,
+      "step": 6351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.016581856847094e-05,
+      "loss": 0.693,
+      "step": 6352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.015124459784103e-05,
+      "loss": 0.7588,
+      "step": 6353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0136674566289344e-05,
+      "loss": 0.7914,
+      "step": 6354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0122108475738667e-05,
+      "loss": 0.6977,
+      "step": 6355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0107546328111232e-05,
+      "loss": 0.7284,
+      "step": 6356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0092988125328817e-05,
+      "loss": 0.7014,
+      "step": 6357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0078433869312612e-05,
+      "loss": 0.6824,
+      "step": 6358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.006388356198332e-05,
+      "loss": 0.7635,
+      "step": 6359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.004933720526111e-05,
+      "loss": 0.6748,
+      "step": 6360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.003479480106564e-05,
+      "loss": 0.7538,
+      "step": 6361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.002025635131603e-05,
+      "loss": 0.6855,
+      "step": 6362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0005721857930905e-05,
+      "loss": 0.7746,
+      "step": 6363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9991191322828334e-05,
+      "loss": 0.797,
+      "step": 6364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9976664747925894e-05,
+      "loss": 0.714,
+      "step": 6365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9962142135140605e-05,
+      "loss": 0.8319,
+      "step": 6366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9947623486389024e-05,
+      "loss": 0.6671,
+      "step": 6367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9933108803587102e-05,
+      "loss": 0.7742,
+      "step": 6368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.991859808865036e-05,
+      "loss": 0.7669,
+      "step": 6369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9904091343493724e-05,
+      "loss": 0.753,
+      "step": 6370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.988958857003162e-05,
+      "loss": 0.7236,
+      "step": 6371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.987508977017794e-05,
+      "loss": 0.8115,
+      "step": 6372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9860594945846074e-05,
+      "loss": 0.7623,
+      "step": 6373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9846104098948865e-05,
+      "loss": 0.7541,
+      "step": 6374
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9831617231398645e-05,
+      "loss": 0.7739,
+      "step": 6375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9817134345107196e-05,
+      "loss": 0.7613,
+      "step": 6376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.980265544198583e-05,
+      "loss": 0.7837,
+      "step": 6377
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9788180523945277e-05,
+      "loss": 0.7181,
+      "step": 6378
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9773709592895766e-05,
+      "loss": 0.7962,
+      "step": 6379
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975924265074699e-05,
+      "loss": 0.8577,
+      "step": 6380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9744779699408132e-05,
+      "loss": 0.7765,
+      "step": 6381
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9730320740787827e-05,
+      "loss": 0.7821,
+      "step": 6382
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.97158657767942e-05,
+      "loss": 0.7655,
+      "step": 6383
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9701414809334823e-05,
+      "loss": 0.7214,
+      "step": 6384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9686967840316795e-05,
+      "loss": 0.7531,
+      "step": 6385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.967252487164663e-05,
+      "loss": 0.8084,
+      "step": 6386
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9658085905230334e-05,
+      "loss": 0.6822,
+      "step": 6387
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.96436509429734e-05,
+      "loss": 0.8161,
+      "step": 6388
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9629219986780783e-05,
+      "loss": 0.7553,
+      "step": 6389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.96147930385569e-05,
+      "loss": 0.8003,
+      "step": 6390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.960037010020564e-05,
+      "loss": 0.7881,
+      "step": 6391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9585951173630375e-05,
+      "loss": 0.6776,
+      "step": 6392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9571536260733935e-05,
+      "loss": 0.7016,
+      "step": 6393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9557125363418627e-05,
+      "loss": 0.6686,
+      "step": 6394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9542718483586232e-05,
+      "loss": 0.7608,
+      "step": 6395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9528315623137977e-05,
+      "loss": 0.7687,
+      "step": 6396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9513916783974607e-05,
+      "loss": 0.7508,
+      "step": 6397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.94995219679963e-05,
+      "loss": 0.6418,
+      "step": 6398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9485131177102696e-05,
+      "loss": 0.7764,
+      "step": 6399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9470744413192928e-05,
+      "loss": 0.7227,
+      "step": 6400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.945636167816558e-05,
+      "loss": 0.7657,
+      "step": 6401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9441982973918705e-05,
+      "loss": 0.7653,
+      "step": 6402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.942760830234986e-05,
+      "loss": 0.6803,
+      "step": 6403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9413237665356015e-05,
+      "loss": 0.694,
+      "step": 6404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9398871064833646e-05,
+      "loss": 0.7941,
+      "step": 6405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.938450850267866e-05,
+      "loss": 0.7874,
+      "step": 6406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.937014998078649e-05,
+      "loss": 0.7211,
+      "step": 6407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.935579550105198e-05,
+      "loss": 0.7697,
+      "step": 6408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9341445065369463e-05,
+      "loss": 0.7751,
+      "step": 6409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932709867563274e-05,
+      "loss": 0.7295,
+      "step": 6410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9312756333735073e-05,
+      "loss": 0.7688,
+      "step": 6411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9298418041569183e-05,
+      "loss": 0.7175,
+      "step": 6412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9284083801027275e-05,
+      "loss": 0.7574,
+      "step": 6413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9269753614001e-05,
+      "loss": 0.7242,
+      "step": 6414
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9255427482381478e-05,
+      "loss": 0.6867,
+      "step": 6415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9241105408059316e-05,
+      "loss": 0.8275,
+      "step": 6416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9226787392924572e-05,
+      "loss": 0.7361,
+      "step": 6417
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9212473438866746e-05,
+      "loss": 0.7082,
+      "step": 6418
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.919816354777483e-05,
+      "loss": 0.7935,
+      "step": 6419
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9183857721537268e-05,
+      "loss": 0.6663,
+      "step": 6420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9169555962041956e-05,
+      "loss": 0.8438,
+      "step": 6421
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9155258271176295e-05,
+      "loss": 0.7221,
+      "step": 6422
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9140964650827113e-05,
+      "loss": 0.7764,
+      "step": 6423
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.912667510288071e-05,
+      "loss": 0.7466,
+      "step": 6424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9112389629222823e-05,
+      "loss": 0.7462,
+      "step": 6425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.909810823173871e-05,
+      "loss": 0.7802,
+      "step": 6426
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9083830912313045e-05,
+      "loss": 0.762,
+      "step": 6427
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9069557672829973e-05,
+      "loss": 0.8177,
+      "step": 6428
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.905528851517311e-05,
+      "loss": 0.8389,
+      "step": 6429
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9041023441225526e-05,
+      "loss": 0.6953,
+      "step": 6430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9026762452869745e-05,
+      "loss": 0.711,
+      "step": 6431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9012505551987765e-05,
+      "loss": 0.7288,
+      "step": 6432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8998252740461043e-05,
+      "loss": 0.775,
+      "step": 6433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8984004020170488e-05,
+      "loss": 0.7944,
+      "step": 6434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.896975939299646e-05,
+      "loss": 0.836,
+      "step": 6435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8955518860818823e-05,
+      "loss": 0.802,
+      "step": 6436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.894128242551686e-05,
+      "loss": 0.7793,
+      "step": 6437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8927050088969316e-05,
+      "loss": 0.7876,
+      "step": 6438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.891282185305439e-05,
+      "loss": 0.7534,
+      "step": 6439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.889859771964979e-05,
+      "loss": 0.7661,
+      "step": 6440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8884377690632622e-05,
+      "loss": 0.7335,
+      "step": 6441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.887016176787948e-05,
+      "loss": 0.7158,
+      "step": 6442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.88559499532664e-05,
+      "loss": 0.7601,
+      "step": 6443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8841742248668898e-05,
+      "loss": 0.8083,
+      "step": 6444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8827538655961907e-05,
+      "loss": 0.7153,
+      "step": 6445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8813339177019882e-05,
+      "loss": 0.817,
+      "step": 6446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8799143813716697e-05,
+      "loss": 0.6292,
+      "step": 6447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8784952567925663e-05,
+      "loss": 0.8427,
+      "step": 6448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8770765441519577e-05,
+      "loss": 0.7045,
+      "step": 6449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.875658243637069e-05,
+      "loss": 0.7584,
+      "step": 6450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.87424035543507e-05,
+      "loss": 0.7608,
+      "step": 6451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.872822879733076e-05,
+      "loss": 0.8316,
+      "step": 6452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.871405816718149e-05,
+      "loss": 0.6717,
+      "step": 6453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8699891665772957e-05,
+      "loss": 0.7677,
+      "step": 6454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.868572929497467e-05,
+      "loss": 0.817,
+      "step": 6455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8671571056655628e-05,
+      "loss": 0.7667,
+      "step": 6456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.865741695268427e-05,
+      "loss": 0.7552,
+      "step": 6457
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.864326698492845e-05,
+      "loss": 0.7422,
+      "step": 6458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.862912115525554e-05,
+      "loss": 0.7346,
+      "step": 6459
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.861497946553234e-05,
+      "loss": 0.7933,
+      "step": 6460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.860084191762508e-05,
+      "loss": 0.6845,
+      "step": 6461
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.858670851339947e-05,
+      "loss": 0.7994,
+      "step": 6462
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.857257925472066e-05,
+      "loss": 0.6536,
+      "step": 6463
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8558454143453276e-05,
+      "loss": 0.779,
+      "step": 6464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8544333181461337e-05,
+      "loss": 0.8455,
+      "step": 6465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8530216370608415e-05,
+      "loss": 0.7231,
+      "step": 6466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8516103712757444e-05,
+      "loss": 0.8012,
+      "step": 6467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.850199520977085e-05,
+      "loss": 0.7225,
+      "step": 6468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.84878908635105e-05,
+      "loss": 0.7926,
+      "step": 6469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.847379067583771e-05,
+      "loss": 0.8583,
+      "step": 6470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8459694648613263e-05,
+      "loss": 0.8025,
+      "step": 6471
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8445602783697374e-05,
+      "loss": 0.7195,
+      "step": 6472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8431515082949725e-05,
+      "loss": 0.7709,
+      "step": 6473
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8417431548229413e-05,
+      "loss": 0.7277,
+      "step": 6474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8403352181395056e-05,
+      "loss": 0.8378,
+      "step": 6475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8389276984304643e-05,
+      "loss": 0.774,
+      "step": 6476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.837520595881568e-05,
+      "loss": 0.7751,
+      "step": 6477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.836113910678507e-05,
+      "loss": 0.6943,
+      "step": 6478
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8347076430069192e-05,
+      "loss": 0.7264,
+      "step": 6479
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8333017930523872e-05,
+      "loss": 0.6833,
+      "step": 6480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8318963610004374e-05,
+      "loss": 0.7511,
+      "step": 6481
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.830491347036542e-05,
+      "loss": 0.7389,
+      "step": 6482
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.829086751346118e-05,
+      "loss": 0.6907,
+      "step": 6483
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.827682574114525e-05,
+      "loss": 0.5937,
+      "step": 6484
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8262788155270722e-05,
+      "loss": 0.6813,
+      "step": 6485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8248754757690102e-05,
+      "loss": 0.7702,
+      "step": 6486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.823472555025534e-05,
+      "loss": 0.7929,
+      "step": 6487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8220700534817843e-05,
+      "loss": 0.7389,
+      "step": 6488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8206679713228463e-05,
+      "loss": 0.8013,
+      "step": 6489
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.81926630873375e-05,
+      "loss": 0.7747,
+      "step": 6490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8178650658994702e-05,
+      "loss": 0.8362,
+      "step": 6491
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8164642430049233e-05,
+      "loss": 0.7068,
+      "step": 6492
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8150638402349773e-05,
+      "loss": 0.7636,
+      "step": 6493
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8136638577744388e-05,
+      "loss": 0.6988,
+      "step": 6494
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8122642958080583e-05,
+      "loss": 0.7269,
+      "step": 6495
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.810865154520536e-05,
+      "loss": 0.7084,
+      "step": 6496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.809466434096514e-05,
+      "loss": 0.7268,
+      "step": 6497
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8080681347205764e-05,
+      "loss": 0.8243,
+      "step": 6498
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8066702565772553e-05,
+      "loss": 0.7979,
+      "step": 6499
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8052727998510243e-05,
+      "loss": 0.8199,
+      "step": 6500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8038757647263045e-05,
+      "loss": 0.7447,
+      "step": 6501
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.802479151387459e-05,
+      "loss": 0.7082,
+      "step": 6502
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.801082960018795e-05,
+      "loss": 0.7563,
+      "step": 6503
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.799687190804565e-05,
+      "loss": 0.7349,
+      "step": 6504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.798291843928968e-05,
+      "loss": 0.7634,
+      "step": 6505
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7968969195761436e-05,
+      "loss": 0.7048,
+      "step": 6506
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7955024179301776e-05,
+      "loss": 0.6856,
+      "step": 6507
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.794108339175099e-05,
+      "loss": 0.7505,
+      "step": 6508
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7927146834948812e-05,
+      "loss": 0.7828,
+      "step": 6509
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.791321451073442e-05,
+      "loss": 0.7491,
+      "step": 6510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7899286420946425e-05,
+      "loss": 0.741,
+      "step": 6511
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.788536256742292e-05,
+      "loss": 0.7288,
+      "step": 6512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.787144295200138e-05,
+      "loss": 0.7597,
+      "step": 6513
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7857527576518746e-05,
+      "loss": 0.7233,
+      "step": 6514
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7843616442811424e-05,
+      "loss": 0.7761,
+      "step": 6515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7829709552715223e-05,
+      "loss": 0.7314,
+      "step": 6516
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7815806908065408e-05,
+      "loss": 0.7858,
+      "step": 6517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7801908510696684e-05,
+      "loss": 0.8239,
+      "step": 6518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.778801436244319e-05,
+      "loss": 0.7722,
+      "step": 6519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7774124465138503e-05,
+      "loss": 0.7715,
+      "step": 6520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7760238820615654e-05,
+      "loss": 0.7238,
+      "step": 6521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.77463574307071e-05,
+      "loss": 0.6611,
+      "step": 6522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7732480297244734e-05,
+      "loss": 0.7031,
+      "step": 6523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771860742205988e-05,
+      "loss": 0.715,
+      "step": 6524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7704738806983346e-05,
+      "loss": 0.7989,
+      "step": 6525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769087445384533e-05,
+      "loss": 0.6924,
+      "step": 6526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677014364475474e-05,
+      "loss": 0.7719,
+      "step": 6527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7663158540702874e-05,
+      "loss": 0.7879,
+      "step": 6528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.764930698435603e-05,
+      "loss": 0.8427,
+      "step": 6529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.763545969726294e-05,
+      "loss": 0.7582,
+      "step": 6530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7621616681250986e-05,
+      "loss": 0.8055,
+      "step": 6531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7607777938146995e-05,
+      "loss": 0.8164,
+      "step": 6532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7593943469777234e-05,
+      "loss": 0.6584,
+      "step": 6533
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7580113277967435e-05,
+      "loss": 0.7962,
+      "step": 6534
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7566287364542722e-05,
+      "loss": 0.773,
+      "step": 6535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7552465731327673e-05,
+      "loss": 0.8186,
+      "step": 6536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.75386483801463e-05,
+      "loss": 0.7281,
+      "step": 6537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7524835312822045e-05,
+      "loss": 0.789,
+      "step": 6538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7511026531177804e-05,
+      "loss": 0.7799,
+      "step": 6539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7497222037035878e-05,
+      "loss": 0.7118,
+      "step": 6540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7483421832218026e-05,
+      "loss": 0.7763,
+      "step": 6541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.746962591854543e-05,
+      "loss": 0.7436,
+      "step": 6542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.745583429783869e-05,
+      "loss": 0.6869,
+      "step": 6543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.74420469719179e-05,
+      "loss": 0.8217,
+      "step": 6544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.742826394260251e-05,
+      "loss": 0.7957,
+      "step": 6545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.741448521171145e-05,
+      "loss": 0.6959,
+      "step": 6546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7400710781063072e-05,
+      "loss": 0.8253,
+      "step": 6547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.738694065247514e-05,
+      "loss": 0.8219,
+      "step": 6548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.73731748277649e-05,
+      "loss": 0.6954,
+      "step": 6549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7359413308748985e-05,
+      "loss": 0.7946,
+      "step": 6550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7345656097243474e-05,
+      "loss": 0.8162,
+      "step": 6551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.733190319506387e-05,
+      "loss": 0.6724,
+      "step": 6552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.731815460402511e-05,
+      "loss": 0.7308,
+      "step": 6553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.73044103259416e-05,
+      "loss": 0.7428,
+      "step": 6554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7290670362627115e-05,
+      "loss": 0.7523,
+      "step": 6555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.72769347158949e-05,
+      "loss": 0.7178,
+      "step": 6556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.726320338755762e-05,
+      "loss": 0.7672,
+      "step": 6557
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7249476379427364e-05,
+      "loss": 0.7248,
+      "step": 6558
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7235753693315658e-05,
+      "loss": 0.7003,
+      "step": 6559
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.722203533103346e-05,
+      "loss": 0.7871,
+      "step": 6560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.720832129439115e-05,
+      "loss": 0.6974,
+      "step": 6561
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7194611585198546e-05,
+      "loss": 0.6895,
+      "step": 6562
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7180906205264862e-05,
+      "loss": 0.6969,
+      "step": 6563
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7167205156398814e-05,
+      "loss": 0.7087,
+      "step": 6564
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7153508440408482e-05,
+      "loss": 0.769,
+      "step": 6565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.713981605910137e-05,
+      "loss": 0.7844,
+      "step": 6566
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7126128014284477e-05,
+      "loss": 0.7892,
+      "step": 6567
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.711244430776416e-05,
+      "loss": 0.6955,
+      "step": 6568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.709876494134623e-05,
+      "loss": 0.6898,
+      "step": 6569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7085089916835923e-05,
+      "loss": 0.8262,
+      "step": 6570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7071419236037917e-05,
+      "loss": 0.7526,
+      "step": 6571
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7057752900756286e-05,
+      "loss": 0.8311,
+      "step": 6572
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.704409091279454e-05,
+      "loss": 0.6826,
+      "step": 6573
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.703043327395566e-05,
+      "loss": 0.7204,
+      "step": 6574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7016779986041993e-05,
+      "loss": 0.7149,
+      "step": 6575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.700313105085533e-05,
+      "loss": 0.733,
+      "step": 6576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.698948647019691e-05,
+      "loss": 0.7162,
+      "step": 6577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.697584624586736e-05,
+      "loss": 0.7552,
+      "step": 6578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6962210379666766e-05,
+      "loss": 0.723,
+      "step": 6579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.694857887339462e-05,
+      "loss": 0.7794,
+      "step": 6580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6934951728849848e-05,
+      "loss": 0.7744,
+      "step": 6581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6921328947830773e-05,
+      "loss": 0.77,
+      "step": 6582
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6907710532135203e-05,
+      "loss": 0.8113,
+      "step": 6583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.689409648356029e-05,
+      "loss": 0.6043,
+      "step": 6584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6880486803902694e-05,
+      "loss": 0.8148,
+      "step": 6585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6866881494958435e-05,
+      "loss": 0.77,
+      "step": 6586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6853280558522984e-05,
+      "loss": 0.8161,
+      "step": 6587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6839683996391227e-05,
+      "loss": 0.7243,
+      "step": 6588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6826091810357464e-05,
+      "loss": 0.7314,
+      "step": 6589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6812504002215435e-05,
+      "loss": 0.7359,
+      "step": 6590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6798920573758293e-05,
+      "loss": 0.8648,
+      "step": 6591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.67853415267786e-05,
+      "loss": 0.7132,
+      "step": 6592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.677176686306839e-05,
+      "loss": 0.7808,
+      "step": 6593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6758196584419063e-05,
+      "loss": 0.8122,
+      "step": 6594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.674463069262146e-05,
+      "loss": 0.78,
+      "step": 6595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6731069189465842e-05,
+      "loss": 0.8205,
+      "step": 6596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6717512076741896e-05,
+      "loss": 0.7884,
+      "step": 6597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.670395935623873e-05,
+      "loss": 0.7873,
+      "step": 6598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6690411029744862e-05,
+      "loss": 0.7001,
+      "step": 6599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.667686709904822e-05,
+      "loss": 0.8096,
+      "step": 6600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6663327565936198e-05,
+      "loss": 0.8062,
+      "step": 6601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6649792432195576e-05,
+      "loss": 0.6962,
+      "step": 6602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6636261699612533e-05,
+      "loss": 0.7256,
+      "step": 6603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.662273536997272e-05,
+      "loss": 0.7689,
+      "step": 6604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6609213445061172e-05,
+      "loss": 0.748,
+      "step": 6605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6595695926662343e-05,
+      "loss": 0.8223,
+      "step": 6606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.658218281656011e-05,
+      "loss": 0.7832,
+      "step": 6607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6568674116537776e-05,
+      "loss": 0.8135,
+      "step": 6608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6555169828378052e-05,
+      "loss": 0.7358,
+      "step": 6609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6541669953863077e-05,
+      "loss": 0.761,
+      "step": 6610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.652817449477439e-05,
+      "loss": 0.6498,
+      "step": 6611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6514683452892955e-05,
+      "loss": 0.7682,
+      "step": 6612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.650119682999918e-05,
+      "loss": 0.7657,
+      "step": 6613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6487714627872852e-05,
+      "loss": 0.7705,
+      "step": 6614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6474236848293194e-05,
+      "loss": 0.7909,
+      "step": 6615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.646076349303884e-05,
+      "loss": 0.8129,
+      "step": 6616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6447294563887832e-05,
+      "loss": 0.7184,
+      "step": 6617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6433830062617643e-05,
+      "loss": 0.7676,
+      "step": 6618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.642036999100514e-05,
+      "loss": 0.7344,
+      "step": 6619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6406914350826656e-05,
+      "loss": 0.7424,
+      "step": 6620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6393463143857884e-05,
+      "loss": 0.7661,
+      "step": 6621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6380016371873928e-05,
+      "loss": 0.7985,
+      "step": 6622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.636657403664938e-05,
+      "loss": 0.7745,
+      "step": 6623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6353136139958164e-05,
+      "loss": 0.7238,
+      "step": 6624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.633970268357366e-05,
+      "loss": 0.7322,
+      "step": 6625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.632627366926865e-05,
+      "loss": 0.7635,
+      "step": 6626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6312849098815336e-05,
+      "loss": 0.7766,
+      "step": 6627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6299428973985332e-05,
+      "loss": 0.7366,
+      "step": 6628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.628601329654966e-05,
+      "loss": 0.7937,
+      "step": 6629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.627260206827876e-05,
+      "loss": 0.7382,
+      "step": 6630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.625919529094248e-05,
+      "loss": 0.7496,
+      "step": 6631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.624579296631008e-05,
+      "loss": 0.805,
+      "step": 6632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.623239509615026e-05,
+      "loss": 0.6989,
+      "step": 6633
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6219001682231094e-05,
+      "loss": 0.87,
+      "step": 6634
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6205612726320085e-05,
+      "loss": 0.6746,
+      "step": 6635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.619222823018414e-05,
+      "loss": 0.8095,
+      "step": 6636
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.617884819558958e-05,
+      "loss": 0.7531,
+      "step": 6637
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6165472624302157e-05,
+      "loss": 0.6962,
+      "step": 6638
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6152101518087008e-05,
+      "loss": 0.7331,
+      "step": 6639
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.61387348787087e-05,
+      "loss": 0.8576,
+      "step": 6640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.612537270793117e-05,
+      "loss": 0.7971,
+      "step": 6641
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.611201500751783e-05,
+      "loss": 0.7023,
+      "step": 6642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6098661779231462e-05,
+      "loss": 0.7363,
+      "step": 6643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.608531302483425e-05,
+      "loss": 0.797,
+      "step": 6644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6071968746087813e-05,
+      "loss": 0.823,
+      "step": 6645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6058628944753155e-05,
+      "loss": 0.8039,
+      "step": 6646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.604529362259072e-05,
+      "loss": 0.7993,
+      "step": 6647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6031962781360326e-05,
+      "loss": 0.7183,
+      "step": 6648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.601863642282123e-05,
+      "loss": 0.6917,
+      "step": 6649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6005314548732076e-05,
+      "loss": 0.7339,
+      "step": 6650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5991997160850903e-05,
+      "loss": 0.799,
+      "step": 6651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5978684260935218e-05,
+      "loss": 0.8217,
+      "step": 6652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.596537585074188e-05,
+      "loss": 0.7505,
+      "step": 6653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.595207193202718e-05,
+      "loss": 0.7917,
+      "step": 6654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.593877250654679e-05,
+      "loss": 0.7844,
+      "step": 6655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5925477576055807e-05,
+      "loss": 0.6982,
+      "step": 6656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5912187142308766e-05,
+      "loss": 0.6748,
+      "step": 6657
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5898901207059553e-05,
+      "loss": 0.8222,
+      "step": 6658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.588561977206149e-05,
+      "loss": 0.8059,
+      "step": 6659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5872342839067306e-05,
+      "loss": 0.8354,
+      "step": 6660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5859070409829102e-05,
+      "loss": 0.7035,
+      "step": 6661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.584580248609846e-05,
+      "loss": 0.7123,
+      "step": 6662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.583253906962629e-05,
+      "loss": 0.8269,
+      "step": 6663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5819280162162948e-05,
+      "loss": 0.8348,
+      "step": 6664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5806025765458178e-05,
+      "loss": 0.7588,
+      "step": 6665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.579277588126114e-05,
+      "loss": 0.7722,
+      "step": 6666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.577953051132039e-05,
+      "loss": 0.749,
+      "step": 6667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.576628965738389e-05,
+      "loss": 0.7605,
+      "step": 6668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.575305332119902e-05,
+      "loss": 0.769,
+      "step": 6669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5739821504512538e-05,
+      "loss": 0.7036,
+      "step": 6670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5726594209070604e-05,
+      "loss": 0.8385,
+      "step": 6671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.571337143661884e-05,
+      "loss": 0.7585,
+      "step": 6672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.57001531889022e-05,
+      "loss": 0.7016,
+      "step": 6673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.568693946766506e-05,
+      "loss": 0.8058,
+      "step": 6674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5673730274651237e-05,
+      "loss": 0.759,
+      "step": 6675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5660525611603906e-05,
+      "loss": 0.6815,
+      "step": 6676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5647325480265655e-05,
+      "loss": 0.697,
+      "step": 6677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.563412988237848e-05,
+      "loss": 0.7954,
+      "step": 6678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5620938819683777e-05,
+      "loss": 0.7633,
+      "step": 6679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5607752293922352e-05,
+      "loss": 0.7663,
+      "step": 6680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5594570306834367e-05,
+      "loss": 0.7015,
+      "step": 6681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5581392860159472e-05,
+      "loss": 0.6943,
+      "step": 6682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5568219955636642e-05,
+      "loss": 0.8713,
+      "step": 6683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5555051595004276e-05,
+      "loss": 0.7605,
+      "step": 6684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5541887780000185e-05,
+      "loss": 0.7818,
+      "step": 6685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.552872851236157e-05,
+      "loss": 0.7029,
+      "step": 6686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5515573793825018e-05,
+      "loss": 0.7186,
+      "step": 6687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.550242362612654e-05,
+      "loss": 0.8106,
+      "step": 6688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.548927801100154e-05,
+      "loss": 0.7638,
+      "step": 6689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.547613695018479e-05,
+      "loss": 0.762,
+      "step": 6690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.546300044541053e-05,
+      "loss": 0.7756,
+      "step": 6691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.544986849841234e-05,
+      "loss": 0.7024,
+      "step": 6692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5436741110923196e-05,
+      "loss": 0.7246,
+      "step": 6693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5423618284675527e-05,
+      "loss": 0.7122,
+      "step": 6694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5410500021401108e-05,
+      "loss": 0.7308,
+      "step": 6695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5397386322831132e-05,
+      "loss": 0.7797,
+      "step": 6696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5384277190696185e-05,
+      "loss": 0.7713,
+      "step": 6697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5371172626726243e-05,
+      "loss": 0.7144,
+      "step": 6698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.53580726326507e-05,
+      "loss": 0.8638,
+      "step": 6699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5344977210198314e-05,
+      "loss": 0.7686,
+      "step": 6700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5331886361097297e-05,
+      "loss": 0.8843,
+      "step": 6701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5318800087075197e-05,
+      "loss": 0.7594,
+      "step": 6702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5305718389858993e-05,
+      "loss": 0.63,
+      "step": 6703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5292641271175035e-05,
+      "loss": 0.7231,
+      "step": 6704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.52795687327491e-05,
+      "loss": 0.7108,
+      "step": 6705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.526650077630633e-05,
+      "loss": 0.7342,
+      "step": 6706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.525343740357128e-05,
+      "loss": 0.8307,
+      "step": 6707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5240378616267886e-05,
+      "loss": 0.7958,
+      "step": 6708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5227324416119514e-05,
+      "loss": 0.7593,
+      "step": 6709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5214274804848889e-05,
+      "loss": 0.7717,
+      "step": 6710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5201229784178122e-05,
+      "loss": 0.7276,
+      "step": 6711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.518818935582877e-05,
+      "loss": 0.6848,
+      "step": 6712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.517515352152173e-05,
+      "loss": 0.7735,
+      "step": 6713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5162122282977326e-05,
+      "loss": 0.6819,
+      "step": 6714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.514909564191525e-05,
+      "loss": 0.7102,
+      "step": 6715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5136073600054606e-05,
+      "loss": 0.6862,
+      "step": 6716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5123056159113891e-05,
+      "loss": 0.7096,
+      "step": 6717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5110043320810985e-05,
+      "loss": 0.7635,
+      "step": 6718
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5097035086863165e-05,
+      "loss": 0.7152,
+      "step": 6719
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5084031458987086e-05,
+      "loss": 0.7236,
+      "step": 6720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.507103243889884e-05,
+      "loss": 0.7382,
+      "step": 6721
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.505803802831387e-05,
+      "loss": 0.7179,
+      "step": 6722
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5045048228947012e-05,
+      "loss": 0.6867,
+      "step": 6723
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5032063042512513e-05,
+      "loss": 0.6248,
+      "step": 6724
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5019082470723989e-05,
+      "loss": 0.7318,
+      "step": 6725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5006106515294472e-05,
+      "loss": 0.7625,
+      "step": 6726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4993135177936346e-05,
+      "loss": 0.7918,
+      "step": 6727
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4980168460361448e-05,
+      "loss": 0.8139,
+      "step": 6728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4967206364280955e-05,
+      "loss": 0.7551,
+      "step": 6729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4954248891405432e-05,
+      "loss": 0.7434,
+      "step": 6730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4941296043444868e-05,
+      "loss": 0.7009,
+      "step": 6731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.492834782210863e-05,
+      "loss": 0.7542,
+      "step": 6732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.491540422910545e-05,
+      "loss": 0.7741,
+      "step": 6733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4902465266143472e-05,
+      "loss": 0.8507,
+      "step": 6734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.488953093493023e-05,
+      "loss": 0.7987,
+      "step": 6735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4876601237172632e-05,
+      "loss": 0.7328,
+      "step": 6736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4863676174576991e-05,
+      "loss": 0.8028,
+      "step": 6737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4850755748848993e-05,
+      "loss": 0.768,
+      "step": 6738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4837839961693728e-05,
+      "loss": 0.6551,
+      "step": 6739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4824928814815642e-05,
+      "loss": 0.7277,
+      "step": 6740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4812022309918627e-05,
+      "loss": 0.8063,
+      "step": 6741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4799120448705917e-05,
+      "loss": 0.7404,
+      "step": 6742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4786223232880137e-05,
+      "loss": 0.7796,
+      "step": 6743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4773330664143303e-05,
+      "loss": 0.7312,
+      "step": 6744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4760442744196817e-05,
+      "loss": 0.8083,
+      "step": 6745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.474755947474149e-05,
+      "loss": 0.8012,
+      "step": 6746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.473468085747749e-05,
+      "loss": 0.7226,
+      "step": 6747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4721806894104384e-05,
+      "loss": 0.7977,
+      "step": 6748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.470893758632112e-05,
+      "loss": 0.765,
+      "step": 6749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.469607293582601e-05,
+      "loss": 0.6896,
+      "step": 6750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.468321294431682e-05,
+      "loss": 0.8641,
+      "step": 6751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4670357613490638e-05,
+      "loss": 0.7851,
+      "step": 6752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4657506945043947e-05,
+      "loss": 0.6473,
+      "step": 6753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4644660940672627e-05,
+      "loss": 0.8328,
+      "step": 6754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4631819602071945e-05,
+      "loss": 0.7123,
+      "step": 6755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4618982930936532e-05,
+      "loss": 0.8584,
+      "step": 6756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4606150928960433e-05,
+      "loss": 0.8111,
+      "step": 6757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4593323597837044e-05,
+      "loss": 0.6257,
+      "step": 6758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4580500939259162e-05,
+      "loss": 0.6877,
+      "step": 6759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4567682954918987e-05,
+      "loss": 0.7839,
+      "step": 6760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4554869646508069e-05,
+      "loss": 0.7763,
+      "step": 6761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4542061015717356e-05,
+      "loss": 0.7459,
+      "step": 6762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4529257064237174e-05,
+      "loss": 0.7491,
+      "step": 6763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4516457793757216e-05,
+      "loss": 0.8029,
+      "step": 6764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.450366320596661e-05,
+      "loss": 0.708,
+      "step": 6765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4490873302553814e-05,
+      "loss": 0.6976,
+      "step": 6766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4478088085206686e-05,
+      "loss": 0.8464,
+      "step": 6767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4465307555612467e-05,
+      "loss": 0.7033,
+      "step": 6768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4452531715457757e-05,
+      "loss": 0.7142,
+      "step": 6769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.443976056642859e-05,
+      "loss": 0.7508,
+      "step": 6770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.442699411021033e-05,
+      "loss": 0.7385,
+      "step": 6771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4414232348487738e-05,
+      "loss": 0.7201,
+      "step": 6772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.440147528294497e-05,
+      "loss": 0.6835,
+      "step": 6773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4388722915265534e-05,
+      "loss": 0.665,
+      "step": 6774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.437597524713234e-05,
+      "loss": 0.7445,
+      "step": 6775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4363232280227673e-05,
+      "loss": 0.792,
+      "step": 6776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4350494016233196e-05,
+      "loss": 0.7524,
+      "step": 6777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4337760456829946e-05,
+      "loss": 0.7416,
+      "step": 6778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4325031603698336e-05,
+      "loss": 0.7039,
+      "step": 6779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4312307458518187e-05,
+      "loss": 0.768,
+      "step": 6780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4299588022968674e-05,
+      "loss": 0.7467,
+      "step": 6781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4286873298728332e-05,
+      "loss": 0.8574,
+      "step": 6782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4274163287475129e-05,
+      "loss": 0.6843,
+      "step": 6783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4261457990886362e-05,
+      "loss": 0.7208,
+      "step": 6784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.424875741063873e-05,
+      "loss": 0.763,
+      "step": 6785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4236061548408291e-05,
+      "loss": 0.7074,
+      "step": 6786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.42233704058705e-05,
+      "loss": 0.8082,
+      "step": 6787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.421068398470018e-05,
+      "loss": 0.7266,
+      "step": 6788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.419800228657151e-05,
+      "loss": 0.7325,
+      "step": 6789
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4185325313158109e-05,
+      "loss": 0.8021,
+      "step": 6790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.41726530661329e-05,
+      "loss": 0.7115,
+      "step": 6791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4159985547168231e-05,
+      "loss": 0.6956,
+      "step": 6792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4147322757935794e-05,
+      "loss": 0.7992,
+      "step": 6793
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4134664700106675e-05,
+      "loss": 0.6894,
+      "step": 6794
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4122011375351336e-05,
+      "loss": 0.7754,
+      "step": 6795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4109362785339603e-05,
+      "loss": 0.718,
+      "step": 6796
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4096718931740688e-05,
+      "loss": 0.6915,
+      "step": 6797
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4084079816223161e-05,
+      "loss": 0.6599,
+      "step": 6798
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4071445440455005e-05,
+      "loss": 0.6956,
+      "step": 6799
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4058815806103542e-05,
+      "loss": 0.7546,
+      "step": 6800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.404619091483546e-05,
+      "loss": 0.7931,
+      "step": 6801
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4033570768316873e-05,
+      "loss": 0.6825,
+      "step": 6802
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4020955368213218e-05,
+      "loss": 0.7595,
+      "step": 6803
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.400834471618932e-05,
+      "loss": 0.7235,
+      "step": 6804
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3995738813909393e-05,
+      "loss": 0.6687,
+      "step": 6805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3983137663036994e-05,
+      "loss": 0.7675,
+      "step": 6806
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3970541265235088e-05,
+      "loss": 0.7217,
+      "step": 6807
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3957949622165984e-05,
+      "loss": 0.6418,
+      "step": 6808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3945362735491357e-05,
+      "loss": 0.8076,
+      "step": 6809
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3932780606872315e-05,
+      "loss": 0.7691,
+      "step": 6810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3920203237969276e-05,
+      "loss": 0.7166,
+      "step": 6811
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3907630630442043e-05,
+      "loss": 0.7044,
+      "step": 6812
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3895062785949803e-05,
+      "loss": 0.7534,
+      "step": 6813
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3882499706151102e-05,
+      "loss": 0.7637,
+      "step": 6814
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3869941392703873e-05,
+      "loss": 0.8054,
+      "step": 6815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.385738784726539e-05,
+      "loss": 0.7472,
+      "step": 6816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3844839071492349e-05,
+      "loss": 0.76,
+      "step": 6817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3832295067040768e-05,
+      "loss": 0.6974,
+      "step": 6818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3819755835566045e-05,
+      "loss": 0.8539,
+      "step": 6819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3807221378722979e-05,
+      "loss": 0.747,
+      "step": 6820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3794691698165701e-05,
+      "loss": 0.7878,
+      "step": 6821
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3782166795547735e-05,
+      "loss": 0.6931,
+      "step": 6822
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3769646672521963e-05,
+      "loss": 0.7068,
+      "step": 6823
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3757131330740635e-05,
+      "loss": 0.7005,
+      "step": 6824
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3744620771855377e-05,
+      "loss": 0.8287,
+      "step": 6825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3732114997517182e-05,
+      "loss": 0.7435,
+      "step": 6826
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3719614009376414e-05,
+      "loss": 0.7566,
+      "step": 6827
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3707117809082787e-05,
+      "loss": 0.8523,
+      "step": 6828
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3694626398285421e-05,
+      "loss": 0.7747,
+      "step": 6829
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3682139778632774e-05,
+      "loss": 0.8135,
+      "step": 6830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3669657951772679e-05,
+      "loss": 0.6924,
+      "step": 6831
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3657180919352336e-05,
+      "loss": 0.7752,
+      "step": 6832
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3644708683018309e-05,
+      "loss": 0.7281,
+      "step": 6833
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3632241244416538e-05,
+      "loss": 0.7292,
+      "step": 6834
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3619778605192308e-05,
+      "loss": 0.8234,
+      "step": 6835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3607320766990311e-05,
+      "loss": 0.776,
+      "step": 6836
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.359486773145458e-05,
+      "loss": 0.754,
+      "step": 6837
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3582419500228495e-05,
+      "loss": 0.7319,
+      "step": 6838
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3569976074954855e-05,
+      "loss": 0.7101,
+      "step": 6839
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3557537457275771e-05,
+      "loss": 0.6481,
+      "step": 6840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3545103648832752e-05,
+      "loss": 0.7607,
+      "step": 6841
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3532674651266652e-05,
+      "loss": 0.7758,
+      "step": 6842
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3520250466217705e-05,
+      "loss": 0.6771,
+      "step": 6843
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3507831095325507e-05,
+      "loss": 0.7764,
+      "step": 6844
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3495416540229011e-05,
+      "loss": 0.7531,
+      "step": 6845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3483006802566544e-05,
+      "loss": 0.6893,
+      "step": 6846
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3470601883975797e-05,
+      "loss": 0.8324,
+      "step": 6847
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3458201786093794e-05,
+      "loss": 0.766,
+      "step": 6848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.344580651055699e-05,
+      "loss": 0.7942,
+      "step": 6849
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3433416059001152e-05,
+      "loss": 0.7491,
+      "step": 6850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.342103043306141e-05,
+      "loss": 0.7926,
+      "step": 6851
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3408649634372278e-05,
+      "loss": 0.7459,
+      "step": 6852
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.339627366456761e-05,
+      "loss": 0.6636,
+      "step": 6853
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3383902525280661e-05,
+      "loss": 0.7694,
+      "step": 6854
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3371536218144016e-05,
+      "loss": 0.7772,
+      "step": 6855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3359174744789626e-05,
+      "loss": 0.7524,
+      "step": 6856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3346818106848807e-05,
+      "loss": 0.8117,
+      "step": 6857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3334466305952232e-05,
+      "loss": 0.7528,
+      "step": 6858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3322119343729966e-05,
+      "loss": 0.8301,
+      "step": 6859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3309777221811398e-05,
+      "loss": 0.765,
+      "step": 6860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3297439941825296e-05,
+      "loss": 0.6629,
+      "step": 6861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.328510750539978e-05,
+      "loss": 0.8116,
+      "step": 6862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3272779914162341e-05,
+      "loss": 0.7073,
+      "step": 6863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3260457169739821e-05,
+      "loss": 0.7317,
+      "step": 6864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.324813927375843e-05,
+      "loss": 0.7142,
+      "step": 6865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3235826227843734e-05,
+      "loss": 0.7404,
+      "step": 6866
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3223518033620658e-05,
+      "loss": 0.6305,
+      "step": 6867
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.321121469271348e-05,
+      "loss": 0.732,
+      "step": 6868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.319891620674587e-05,
+      "loss": 0.7063,
+      "step": 6869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3186622577340818e-05,
+      "loss": 0.7732,
+      "step": 6870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3174333806120697e-05,
+      "loss": 0.6208,
+      "step": 6871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3162049894707207e-05,
+      "loss": 0.6931,
+      "step": 6872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3149770844721466e-05,
+      "loss": 0.6936,
+      "step": 6873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3137496657783893e-05,
+      "loss": 0.7556,
+      "step": 6874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3125227335514295e-05,
+      "loss": 0.7626,
+      "step": 6875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.311296287953182e-05,
+      "loss": 0.8029,
+      "step": 6876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.310070329145498e-05,
+      "loss": 0.7666,
+      "step": 6877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3088448572901662e-05,
+      "loss": 0.7476,
+      "step": 6878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3076198725489098e-05,
+      "loss": 0.7267,
+      "step": 6879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3063953750833862e-05,
+      "loss": 0.79,
+      "step": 6880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.30517136505519e-05,
+      "loss": 0.8179,
+      "step": 6881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3039478426258517e-05,
+      "loss": 0.7568,
+      "step": 6882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3027248079568367e-05,
+      "loss": 0.7583,
+      "step": 6883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3015022612095462e-05,
+      "loss": 0.6957,
+      "step": 6884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3002802025453176e-05,
+      "loss": 0.7553,
+      "step": 6885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2990586321254227e-05,
+      "loss": 0.829,
+      "step": 6886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2978375501110696e-05,
+      "loss": 0.7791,
+      "step": 6887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2966169566634028e-05,
+      "loss": 0.8601,
+      "step": 6888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2953968519435017e-05,
+      "loss": 0.7485,
+      "step": 6889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2941772361123793e-05,
+      "loss": 0.6803,
+      "step": 6890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2929581093309878e-05,
+      "loss": 0.7911,
+      "step": 6891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2917394717602121e-05,
+      "loss": 0.7374,
+      "step": 6892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2905213235608732e-05,
+      "loss": 0.7436,
+      "step": 6893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2893036648937274e-05,
+      "loss": 0.7901,
+      "step": 6894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2880864959194665e-05,
+      "loss": 0.7632,
+      "step": 6895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2868698167987181e-05,
+      "loss": 0.7671,
+      "step": 6896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2856536276920434e-05,
+      "loss": 0.6733,
+      "step": 6897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2844379287599428e-05,
+      "loss": 0.7313,
+      "step": 6898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2832227201628483e-05,
+      "loss": 0.7652,
+      "step": 6899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2820080020611286e-05,
+      "loss": 0.7475,
+      "step": 6900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2807937746150872e-05,
+      "loss": 0.7644,
+      "step": 6901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2795800379849637e-05,
+      "loss": 0.6719,
+      "step": 6902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2783667923309317e-05,
+      "loss": 0.8314,
+      "step": 6903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2771540378131014e-05,
+      "loss": 0.7638,
+      "step": 6904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2759417745915176e-05,
+      "loss": 0.6938,
+      "step": 6905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2747300028261577e-05,
+      "loss": 0.7534,
+      "step": 6906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2735187226769401e-05,
+      "loss": 0.7543,
+      "step": 6907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2723079343037142e-05,
+      "loss": 0.6923,
+      "step": 6908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2710976378662625e-05,
+      "loss": 0.7321,
+      "step": 6909
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2698878335243092e-05,
+      "loss": 0.777,
+      "step": 6910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.268678521437508e-05,
+      "loss": 0.7896,
+      "step": 6911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2674697017654492e-05,
+      "loss": 0.7828,
+      "step": 6912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2662613746676577e-05,
+      "loss": 0.6672,
+      "step": 6913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2650535403035952e-05,
+      "loss": 0.7556,
+      "step": 6914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2638461988326555e-05,
+      "loss": 0.7194,
+      "step": 6915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2626393504141704e-05,
+      "loss": 0.7824,
+      "step": 6916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2614329952074028e-05,
+      "loss": 0.7696,
+      "step": 6917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2602271333715559e-05,
+      "loss": 0.6639,
+      "step": 6918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2590217650657644e-05,
+      "loss": 0.7839,
+      "step": 6919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2578168904490961e-05,
+      "loss": 0.7901,
+      "step": 6920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2566125096805581e-05,
+      "loss": 0.7942,
+      "step": 6921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2554086229190882e-05,
+      "loss": 0.7651,
+      "step": 6922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2542052303235618e-05,
+      "loss": 0.7205,
+      "step": 6923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2530023320527884e-05,
+      "loss": 0.6883,
+      "step": 6924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2517999282655096e-05,
+      "loss": 0.7069,
+      "step": 6925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2505980191204082e-05,
+      "loss": 0.7679,
+      "step": 6926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.249396604776093e-05,
+      "loss": 0.7616,
+      "step": 6927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.248195685391117e-05,
+      "loss": 0.7681,
+      "step": 6928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.24699526112396e-05,
+      "loss": 0.6302,
+      "step": 6929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2457953321330407e-05,
+      "loss": 0.6601,
+      "step": 6930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2445958985767104e-05,
+      "loss": 0.7392,
+      "step": 6931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2433969606132567e-05,
+      "loss": 0.6826,
+      "step": 6932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2421985184009e-05,
+      "loss": 0.745,
+      "step": 6933
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2410005720977968e-05,
+      "loss": 0.7477,
+      "step": 6934
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2398031218620371e-05,
+      "loss": 0.6641,
+      "step": 6935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2386061678516454e-05,
+      "loss": 0.8677,
+      "step": 6936
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2374097102245829e-05,
+      "loss": 0.7433,
+      "step": 6937
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2362137491387432e-05,
+      "loss": 0.7936,
+      "step": 6938
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2350182847519547e-05,
+      "loss": 0.8212,
+      "step": 6939
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2338233172219798e-05,
+      "loss": 0.7598,
+      "step": 6940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2326288467065161e-05,
+      "loss": 0.753,
+      "step": 6941
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2314348733631959e-05,
+      "loss": 0.7556,
+      "step": 6942
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.230241397349583e-05,
+      "loss": 0.8208,
+      "step": 6943
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2290484188231815e-05,
+      "loss": 0.7166,
+      "step": 6944
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2278559379414245e-05,
+      "loss": 0.6718,
+      "step": 6945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2266639548616798e-05,
+      "loss": 0.8104,
+      "step": 6946
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2254724697412545e-05,
+      "loss": 0.712,
+      "step": 6947
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2242814827373844e-05,
+      "loss": 0.854,
+      "step": 6948
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.223090994007241e-05,
+      "loss": 0.7581,
+      "step": 6949
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.221901003707932e-05,
+      "loss": 0.7525,
+      "step": 6950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2207115119964968e-05,
+      "loss": 0.7419,
+      "step": 6951
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.219522519029911e-05,
+      "loss": 0.7937,
+      "step": 6952
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2183340249650827e-05,
+      "loss": 0.8101,
+      "step": 6953
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.217146029958855e-05,
+      "loss": 0.7278,
+      "step": 6954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2159585341680064e-05,
+      "loss": 0.7135,
+      "step": 6955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2147715377492452e-05,
+      "loss": 0.8109,
+      "step": 6956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2135850408592209e-05,
+      "loss": 0.7227,
+      "step": 6957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2123990436545108e-05,
+      "loss": 0.6933,
+      "step": 6958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.211213546291629e-05,
+      "loss": 0.7705,
+      "step": 6959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.210028548927023e-05,
+      "loss": 0.6948,
+      "step": 6960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.208844051717073e-05,
+      "loss": 0.6734,
+      "step": 6961
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2076600548180972e-05,
+      "loss": 0.7141,
+      "step": 6962
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.206476558386344e-05,
+      "loss": 0.6585,
+      "step": 6963
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2052935625779971e-05,
+      "loss": 0.6953,
+      "step": 6964
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.204111067549174e-05,
+      "loss": 0.7321,
+      "step": 6965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2029290734559245e-05,
+      "loss": 0.7087,
+      "step": 6966
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2017475804542373e-05,
+      "loss": 0.813,
+      "step": 6967
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2005665887000289e-05,
+      "loss": 0.7485,
+      "step": 6968
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.199386098349154e-05,
+      "loss": 0.7699,
+      "step": 6969
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1982061095573982e-05,
+      "loss": 0.7083,
+      "step": 6970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1970266224804821e-05,
+      "loss": 0.6958,
+      "step": 6971
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1958476372740613e-05,
+      "loss": 0.8325,
+      "step": 6972
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1946691540937226e-05,
+      "loss": 0.6761,
+      "step": 6973
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1934911730949894e-05,
+      "loss": 0.773,
+      "step": 6974
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1923136944333163e-05,
+      "loss": 0.7652,
+      "step": 6975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.191136718264092e-05,
+      "loss": 0.7923,
+      "step": 6976
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1899602447426417e-05,
+      "loss": 0.7331,
+      "step": 6977
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1887842740242216e-05,
+      "loss": 0.7673,
+      "step": 6978
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1876088062640218e-05,
+      "loss": 0.765,
+      "step": 6979
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1864338416171644e-05,
+      "loss": 0.7197,
+      "step": 6980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1852593802387108e-05,
+      "loss": 0.7731,
+      "step": 6981
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.18408542228365e-05,
+      "loss": 0.7874,
+      "step": 6982
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1829119679069072e-05,
+      "loss": 0.7495,
+      "step": 6983
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1817390172633403e-05,
+      "loss": 0.8128,
+      "step": 6984
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1805665705077407e-05,
+      "loss": 0.7437,
+      "step": 6985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1793946277948354e-05,
+      "loss": 0.8115,
+      "step": 6986
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1782231892792828e-05,
+      "loss": 0.771,
+      "step": 6987
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1770522551156748e-05,
+      "loss": 0.7715,
+      "step": 6988
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1758818254585369e-05,
+      "loss": 0.6156,
+      "step": 6989
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1747119004623285e-05,
+      "loss": 0.7017,
+      "step": 6990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1735424802814426e-05,
+      "loss": 0.7183,
+      "step": 6991
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.172373565070205e-05,
+      "loss": 0.7704,
+      "step": 6992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1712051549828739e-05,
+      "loss": 0.8536,
+      "step": 6993
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1700372501736433e-05,
+      "loss": 0.7496,
+      "step": 6994
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1688698507966372e-05,
+      "loss": 0.7241,
+      "step": 6995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1677029570059178e-05,
+      "loss": 0.735,
+      "step": 6996
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1665365689554763e-05,
+      "loss": 0.7355,
+      "step": 6997
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.165370686799237e-05,
+      "loss": 0.7086,
+      "step": 6998
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1642053106910616e-05,
+      "loss": 0.7672,
+      "step": 6999
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1630404407847412e-05,
+      "loss": 0.7656,
+      "step": 7000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1618760772340015e-05,
+      "loss": 0.7254,
+      "step": 7001
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1607122201925002e-05,
+      "loss": 0.8292,
+      "step": 7002
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1595488698138302e-05,
+      "loss": 0.7219,
+      "step": 7003
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1583860262515161e-05,
+      "loss": 0.7702,
+      "step": 7004
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1572236896590139e-05,
+      "loss": 0.8065,
+      "step": 7005
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1560618601897182e-05,
+      "loss": 0.7098,
+      "step": 7006
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1549005379969518e-05,
+      "loss": 0.8288,
+      "step": 7007
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1537397232339719e-05,
+      "loss": 0.7763,
+      "step": 7008
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1525794160539689e-05,
+      "loss": 0.8265,
+      "step": 7009
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1514196166100655e-05,
+      "loss": 0.6761,
+      "step": 7010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.150260325055319e-05,
+      "loss": 0.7225,
+      "step": 7011
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1491015415427175e-05,
+      "loss": 0.7999,
+      "step": 7012
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1479432662251849e-05,
+      "loss": 0.7752,
+      "step": 7013
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1467854992555732e-05,
+      "loss": 0.7504,
+      "step": 7014
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1456282407866747e-05,
+      "loss": 0.82,
+      "step": 7015
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1444714909712085e-05,
+      "loss": 0.714,
+      "step": 7016
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1433152499618265e-05,
+      "loss": 0.7172,
+      "step": 7017
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1421595179111189e-05,
+      "loss": 0.7866,
+      "step": 7018
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1410042949716038e-05,
+      "loss": 0.7512,
+      "step": 7019
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1398495812957333e-05,
+      "loss": 0.77,
+      "step": 7020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1386953770358933e-05,
+      "loss": 0.8644,
+      "step": 7021
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1375416823444008e-05,
+      "loss": 0.6443,
+      "step": 7022
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1363884973735067e-05,
+      "loss": 0.7364,
+      "step": 7023
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1352358222753944e-05,
+      "loss": 0.6162,
+      "step": 7024
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.134083657202179e-05,
+      "loss": 0.7903,
+      "step": 7025
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.132932002305912e-05,
+      "loss": 0.7243,
+      "step": 7026
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1317808577385735e-05,
+      "loss": 0.7372,
+      "step": 7027
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1306302236520777e-05,
+      "loss": 0.8255,
+      "step": 7028
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1294801001982713e-05,
+      "loss": 0.7423,
+      "step": 7029
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1283304875289336e-05,
+      "loss": 0.7181,
+      "step": 7030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.127181385795777e-05,
+      "loss": 0.8722,
+      "step": 7031
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1260327951504451e-05,
+      "loss": 0.7449,
+      "step": 7032
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1248847157445148e-05,
+      "loss": 0.6511,
+      "step": 7033
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1237371477294977e-05,
+      "loss": 0.7352,
+      "step": 7034
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1225900912568337e-05,
+      "loss": 0.7694,
+      "step": 7035
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1214435464779006e-05,
+      "loss": 0.7308,
+      "step": 7036
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.120297513544003e-05,
+      "loss": 0.7728,
+      "step": 7037
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.119151992606381e-05,
+      "loss": 0.7569,
+      "step": 7038
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1180069838162071e-05,
+      "loss": 0.7638,
+      "step": 7039
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1168624873245848e-05,
+      "loss": 0.7441,
+      "step": 7040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1157185032825518e-05,
+      "loss": 0.6864,
+      "step": 7041
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.114575031841077e-05,
+      "loss": 0.7054,
+      "step": 7042
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1134320731510617e-05,
+      "loss": 0.7923,
+      "step": 7043
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1122896273633388e-05,
+      "loss": 0.7658,
+      "step": 7044
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.111147694628677e-05,
+      "loss": 0.5858,
+      "step": 7045
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1100062750977736e-05,
+      "loss": 0.776,
+      "step": 7046
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1088653689212592e-05,
+      "loss": 0.7092,
+      "step": 7047
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1077249762496966e-05,
+      "loss": 0.7559,
+      "step": 7048
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1065850972335812e-05,
+      "loss": 0.6752,
+      "step": 7049
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1054457320233408e-05,
+      "loss": 0.6555,
+      "step": 7050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1043068807693336e-05,
+      "loss": 0.7394,
+      "step": 7051
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1031685436218541e-05,
+      "loss": 0.8213,
+      "step": 7052
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1020307207311242e-05,
+      "loss": 0.743,
+      "step": 7053
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1008934122473002e-05,
+      "loss": 0.7415,
+      "step": 7054
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0997566183204715e-05,
+      "loss": 0.6937,
+      "step": 7055
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0986203391006583e-05,
+      "loss": 0.7532,
+      "step": 7056
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.097484574737812e-05,
+      "loss": 0.7271,
+      "step": 7057
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0963493253818175e-05,
+      "loss": 0.7819,
+      "step": 7058
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0952145911824918e-05,
+      "loss": 0.7029,
+      "step": 7059
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0940803722895826e-05,
+      "loss": 0.803,
+      "step": 7060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0929466688527707e-05,
+      "loss": 0.8109,
+      "step": 7061
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0918134810216685e-05,
+      "loss": 0.7052,
+      "step": 7062
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.090680808945821e-05,
+      "loss": 0.7699,
+      "step": 7063
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0895486527747023e-05,
+      "loss": 0.7085,
+      "step": 7064
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0884170126577242e-05,
+      "loss": 0.7355,
+      "step": 7065
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.087285888744225e-05,
+      "loss": 0.7869,
+      "step": 7066
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0861552811834774e-05,
+      "loss": 0.6822,
+      "step": 7067
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0850251901246843e-05,
+      "loss": 0.7032,
+      "step": 7068
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0838956157169827e-05,
+      "loss": 0.8057,
+      "step": 7069
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0827665581094382e-05,
+      "loss": 0.7568,
+      "step": 7070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0816380174510526e-05,
+      "loss": 0.7903,
+      "step": 7071
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0805099938907565e-05,
+      "loss": 0.6529,
+      "step": 7072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0793824875774122e-05,
+      "loss": 0.7354,
+      "step": 7073
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.078255498659813e-05,
+      "loss": 0.7572,
+      "step": 7074
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0771290272866886e-05,
+      "loss": 0.6898,
+      "step": 7075
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0760030736066951e-05,
+      "loss": 0.7226,
+      "step": 7076
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0748776377684223e-05,
+      "loss": 0.6368,
+      "step": 7077
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0737527199203923e-05,
+      "loss": 0.7266,
+      "step": 7078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0726283202110572e-05,
+      "loss": 0.7659,
+      "step": 7079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0715044387888023e-05,
+      "loss": 0.7316,
+      "step": 7080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0703810758019439e-05,
+      "loss": 0.7735,
+      "step": 7081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0692582313987293e-05,
+      "loss": 0.8433,
+      "step": 7082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0681359057273388e-05,
+      "loss": 0.842,
+      "step": 7083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0670140989358818e-05,
+      "loss": 0.7454,
+      "step": 7084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.065892811172403e-05,
+      "loss": 0.7624,
+      "step": 7085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0647720425848756e-05,
+      "loss": 0.7108,
+      "step": 7086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0636517933212048e-05,
+      "loss": 0.727,
+      "step": 7087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0625320635292264e-05,
+      "loss": 0.7582,
+      "step": 7088
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0614128533567113e-05,
+      "loss": 0.7149,
+      "step": 7089
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0602941629513586e-05,
+      "loss": 0.8555,
+      "step": 7090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0591759924607986e-05,
+      "loss": 0.7825,
+      "step": 7091
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0580583420325946e-05,
+      "loss": 0.7642,
+      "step": 7092
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0569412118142402e-05,
+      "loss": 0.6669,
+      "step": 7093
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0558246019531598e-05,
+      "loss": 0.7166,
+      "step": 7094
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0547085125967126e-05,
+      "loss": 0.6935,
+      "step": 7095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0535929438921854e-05,
+      "loss": 0.6536,
+      "step": 7096
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.052477895986797e-05,
+      "loss": 0.7838,
+      "step": 7097
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0513633690276985e-05,
+      "loss": 0.8159,
+      "step": 7098
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0502493631619715e-05,
+      "loss": 0.752,
+      "step": 7099
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.049135878536629e-05,
+      "loss": 0.7107,
+      "step": 7100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0480229152986154e-05,
+      "loss": 0.7646,
+      "step": 7101
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.046910473594806e-05,
+      "loss": 0.6537,
+      "step": 7102
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0457985535720061e-05,
+      "loss": 0.8187,
+      "step": 7103
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.044687155376956e-05,
+      "loss": 0.7101,
+      "step": 7104
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0435762791563237e-05,
+      "loss": 0.7992,
+      "step": 7105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0424659250567087e-05,
+      "loss": 0.7393,
+      "step": 7106
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0413560932246414e-05,
+      "loss": 0.7014,
+      "step": 7107
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.040246783806586e-05,
+      "loss": 0.8488,
+      "step": 7108
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.039137996948935e-05,
+      "loss": 0.7152,
+      "step": 7109
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0380297327980127e-05,
+      "loss": 0.7167,
+      "step": 7110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0369219915000734e-05,
+      "loss": 0.7724,
+      "step": 7111
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0358147732013046e-05,
+      "loss": 0.7346,
+      "step": 7112
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.034708078047822e-05,
+      "loss": 0.7746,
+      "step": 7113
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.033601906185676e-05,
+      "loss": 0.7421,
+      "step": 7114
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0324962577608455e-05,
+      "loss": 0.773,
+      "step": 7115
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0313911329192394e-05,
+      "loss": 0.7499,
+      "step": 7116
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0302865318066996e-05,
+      "loss": 0.7709,
+      "step": 7117
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0291824545689976e-05,
+      "loss": 0.7356,
+      "step": 7118
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0280789013518361e-05,
+      "loss": 0.6831,
+      "step": 7119
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0269758723008489e-05,
+      "loss": 0.7648,
+      "step": 7120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0258733675616006e-05,
+      "loss": 0.6722,
+      "step": 7121
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.024771387279585e-05,
+      "loss": 0.7509,
+      "step": 7122
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0236699316002313e-05,
+      "loss": 0.7937,
+      "step": 7123
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0225690006688937e-05,
+      "loss": 0.6773,
+      "step": 7124
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0214685946308595e-05,
+      "loss": 0.8404,
+      "step": 7125
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0203687136313495e-05,
+      "loss": 0.773,
+      "step": 7126
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0192693578155115e-05,
+      "loss": 0.6649,
+      "step": 7127
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0181705273284248e-05,
+      "loss": 0.785,
+      "step": 7128
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0170722223150997e-05,
+      "loss": 0.7175,
+      "step": 7129
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0159744429204777e-05,
+      "loss": 0.7122,
+      "step": 7130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0148771892894298e-05,
+      "loss": 0.7393,
+      "step": 7131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0137804615667596e-05,
+      "loss": 0.7266,
+      "step": 7132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.012684259897197e-05,
+      "loss": 0.7441,
+      "step": 7133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0115885844254091e-05,
+      "loss": 0.7491,
+      "step": 7134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0104934352959883e-05,
+      "loss": 0.6931,
+      "step": 7135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0093988126534599e-05,
+      "loss": 0.7174,
+      "step": 7136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0083047166422777e-05,
+      "loss": 0.7453,
+      "step": 7137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.007211147406828e-05,
+      "loss": 0.6819,
+      "step": 7138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0061181050914265e-05,
+      "loss": 0.721,
+      "step": 7139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0050255898403205e-05,
+      "loss": 0.7351,
+      "step": 7140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0039336017976847e-05,
+      "loss": 0.7708,
+      "step": 7141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0028421411076293e-05,
+      "loss": 0.7011,
+      "step": 7142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.00175120791419e-05,
+      "loss": 0.7416,
+      "step": 7143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0006608023613378e-05,
+      "loss": 0.7365,
+      "step": 7144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.99570924592969e-06,
+      "loss": 0.7166,
+      "step": 7145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.98481574752913e-06,
+      "loss": 0.7891,
+      "step": 7146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.973927529849292e-06,
+      "loss": 0.6871,
+      "step": 7147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.963044594327064e-06,
+      "loss": 0.6351,
+      "step": 7148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.952166942398655e-06,
+      "loss": 0.6356,
+      "step": 7149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.941294575499561e-06,
+      "loss": 0.7047,
+      "step": 7150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.930427495064587e-06,
+      "loss": 0.7498,
+      "step": 7151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.919565702527834e-06,
+      "loss": 0.7572,
+      "step": 7152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.908709199322702e-06,
+      "loss": 0.8433,
+      "step": 7153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.897857986881925e-06,
+      "loss": 0.7526,
+      "step": 7154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.887012066637508e-06,
+      "loss": 0.8109,
+      "step": 7155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.876171440020754e-06,
+      "loss": 0.8074,
+      "step": 7156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.865336108462286e-06,
+      "loss": 0.7833,
+      "step": 7157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.854506073392017e-06,
+      "loss": 0.6648,
+      "step": 7158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.843681336239152e-06,
+      "loss": 0.7508,
+      "step": 7159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.832861898432238e-06,
+      "loss": 0.7713,
+      "step": 7160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.822047761399079e-06,
+      "loss": 0.7814,
+      "step": 7161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.811238926566774e-06,
+      "loss": 0.7762,
+      "step": 7162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.800435395361779e-06,
+      "loss": 0.7222,
+      "step": 7163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.789637169209791e-06,
+      "loss": 0.7618,
+      "step": 7164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.778844249535841e-06,
+      "loss": 0.8495,
+      "step": 7165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.768056637764239e-06,
+      "loss": 0.6698,
+      "step": 7166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.757274335318605e-06,
+      "loss": 0.8521,
+      "step": 7167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.746497343621857e-06,
+      "loss": 0.7463,
+      "step": 7168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.735725664096218e-06,
+      "loss": 0.726,
+      "step": 7169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.724959298163199e-06,
+      "loss": 0.7673,
+      "step": 7170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.71419824724361e-06,
+      "loss": 0.7041,
+      "step": 7171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.703442512757565e-06,
+      "loss": 0.7655,
+      "step": 7172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.69269209612449e-06,
+      "loss": 0.7008,
+      "step": 7173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.681946998763091e-06,
+      "loss": 0.7155,
+      "step": 7174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.671207222091367e-06,
+      "loss": 0.8323,
+      "step": 7175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.66047276752663e-06,
+      "loss": 0.7037,
+      "step": 7176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.64974363648548e-06,
+      "loss": 0.7096,
+      "step": 7177
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.63901983038381e-06,
+      "loss": 0.7295,
+      "step": 7178
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.628301350636842e-06,
+      "loss": 0.7264,
+      "step": 7179
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.617588198659055e-06,
+      "loss": 0.7322,
+      "step": 7180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.606880375864246e-06,
+      "loss": 0.7966,
+      "step": 7181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.596177883665491e-06,
+      "loss": 0.7163,
+      "step": 7182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.585480723475205e-06,
+      "loss": 0.7294,
+      "step": 7183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.574788896705045e-06,
+      "loss": 0.7717,
+      "step": 7184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.564102404766001e-06,
+      "loss": 0.7265,
+      "step": 7185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.553421249068339e-06,
+      "loss": 0.7968,
+      "step": 7186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.542745431021639e-06,
+      "loss": 0.6352,
+      "step": 7187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.532074952034753e-06,
+      "loss": 0.7825,
+      "step": 7188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.521409813515853e-06,
+      "loss": 0.7994,
+      "step": 7189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.510750016872388e-06,
+      "loss": 0.7966,
+      "step": 7190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.500095563511118e-06,
+      "loss": 0.7622,
+      "step": 7191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.489446454838068e-06,
+      "loss": 0.7211,
+      "step": 7192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.478802692258603e-06,
+      "loss": 0.7357,
+      "step": 7193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.468164277177354e-06,
+      "loss": 0.7555,
+      "step": 7194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.457531210998249e-06,
+      "loss": 0.7241,
+      "step": 7195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.446903495124492e-06,
+      "loss": 0.7722,
+      "step": 7196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.43628113095863e-06,
+      "loss": 0.7607,
+      "step": 7197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.42566411990246e-06,
+      "loss": 0.6916,
+      "step": 7198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.415052463357093e-06,
+      "loss": 0.7814,
+      "step": 7199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.404446162722924e-06,
+      "loss": 0.6718,
+      "step": 7200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.393845219399639e-06,
+      "loss": 0.6909,
+      "step": 7201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.383249634786212e-06,
+      "loss": 0.7851,
+      "step": 7202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.372659410280954e-06,
+      "loss": 0.7837,
+      "step": 7203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.362074547281413e-06,
+      "loss": 0.7222,
+      "step": 7204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.351495047184455e-06,
+      "loss": 0.7795,
+      "step": 7205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.34092091138623e-06,
+      "loss": 0.7355,
+      "step": 7206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.330352141282195e-06,
+      "loss": 0.7761,
+      "step": 7207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.319788738267077e-06,
+      "loss": 0.7709,
+      "step": 7208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.309230703734911e-06,
+      "loss": 0.7497,
+      "step": 7209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.298678039079023e-06,
+      "loss": 0.7683,
+      "step": 7210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.288130745692013e-06,
+      "loss": 0.7855,
+      "step": 7211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.277588824965806e-06,
+      "loss": 0.738,
+      "step": 7212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.267052278291582e-06,
+      "loss": 0.7685,
+      "step": 7213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.256521107059834e-06,
+      "loss": 0.7071,
+      "step": 7214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.245995312660327e-06,
+      "loss": 0.7657,
+      "step": 7215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.235474896482143e-06,
+      "loss": 0.7103,
+      "step": 7216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.22495985991364e-06,
+      "loss": 0.7034,
+      "step": 7217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.214450204342461e-06,
+      "loss": 0.7509,
+      "step": 7218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.203945931155538e-06,
+      "loss": 0.7932,
+      "step": 7219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.193447041739106e-06,
+      "loss": 0.7126,
+      "step": 7220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.182953537478661e-06,
+      "loss": 0.6784,
+      "step": 7221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.17246541975904e-06,
+      "loss": 0.7737,
+      "step": 7222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.161982689964327e-06,
+      "loss": 0.7685,
+      "step": 7223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.151505349477902e-06,
+      "loss": 0.754,
+      "step": 7224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.141033399682437e-06,
+      "loss": 0.6675,
+      "step": 7225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.130566841959898e-06,
+      "loss": 0.747,
+      "step": 7226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.120105677691526e-06,
+      "loss": 0.6903,
+      "step": 7227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.109649908257872e-06,
+      "loss": 0.6497,
+      "step": 7228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.099199535038749e-06,
+      "loss": 0.7443,
+      "step": 7229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.08875455941327e-06,
+      "loss": 0.7542,
+      "step": 7230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.078314982759856e-06,
+      "loss": 0.7469,
+      "step": 7231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.067880806456187e-06,
+      "loss": 0.8664,
+      "step": 7232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.057452031879216e-06,
+      "loss": 0.7928,
+      "step": 7233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.04702866040525e-06,
+      "loss": 0.7403,
+      "step": 7234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.036610693409809e-06,
+      "loss": 0.81,
+      "step": 7235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.026198132267743e-06,
+      "loss": 0.7627,
+      "step": 7236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.015790978353173e-06,
+      "loss": 0.7233,
+      "step": 7237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.005389233039508e-06,
+      "loss": 0.8013,
+      "step": 7238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.994992897699445e-06,
+      "loss": 0.6886,
+      "step": 7239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.984601973704964e-06,
+      "loss": 0.7438,
+      "step": 7240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.974216462427326e-06,
+      "loss": 0.7413,
+      "step": 7241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.963836365237111e-06,
+      "loss": 0.6369,
+      "step": 7242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.953461683504138e-06,
+      "loss": 0.7061,
+      "step": 7243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.943092418597542e-06,
+      "loss": 0.7275,
+      "step": 7244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.93272857188573e-06,
+      "loss": 0.7033,
+      "step": 7245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.922370144736391e-06,
+      "loss": 0.7262,
+      "step": 7246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.912017138516514e-06,
+      "loss": 0.7896,
+      "step": 7247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.901669554592362e-06,
+      "loss": 0.7603,
+      "step": 7248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.891327394329468e-06,
+      "loss": 0.8061,
+      "step": 7249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.880990659092697e-06,
+      "loss": 0.6752,
+      "step": 7250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.870659350246146e-06,
+      "loss": 0.6307,
+      "step": 7251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.860333469153209e-06,
+      "loss": 0.7442,
+      "step": 7252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.850013017176589e-06,
+      "loss": 0.6161,
+      "step": 7253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.839697995678254e-06,
+      "loss": 0.6964,
+      "step": 7254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.829388406019446e-06,
+      "loss": 0.7323,
+      "step": 7255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.819084249560705e-06,
+      "loss": 0.7682,
+      "step": 7256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.808785527661844e-06,
+      "loss": 0.7974,
+      "step": 7257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.798492241681971e-06,
+      "loss": 0.7049,
+      "step": 7258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.788204392979465e-06,
+      "loss": 0.7438,
+      "step": 7259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.777921982911996e-06,
+      "loss": 0.7758,
+      "step": 7260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.767645012836495e-06,
+      "loss": 0.7276,
+      "step": 7261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.757373484109216e-06,
+      "loss": 0.7326,
+      "step": 7262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.747107398085669e-06,
+      "loss": 0.7793,
+      "step": 7263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.736846756120637e-06,
+      "loss": 0.8276,
+      "step": 7264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.726591559568197e-06,
+      "loss": 0.7815,
+      "step": 7265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.716341809781714e-06,
+      "loss": 0.7112,
+      "step": 7266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.706097508113803e-06,
+      "loss": 0.7436,
+      "step": 7267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.695858655916416e-06,
+      "loss": 0.7573,
+      "step": 7268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.685625254540741e-06,
+      "loss": 0.7379,
+      "step": 7269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.675397305337241e-06,
+      "loss": 0.675,
+      "step": 7270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.665174809655708e-06,
+      "loss": 0.8004,
+      "step": 7271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.654957768845162e-06,
+      "loss": 0.671,
+      "step": 7272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.644746184253938e-06,
+      "loss": 0.6846,
+      "step": 7273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.634540057229623e-06,
+      "loss": 0.6275,
+      "step": 7274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.624339389119107e-06,
+      "loss": 0.8288,
+      "step": 7275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.614144181268552e-06,
+      "loss": 0.7751,
+      "step": 7276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.603954435023393e-06,
+      "loss": 0.7511,
+      "step": 7277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.593770151728353e-06,
+      "loss": 0.7278,
+      "step": 7278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.583591332727426e-06,
+      "loss": 0.8533,
+      "step": 7279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.573417979363885e-06,
+      "loss": 0.6996,
+      "step": 7280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.563250092980307e-06,
+      "loss": 0.7814,
+      "step": 7281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.553087674918514e-06,
+      "loss": 0.7891,
+      "step": 7282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.542930726519622e-06,
+      "loss": 0.7343,
+      "step": 7283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.532779249124018e-06,
+      "loss": 0.7531,
+      "step": 7284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.522633244071375e-06,
+      "loss": 0.732,
+      "step": 7285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.512492712700627e-06,
+      "loss": 0.7421,
+      "step": 7286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.502357656350024e-06,
+      "loss": 0.7639,
+      "step": 7287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.49222807635705e-06,
+      "loss": 0.726,
+      "step": 7288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.482103974058492e-06,
+      "loss": 0.775,
+      "step": 7289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.471985350790396e-06,
+      "loss": 0.7265,
+      "step": 7290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.46187220788811e-06,
+      "loss": 0.7679,
+      "step": 7291
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.451764546686248e-06,
+      "loss": 0.7619,
+      "step": 7292
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.441662368518677e-06,
+      "loss": 0.7429,
+      "step": 7293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.431565674718578e-06,
+      "loss": 0.8014,
+      "step": 7294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.421474466618378e-06,
+      "loss": 0.749,
+      "step": 7295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.411388745549804e-06,
+      "loss": 0.7513,
+      "step": 7296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.401308512843842e-06,
+      "loss": 0.7226,
+      "step": 7297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.391233769830758e-06,
+      "loss": 0.7981,
+      "step": 7298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.381164517840096e-06,
+      "loss": 0.8115,
+      "step": 7299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.371100758200662e-06,
+      "loss": 0.8226,
+      "step": 7300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.361042492240573e-06,
+      "loss": 0.6373,
+      "step": 7301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.35098972128719e-06,
+      "loss": 0.7488,
+      "step": 7302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.340942446667155e-06,
+      "loss": 0.7448,
+      "step": 7303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.330900669706364e-06,
+      "loss": 0.8119,
+      "step": 7304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.32086439173005e-06,
+      "loss": 0.7531,
+      "step": 7305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.310833614062651e-06,
+      "loss": 0.7045,
+      "step": 7306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.30080833802792e-06,
+      "loss": 0.7382,
+      "step": 7307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.29078856494887e-06,
+      "loss": 0.746,
+      "step": 7308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.280774296147786e-06,
+      "loss": 0.7706,
+      "step": 7309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.270765532946217e-06,
+      "loss": 0.7112,
+      "step": 7310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.26076227666503e-06,
+      "loss": 0.7713,
+      "step": 7311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.250764528624311e-06,
+      "loss": 0.7497,
+      "step": 7312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.240772290143451e-06,
+      "loss": 0.7147,
+      "step": 7313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.230785562541104e-06,
+      "loss": 0.7438,
+      "step": 7314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.220804347135197e-06,
+      "loss": 0.7402,
+      "step": 7315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.210828645242924e-06,
+      "loss": 0.8388,
+      "step": 7316
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.200858458180772e-06,
+      "loss": 0.7919,
+      "step": 7317
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.19089378726447e-06,
+      "loss": 0.8185,
+      "step": 7318
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.18093463380904e-06,
+      "loss": 0.8058,
+      "step": 7319
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.170980999128764e-06,
+      "loss": 0.7128,
+      "step": 7320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.161032884537223e-06,
+      "loss": 0.7274,
+      "step": 7321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.151090291347235e-06,
+      "loss": 0.6846,
+      "step": 7322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.141153220870896e-06,
+      "loss": 0.7861,
+      "step": 7323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.131221674419598e-06,
+      "loss": 0.8093,
+      "step": 7324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.121295653303984e-06,
+      "loss": 0.7608,
+      "step": 7325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.111375158833961e-06,
+      "loss": 0.568,
+      "step": 7326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.101460192318721e-06,
+      "loss": 0.7809,
+      "step": 7327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.091550755066718e-06,
+      "loss": 0.7812,
+      "step": 7328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.08164684838567e-06,
+      "loss": 0.6847,
+      "step": 7329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.071748473582603e-06,
+      "loss": 0.6387,
+      "step": 7330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.061855631963765e-06,
+      "loss": 0.7722,
+      "step": 7331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.051968324834702e-06,
+      "loss": 0.8291,
+      "step": 7332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.042086553500217e-06,
+      "loss": 0.8242,
+      "step": 7333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.032210319264388e-06,
+      "loss": 0.7805,
+      "step": 7334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.022339623430558e-06,
+      "loss": 0.6918,
+      "step": 7335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.01247446730134e-06,
+      "loss": 0.6773,
+      "step": 7336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.00261485217863e-06,
+      "loss": 0.7292,
+      "step": 7337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.992760779363556e-06,
+      "loss": 0.8199,
+      "step": 7338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.982912250156572e-06,
+      "loss": 0.7241,
+      "step": 7339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.973069265857352e-06,
+      "loss": 0.645,
+      "step": 7340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.963231827764845e-06,
+      "loss": 0.7935,
+      "step": 7341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.953399937177297e-06,
+      "loss": 0.812,
+      "step": 7342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.943573595392195e-06,
+      "loss": 0.7738,
+      "step": 7343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.933752803706301e-06,
+      "loss": 0.6939,
+      "step": 7344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.923937563415634e-06,
+      "loss": 0.6734,
+      "step": 7345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.914127875815508e-06,
+      "loss": 0.7283,
+      "step": 7346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.90432374220047e-06,
+      "loss": 0.7298,
+      "step": 7347
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.894525163864364e-06,
+      "loss": 0.7443,
+      "step": 7348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.88473214210027e-06,
+      "loss": 0.7867,
+      "step": 7349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.874944678200573e-06,
+      "loss": 0.8183,
+      "step": 7350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.865162773456902e-06,
+      "loss": 0.686,
+      "step": 7351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.85538642916015e-06,
+      "loss": 0.866,
+      "step": 7352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.845615646600479e-06,
+      "loss": 0.7464,
+      "step": 7353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.835850427067321e-06,
+      "loss": 0.7437,
+      "step": 7354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.82609077184937e-06,
+      "loss": 0.7756,
+      "step": 7355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.816336682234593e-06,
+      "loss": 0.7194,
+      "step": 7356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.806588159510198e-06,
+      "loss": 0.6938,
+      "step": 7357
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.796845204962704e-06,
+      "loss": 0.7122,
+      "step": 7358
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.787107819877853e-06,
+      "loss": 0.8316,
+      "step": 7359
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.777376005540665e-06,
+      "loss": 0.6852,
+      "step": 7360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.76764976323544e-06,
+      "loss": 0.7945,
+      "step": 7361
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.757929094245726e-06,
+      "loss": 0.666,
+      "step": 7362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.748213999854332e-06,
+      "loss": 0.7621,
+      "step": 7363
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.738504481343345e-06,
+      "loss": 0.716,
+      "step": 7364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.728800539994113e-06,
+      "loss": 0.7463,
+      "step": 7365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.71910217708724e-06,
+      "loss": 0.8014,
+      "step": 7366
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.709409393902595e-06,
+      "loss": 0.8007,
+      "step": 7367
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.699722191719316e-06,
+      "loss": 0.7488,
+      "step": 7368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.690040571815799e-06,
+      "loss": 0.7461,
+      "step": 7369
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.680364535469725e-06,
+      "loss": 0.7373,
+      "step": 7370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.670694083958007e-06,
+      "loss": 0.746,
+      "step": 7371
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.661029218556843e-06,
+      "loss": 0.7246,
+      "step": 7372
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.651369940541676e-06,
+      "loss": 0.6547,
+      "step": 7373
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.64171625118722e-06,
+      "loss": 0.6896,
+      "step": 7374
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.632068151767447e-06,
+      "loss": 0.685,
+      "step": 7375
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.622425643555614e-06,
+      "loss": 0.6921,
+      "step": 7376
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.612788727824216e-06,
+      "loss": 0.7793,
+      "step": 7377
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.6031574058450186e-06,
+      "loss": 0.7761,
+      "step": 7378
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5935316788890265e-06,
+      "loss": 0.6769,
+      "step": 7379
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.583911548226563e-06,
+      "loss": 0.8245,
+      "step": 7380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.574297015127152e-06,
+      "loss": 0.663,
+      "step": 7381
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.564688080859611e-06,
+      "loss": 0.7667,
+      "step": 7382
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.555084746692009e-06,
+      "loss": 0.6331,
+      "step": 7383
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.545487013891684e-06,
+      "loss": 0.7212,
+      "step": 7384
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.53589488372522e-06,
+      "loss": 0.7302,
+      "step": 7385
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.526308357458473e-06,
+      "loss": 0.7359,
+      "step": 7386
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.516727436356557e-06,
+      "loss": 0.6507,
+      "step": 7387
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.507152121683836e-06,
+      "loss": 0.6947,
+      "step": 7388
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.497582414703968e-06,
+      "loss": 0.7245,
+      "step": 7389
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.488018316679835e-06,
+      "loss": 0.7239,
+      "step": 7390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.478459828873591e-06,
+      "loss": 0.6878,
+      "step": 7391
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.46890695254665e-06,
+      "loss": 0.7764,
+      "step": 7392
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.459359688959689e-06,
+      "loss": 0.6819,
+      "step": 7393
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.449818039372619e-06,
+      "loss": 0.7164,
+      "step": 7394
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.440282005044663e-06,
+      "loss": 0.7407,
+      "step": 7395
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.430751587234258e-06,
+      "loss": 0.731,
+      "step": 7396
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.421226787199115e-06,
+      "loss": 0.7242,
+      "step": 7397
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4117076061961885e-06,
+      "loss": 0.7974,
+      "step": 7398
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.402194045481725e-06,
+      "loss": 0.7367,
+      "step": 7399
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.392686106311203e-06,
+      "loss": 0.7833,
+      "step": 7400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.383183789939363e-06,
+      "loss": 0.7656,
+      "step": 7401
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.373687097620208e-06,
+      "loss": 0.8026,
+      "step": 7402
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.364196030606996e-06,
+      "loss": 0.8137,
+      "step": 7403
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.354710590152242e-06,
+      "loss": 0.7221,
+      "step": 7404
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.345230777507722e-06,
+      "loss": 0.7058,
+      "step": 7405
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.335756593924464e-06,
+      "loss": 0.6004,
+      "step": 7406
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.326288040652762e-06,
+      "loss": 0.6732,
+      "step": 7407
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.316825118942144e-06,
+      "loss": 0.6723,
+      "step": 7408
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.3073678300414336e-06,
+      "loss": 0.8056,
+      "step": 7409
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.297916175198688e-06,
+      "loss": 0.7562,
+      "step": 7410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.288470155661209e-06,
+      "loss": 0.6754,
+      "step": 7411
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.2790297726755716e-06,
+      "loss": 0.7728,
+      "step": 7412
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.269595027487614e-06,
+      "loss": 0.7161,
+      "step": 7413
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.2601659213424104e-06,
+      "loss": 0.7302,
+      "step": 7414
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.250742455484305e-06,
+      "loss": 0.6797,
+      "step": 7415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.241324631156893e-06,
+      "loss": 0.7316,
+      "step": 7416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.231912449603023e-06,
+      "loss": 0.7432,
+      "step": 7417
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.222505912064792e-06,
+      "loss": 0.8095,
+      "step": 7418
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.2131050197835784e-06,
+      "loss": 0.7041,
+      "step": 7419
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.203709773999989e-06,
+      "loss": 0.6771,
+      "step": 7420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1943201759539005e-06,
+      "loss": 0.7771,
+      "step": 7421
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1849362268844315e-06,
+      "loss": 0.6962,
+      "step": 7422
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.175557928029963e-06,
+      "loss": 0.7055,
+      "step": 7423
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.166185280628135e-06,
+      "loss": 0.7203,
+      "step": 7424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.156818285915834e-06,
+      "loss": 0.753,
+      "step": 7425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.147456945129199e-06,
+      "loss": 0.7049,
+      "step": 7426
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.138101259503632e-06,
+      "loss": 0.785,
+      "step": 7427
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.128751230273773e-06,
+      "loss": 0.6809,
+      "step": 7428
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.119406858673539e-06,
+      "loss": 0.7091,
+      "step": 7429
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.110068145936083e-06,
+      "loss": 0.7098,
+      "step": 7430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.100735093293809e-06,
+      "loss": 0.7341,
+      "step": 7431
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.091407701978392e-06,
+      "loss": 0.6965,
+      "step": 7432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.082085973220742e-06,
+      "loss": 0.7272,
+      "step": 7433
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.072769908251031e-06,
+      "loss": 0.633,
+      "step": 7434
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.063459508298681e-06,
+      "loss": 0.8117,
+      "step": 7435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.054154774592364e-06,
+      "loss": 0.8222,
+      "step": 7436
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.044855708359993e-06,
+      "loss": 0.8336,
+      "step": 7437
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.035562310828769e-06,
+      "loss": 0.7022,
+      "step": 7438
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.026274583225112e-06,
+      "loss": 0.7698,
+      "step": 7439
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0169925267747125e-06,
+      "loss": 0.738,
+      "step": 7440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.007716142702492e-06,
+      "loss": 0.6166,
+      "step": 7441
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.998445432232636e-06,
+      "loss": 0.7718,
+      "step": 7442
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.989180396588591e-06,
+      "loss": 0.6821,
+      "step": 7443
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.979921036993042e-06,
+      "loss": 0.5524,
+      "step": 7444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.97066735466792e-06,
+      "loss": 0.7636,
+      "step": 7445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.96141935083442e-06,
+      "loss": 0.7516,
+      "step": 7446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.952177026712964e-06,
+      "loss": 0.7104,
+      "step": 7447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.942940383523277e-06,
+      "loss": 0.7099,
+      "step": 7448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.933709422484264e-06,
+      "loss": 0.7408,
+      "step": 7449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.924484144814147e-06,
+      "loss": 0.6765,
+      "step": 7450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9152645517303545e-06,
+      "loss": 0.715,
+      "step": 7451
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9060506444495745e-06,
+      "loss": 0.7344,
+      "step": 7452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8968424241877494e-06,
+      "loss": 0.6952,
+      "step": 7453
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.887639892160064e-06,
+      "loss": 0.7216,
+      "step": 7454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.878443049580963e-06,
+      "loss": 0.8191,
+      "step": 7455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8692518976641305e-06,
+      "loss": 0.7619,
+      "step": 7456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.860066437622498e-06,
+      "loss": 0.709,
+      "step": 7457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.850886670668266e-06,
+      "loss": 0.7128,
+      "step": 7458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8417125980128675e-06,
+      "loss": 0.7334,
+      "step": 7459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8325442208669765e-06,
+      "loss": 0.7141,
+      "step": 7460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.823381540440532e-06,
+      "loss": 0.7278,
+      "step": 7461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.814224557942705e-06,
+      "loss": 0.7535,
+      "step": 7462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.805073274581936e-06,
+      "loss": 0.7335,
+      "step": 7463
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.795927691565895e-06,
+      "loss": 0.7041,
+      "step": 7464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.786787810101491e-06,
+      "loss": 0.6872,
+      "step": 7465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.77765363139492e-06,
+      "loss": 0.7979,
+      "step": 7466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.768525156651589e-06,
+      "loss": 0.7733,
+      "step": 7467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.759402387076153e-06,
+      "loss": 0.9233,
+      "step": 7468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.750285323872546e-06,
+      "loss": 0.735,
+      "step": 7469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.741173968243919e-06,
+      "loss": 0.8233,
+      "step": 7470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.732068321392682e-06,
+      "loss": 0.7451,
+      "step": 7471
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7229683845204835e-06,
+      "loss": 0.722,
+      "step": 7472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.71387415882822e-06,
+      "loss": 0.7521,
+      "step": 7473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.704785645516043e-06,
+      "loss": 0.8296,
+      "step": 7474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.695702845783341e-06,
+      "loss": 0.5941,
+      "step": 7475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.686625760828763e-06,
+      "loss": 0.8027,
+      "step": 7476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.677554391850166e-06,
+      "loss": 0.7971,
+      "step": 7477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.668488740044709e-06,
+      "loss": 0.7215,
+      "step": 7478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.659428806608758e-06,
+      "loss": 0.7393,
+      "step": 7479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.650374592737934e-06,
+      "loss": 0.7703,
+      "step": 7480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6413260996270985e-06,
+      "loss": 0.7279,
+      "step": 7481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.632283328470368e-06,
+      "loss": 0.7674,
+      "step": 7482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.623246280461076e-06,
+      "loss": 0.7748,
+      "step": 7483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.614214956791859e-06,
+      "loss": 0.7608,
+      "step": 7484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.605189358654545e-06,
+      "loss": 0.7085,
+      "step": 7485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.596169487240222e-06,
+      "loss": 0.8492,
+      "step": 7486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5871553437392155e-06,
+      "loss": 0.7995,
+      "step": 7487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.578146929341122e-06,
+      "loss": 0.7412,
+      "step": 7488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.569144245234754e-06,
+      "loss": 0.7844,
+      "step": 7489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5601472926081766e-06,
+      "loss": 0.7177,
+      "step": 7490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.551156072648701e-06,
+      "loss": 0.7764,
+      "step": 7491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.542170586542873e-06,
+      "loss": 0.6763,
+      "step": 7492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5331908354764995e-06,
+      "loss": 0.7481,
+      "step": 7493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5242168206346095e-06,
+      "loss": 0.7866,
+      "step": 7494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.515248543201497e-06,
+      "loss": 0.7364,
+      "step": 7495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.506286004360662e-06,
+      "loss": 0.7887,
+      "step": 7496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4973292052949e-06,
+      "loss": 0.7043,
+      "step": 7497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4883781471862134e-06,
+      "loss": 0.7612,
+      "step": 7498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.479432831215848e-06,
+      "loss": 0.6922,
+      "step": 7499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.470493258564309e-06,
+      "loss": 0.8199,
+      "step": 7500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.461559430411324e-06,
+      "loss": 0.5795,
+      "step": 7501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4526313479358615e-06,
+      "loss": 0.6622,
+      "step": 7502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4437090123161626e-06,
+      "loss": 0.7542,
+      "step": 7503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.434792424729691e-06,
+      "loss": 0.7846,
+      "step": 7504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.425881586353133e-06,
+      "loss": 0.7042,
+      "step": 7505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.416976498362432e-06,
+      "loss": 0.7899,
+      "step": 7506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.408077161932796e-06,
+      "loss": 0.671,
+      "step": 7507
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.399183578238638e-06,
+      "loss": 0.7049,
+      "step": 7508
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.390295748453628e-06,
+      "loss": 0.6335,
+      "step": 7509
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.381413673750675e-06,
+      "loss": 0.692,
+      "step": 7510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.372537355301927e-06,
+      "loss": 0.6965,
+      "step": 7511
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.363666794278772e-06,
+      "loss": 0.6912,
+      "step": 7512
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.354801991851839e-06,
+      "loss": 0.7152,
+      "step": 7513
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.345942949190997e-06,
+      "loss": 0.658,
+      "step": 7514
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.337089667465357e-06,
+      "loss": 0.7704,
+      "step": 7515
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.328242147843261e-06,
+      "loss": 0.6754,
+      "step": 7516
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.319400391492314e-06,
+      "loss": 0.7535,
+      "step": 7517
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.310564399579339e-06,
+      "loss": 0.7536,
+      "step": 7518
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.301734173270396e-06,
+      "loss": 0.7225,
+      "step": 7519
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.292909713730782e-06,
+      "loss": 0.7911,
+      "step": 7520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2840910221250716e-06,
+      "loss": 0.786,
+      "step": 7521
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.275278099617032e-06,
+      "loss": 0.721,
+      "step": 7522
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.266470947369691e-06,
+      "loss": 0.801,
+      "step": 7523
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.257669566545305e-06,
+      "loss": 0.7143,
+      "step": 7524
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.248873958305373e-06,
+      "loss": 0.7385,
+      "step": 7525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.24008412381063e-06,
+      "loss": 0.8013,
+      "step": 7526
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.231300064221068e-06,
+      "loss": 0.7468,
+      "step": 7527
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.222521780695894e-06,
+      "loss": 0.7624,
+      "step": 7528
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2137492743935545e-06,
+      "loss": 0.7443,
+      "step": 7529
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.204982546471744e-06,
+      "loss": 0.7419,
+      "step": 7530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.1962215980873815e-06,
+      "loss": 0.6484,
+      "step": 7531
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.1874664303966435e-06,
+      "loss": 0.6656,
+      "step": 7532
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.178717044554921e-06,
+      "loss": 0.794,
+      "step": 7533
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.169973441716858e-06,
+      "loss": 0.7725,
+      "step": 7534
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.161235623036326e-06,
+      "loss": 0.6824,
+      "step": 7535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.152503589666425e-06,
+      "loss": 0.7324,
+      "step": 7536
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1437773427595325e-06,
+      "loss": 0.7748,
+      "step": 7537
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.135056883467216e-06,
+      "loss": 0.793,
+      "step": 7538
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.126342212940284e-06,
+      "loss": 0.7635,
+      "step": 7539
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.117633332328821e-06,
+      "loss": 0.7695,
+      "step": 7540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.108930242782101e-06,
+      "loss": 0.7237,
+      "step": 7541
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.100232945448664e-06,
+      "loss": 0.7299,
+      "step": 7542
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.091541441476267e-06,
+      "loss": 0.7172,
+      "step": 7543
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.082855732011905e-06,
+      "loss": 0.7021,
+      "step": 7544
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.074175818201821e-06,
+      "loss": 0.8088,
+      "step": 7545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0655017011914785e-06,
+      "loss": 0.7412,
+      "step": 7546
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.056833382125593e-06,
+      "loss": 0.6976,
+      "step": 7547
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.048170862148095e-06,
+      "loss": 0.7704,
+      "step": 7548
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.03951414240217e-06,
+      "loss": 0.7821,
+      "step": 7549
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.030863224030214e-06,
+      "loss": 0.7671,
+      "step": 7550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.022218108173877e-06,
+      "loss": 0.7114,
+      "step": 7551
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.013578795974039e-06,
+      "loss": 0.6544,
+      "step": 7552
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.004945288570813e-06,
+      "loss": 0.8113,
+      "step": 7553
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.99631758710354e-06,
+      "loss": 0.6461,
+      "step": 7554
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.987695692710787e-06,
+      "loss": 0.7625,
+      "step": 7555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.979079606530397e-06,
+      "loss": 0.7153,
+      "step": 7556
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.970469329699391e-06,
+      "loss": 0.7591,
+      "step": 7557
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.961864863354072e-06,
+      "loss": 0.7888,
+      "step": 7558
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.953266208629943e-06,
+      "loss": 0.7133,
+      "step": 7559
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.944673366661746e-06,
+      "loss": 0.7002,
+      "step": 7560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9360863385834685e-06,
+      "loss": 0.5764,
+      "step": 7561
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.927505125528315e-06,
+      "loss": 0.8702,
+      "step": 7562
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.918929728628736e-06,
+      "loss": 0.8051,
+      "step": 7563
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.910360149016403e-06,
+      "loss": 0.8371,
+      "step": 7564
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.901796387822223e-06,
+      "loss": 0.6812,
+      "step": 7565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.893238446176352e-06,
+      "loss": 0.7611,
+      "step": 7566
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.884686325208161e-06,
+      "loss": 0.7041,
+      "step": 7567
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.876140026046245e-06,
+      "loss": 0.781,
+      "step": 7568
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.867599549818448e-06,
+      "loss": 0.7378,
+      "step": 7569
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.859064897651839e-06,
+      "loss": 0.7224,
+      "step": 7570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.850536070672713e-06,
+      "loss": 0.6975,
+      "step": 7571
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.842013070006613e-06,
+      "loss": 0.7586,
+      "step": 7572
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.833495896778279e-06,
+      "loss": 0.6826,
+      "step": 7573
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.824984552111729e-06,
+      "loss": 0.7634,
+      "step": 7574
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.816479037130185e-06,
+      "loss": 0.7372,
+      "step": 7575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.807979352956078e-06,
+      "loss": 0.7301,
+      "step": 7576
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.799485500711122e-06,
+      "loss": 0.7405,
+      "step": 7577
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.790997481516225e-06,
+      "loss": 0.767,
+      "step": 7578
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.78251529649153e-06,
+      "loss": 0.8402,
+      "step": 7579
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.774038946756416e-06,
+      "loss": 0.8093,
+      "step": 7580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.765568433429486e-06,
+      "loss": 0.8049,
+      "step": 7581
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.757103757628573e-06,
+      "loss": 0.772,
+      "step": 7582
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.748644920470753e-06,
+      "loss": 0.7349,
+      "step": 7583
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.740191923072313e-06,
+      "loss": 0.8429,
+      "step": 7584
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.731744766548769e-06,
+      "loss": 0.787,
+      "step": 7585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.723303452014894e-06,
+      "loss": 0.7805,
+      "step": 7586
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.71486798058466e-06,
+      "loss": 0.7189,
+      "step": 7587
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.706438353371285e-06,
+      "loss": 0.6715,
+      "step": 7588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.698014571487203e-06,
+      "loss": 0.6186,
+      "step": 7589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.689596636044087e-06,
+      "loss": 0.7916,
+      "step": 7590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6811845481528295e-06,
+      "loss": 0.6828,
+      "step": 7591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.672778308923549e-06,
+      "loss": 0.7238,
+      "step": 7592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.664377919465619e-06,
+      "loss": 0.7182,
+      "step": 7593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.655983380887614e-06,
+      "loss": 0.6464,
+      "step": 7594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.647594694297331e-06,
+      "loss": 0.7268,
+      "step": 7595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.63921186080183e-06,
+      "loss": 0.8151,
+      "step": 7596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.630834881507363e-06,
+      "loss": 0.7537,
+      "step": 7597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.622463757519425e-06,
+      "loss": 0.7458,
+      "step": 7598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.614098489942738e-06,
+      "loss": 0.8296,
+      "step": 7599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.605739079881239e-06,
+      "loss": 0.7001,
+      "step": 7600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.597385528438115e-06,
+      "loss": 0.5571,
+      "step": 7601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5890378367157604e-06,
+      "loss": 0.7502,
+      "step": 7602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5806960058158e-06,
+      "loss": 0.7914,
+      "step": 7603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.572360036839092e-06,
+      "loss": 0.7783,
+      "step": 7604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5640299308857e-06,
+      "loss": 0.6712,
+      "step": 7605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.555705689054963e-06,
+      "loss": 0.7658,
+      "step": 7606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5473873124453955e-06,
+      "loss": 0.6717,
+      "step": 7607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.539074802154759e-06,
+      "loss": 0.6773,
+      "step": 7608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.530768159280042e-06,
+      "loss": 0.7457,
+      "step": 7609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5224673849174344e-06,
+      "loss": 0.7664,
+      "step": 7610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5141724801624e-06,
+      "loss": 0.7266,
+      "step": 7611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.505883446109589e-06,
+      "loss": 0.8203,
+      "step": 7612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4976002838528825e-06,
+      "loss": 0.7172,
+      "step": 7613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.489322994485396e-06,
+      "loss": 0.8079,
+      "step": 7614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.481051579099472e-06,
+      "loss": 0.7398,
+      "step": 7615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.472786038786665e-06,
+      "loss": 0.7398,
+      "step": 7616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4645263746377685e-06,
+      "loss": 0.7609,
+      "step": 7617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.456272587742783e-06,
+      "loss": 0.7324,
+      "step": 7618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.448024679190955e-06,
+      "loss": 0.6783,
+      "step": 7619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.439782650070729e-06,
+      "loss": 0.7387,
+      "step": 7620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.431546501469803e-06,
+      "loss": 0.7263,
+      "step": 7621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.423316234475078e-06,
+      "loss": 0.792,
+      "step": 7622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.415091850172688e-06,
+      "loss": 0.7025,
+      "step": 7623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.406873349647973e-06,
+      "loss": 0.7521,
+      "step": 7624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.398660733985533e-06,
+      "loss": 0.7828,
+      "step": 7625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.390454004269163e-06,
+      "loss": 0.8043,
+      "step": 7626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.382253161581885e-06,
+      "loss": 0.6875,
+      "step": 7627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.374058207005944e-06,
+      "loss": 0.6999,
+      "step": 7628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.365869141622809e-06,
+      "loss": 0.7651,
+      "step": 7629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.357685966513193e-06,
+      "loss": 0.7756,
+      "step": 7630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3495086827569986e-06,
+      "loss": 0.7825,
+      "step": 7631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.341337291433362e-06,
+      "loss": 0.7699,
+      "step": 7632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.33317179362065e-06,
+      "loss": 0.6796,
+      "step": 7633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.325012190396439e-06,
+      "loss": 0.7312,
+      "step": 7634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.31685848283755e-06,
+      "loss": 0.7065,
+      "step": 7635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.30871067202e-06,
+      "loss": 0.6169,
+      "step": 7636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3005687590190426e-06,
+      "loss": 0.7033,
+      "step": 7637
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.292432744909143e-06,
+      "loss": 0.6715,
+      "step": 7638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.284302630764004e-06,
+      "loss": 0.7351,
+      "step": 7639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.27617841765653e-06,
+      "loss": 0.7138,
+      "step": 7640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2680601066588615e-06,
+      "loss": 0.7656,
+      "step": 7641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.259947698842354e-06,
+      "loss": 0.7471,
+      "step": 7642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.25184119527759e-06,
+      "loss": 0.7589,
+      "step": 7643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.243740597034346e-06,
+      "loss": 0.7202,
+      "step": 7644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2356459051816745e-06,
+      "loss": 0.7284,
+      "step": 7645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.227557120787796e-06,
+      "loss": 0.727,
+      "step": 7646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.219474244920164e-06,
+      "loss": 0.8086,
+      "step": 7647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.21139727864548e-06,
+      "loss": 0.6454,
+      "step": 7648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.203326223029631e-06,
+      "loss": 0.8071,
+      "step": 7649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.195261079137742e-06,
+      "loss": 0.6848,
+      "step": 7650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.187201848034146e-06,
+      "loss": 0.6998,
+      "step": 7651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1791485307824126e-06,
+      "loss": 0.7149,
+      "step": 7652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.171101128445316e-06,
+      "loss": 0.6576,
+      "step": 7653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.163059642084844e-06,
+      "loss": 0.7973,
+      "step": 7654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.155024072762238e-06,
+      "loss": 0.6705,
+      "step": 7655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.146994421537921e-06,
+      "loss": 0.7663,
+      "step": 7656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.138970689471551e-06,
+      "loss": 0.6719,
+      "step": 7657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.130952877622008e-06,
+      "loss": 0.7499,
+      "step": 7658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.122940987047381e-06,
+      "loss": 0.7857,
+      "step": 7659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.114935018804984e-06,
+      "loss": 0.755,
+      "step": 7660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.106934973951349e-06,
+      "loss": 0.7236,
+      "step": 7661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0989408535422225e-06,
+      "loss": 0.6413,
+      "step": 7662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.090952658632564e-06,
+      "loss": 0.7999,
+      "step": 7663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0829703902765815e-06,
+      "loss": 0.6907,
+      "step": 7664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.074994049527654e-06,
+      "loss": 0.6999,
+      "step": 7665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.067023637438423e-06,
+      "loss": 0.927,
+      "step": 7666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.059059155060719e-06,
+      "loss": 0.673,
+      "step": 7667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0511006034455965e-06,
+      "loss": 0.7127,
+      "step": 7668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.043147983643332e-06,
+      "loss": 0.7083,
+      "step": 7669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.035201296703418e-06,
+      "loss": 0.7292,
+      "step": 7670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.02726054367455e-06,
+      "loss": 0.8204,
+      "step": 7671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.019325725604668e-06,
+      "loss": 0.6807,
+      "step": 7672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0113968435408946e-06,
+      "loss": 0.7768,
+      "step": 7673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0034738985296095e-06,
+      "loss": 0.7388,
+      "step": 7674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.995556891616376e-06,
+      "loss": 0.6911,
+      "step": 7675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.987645823845993e-06,
+      "loss": 0.796,
+      "step": 7676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9797406962624525e-06,
+      "loss": 0.769,
+      "step": 7677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.971841509908992e-06,
+      "loss": 0.7331,
+      "step": 7678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.96394826582805e-06,
+      "loss": 0.6936,
+      "step": 7679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9560609650612714e-06,
+      "loss": 0.7999,
+      "step": 7680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.948179608649523e-06,
+      "loss": 0.837,
+      "step": 7681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.940304197632911e-06,
+      "loss": 0.7902,
+      "step": 7682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.932434733050728e-06,
+      "loss": 0.7008,
+      "step": 7683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.924571215941476e-06,
+      "loss": 0.7508,
+      "step": 7684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.916713647342914e-06,
+      "loss": 0.7701,
+      "step": 7685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.908862028291972e-06,
+      "loss": 0.702,
+      "step": 7686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.901016359824817e-06,
+      "loss": 0.6998,
+      "step": 7687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.893176642976827e-06,
+      "loss": 0.7808,
+      "step": 7688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.885342878782584e-06,
+      "loss": 0.7165,
+      "step": 7689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.877515068275906e-06,
+      "loss": 0.7418,
+      "step": 7690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.869693212489801e-06,
+      "loss": 0.6592,
+      "step": 7691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.861877312456509e-06,
+      "loss": 0.704,
+      "step": 7692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.854067369207471e-06,
+      "loss": 0.7651,
+      "step": 7693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.846263383773364e-06,
+      "loss": 0.7961,
+      "step": 7694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.838465357184058e-06,
+      "loss": 0.7939,
+      "step": 7695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.830673290468635e-06,
+      "loss": 0.8265,
+      "step": 7696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.822887184655405e-06,
+      "loss": 0.7305,
+      "step": 7697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8151070407718825e-06,
+      "loss": 0.7368,
+      "step": 7698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.807332859844799e-06,
+      "loss": 0.7739,
+      "step": 7699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.799564642900078e-06,
+      "loss": 0.7343,
+      "step": 7700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.791802390962902e-06,
+      "loss": 0.6884,
+      "step": 7701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.784046105057632e-06,
+      "loss": 0.8089,
+      "step": 7702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.776295786207835e-06,
+      "loss": 0.7162,
+      "step": 7703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.768551435436325e-06,
+      "loss": 0.744,
+      "step": 7704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.760813053765095e-06,
+      "loss": 0.7334,
+      "step": 7705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7530806422153684e-06,
+      "loss": 0.8215,
+      "step": 7706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.745354201807573e-06,
+      "loss": 0.8026,
+      "step": 7707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737633733561348e-06,
+      "loss": 0.6973,
+      "step": 7708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.729919238495551e-06,
+      "loss": 0.7393,
+      "step": 7709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.72221071762825e-06,
+      "loss": 0.837,
+      "step": 7710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.714508171976717e-06,
+      "loss": 0.7168,
+      "step": 7711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.706811602557448e-06,
+      "loss": 0.7352,
+      "step": 7712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.699121010386126e-06,
+      "loss": 0.8221,
+      "step": 7713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6914363964776884e-06,
+      "loss": 0.7585,
+      "step": 7714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.683757761846241e-06,
+      "loss": 0.7465,
+      "step": 7715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.676085107505124e-06,
+      "loss": 0.7464,
+      "step": 7716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.668418434466876e-06,
+      "loss": 0.8136,
+      "step": 7717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6607577437432494e-06,
+      "loss": 0.752,
+      "step": 7718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.653103036345224e-06,
+      "loss": 0.7642,
+      "step": 7719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.645454313282965e-06,
+      "loss": 0.7186,
+      "step": 7720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.637811575565859e-06,
+      "loss": 0.7451,
+      "step": 7721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.630174824202499e-06,
+      "loss": 0.7446,
+      "step": 7722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.622544060200701e-06,
+      "loss": 0.7794,
+      "step": 7723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.614919284567476e-06,
+      "loss": 0.7508,
+      "step": 7724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.607300498309047e-06,
+      "loss": 0.736,
+      "step": 7725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.599687702430855e-06,
+      "loss": 0.7158,
+      "step": 7726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.592080897937539e-06,
+      "loss": 0.739,
+      "step": 7727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.58448008583296e-06,
+      "loss": 0.8007,
+      "step": 7728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.576885267120173e-06,
+      "loss": 0.6838,
+      "step": 7729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5692964428014575e-06,
+      "loss": 0.7726,
+      "step": 7730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5617136138782936e-06,
+      "loss": 0.7928,
+      "step": 7731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.554136781351359e-06,
+      "loss": 0.7908,
+      "step": 7732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.546565946220571e-06,
+      "loss": 0.7125,
+      "step": 7733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.53900110948503e-06,
+      "loss": 0.8058,
+      "step": 7734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.531442272143055e-06,
+      "loss": 0.6611,
+      "step": 7735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.523889435192164e-06,
+      "loss": 0.7083,
+      "step": 7736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.516342599629081e-06,
+      "loss": 0.7152,
+      "step": 7737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.508801766449766e-06,
+      "loss": 0.799,
+      "step": 7738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.501266936649362e-06,
+      "loss": 0.7048,
+      "step": 7739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.493738111222218e-06,
+      "loss": 0.758,
+      "step": 7740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4862152911618934e-06,
+      "loss": 0.7121,
+      "step": 7741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.478698477461163e-06,
+      "loss": 0.685,
+      "step": 7742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4711876711120205e-06,
+      "loss": 0.7342,
+      "step": 7743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4636828731056345e-06,
+      "loss": 0.7247,
+      "step": 7744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.456184084432402e-06,
+      "loss": 0.7117,
+      "step": 7745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.448691306081926e-06,
+      "loss": 0.713,
+      "step": 7746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4412045390430034e-06,
+      "loss": 0.7932,
+      "step": 7747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.433723784303656e-06,
+      "loss": 0.7543,
+      "step": 7748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4262490428510994e-06,
+      "loss": 0.7317,
+      "step": 7749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.418780315671766e-06,
+      "loss": 0.7599,
+      "step": 7750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.41131760375128e-06,
+      "loss": 0.7548,
+      "step": 7751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.40386090807447e-06,
+      "loss": 0.7908,
+      "step": 7752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.396410229625408e-06,
+      "loss": 0.6506,
+      "step": 7753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.388965569387332e-06,
+      "loss": 0.718,
+      "step": 7754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.381526928342683e-06,
+      "loss": 0.8038,
+      "step": 7755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3740943074731475e-06,
+      "loss": 0.7886,
+      "step": 7756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.366667707759581e-06,
+      "loss": 0.7481,
+      "step": 7757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.359247130182059e-06,
+      "loss": 0.7371,
+      "step": 7758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.351832575719861e-06,
+      "loss": 0.6996,
+      "step": 7759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.344424045351464e-06,
+      "loss": 0.7389,
+      "step": 7760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.337021540054564e-06,
+      "loss": 0.7004,
+      "step": 7761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.329625060806036e-06,
+      "loss": 0.8032,
+      "step": 7762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.322234608582004e-06,
+      "loss": 0.7728,
+      "step": 7763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.31485018435776e-06,
+      "loss": 0.7556,
+      "step": 7764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3074717891078085e-06,
+      "loss": 0.7847,
+      "step": 7765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3000994238058644e-06,
+      "loss": 0.7141,
+      "step": 7766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.292733089424844e-06,
+      "loss": 0.6954,
+      "step": 7767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.285372786936859e-06,
+      "loss": 0.7174,
+      "step": 7768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.278018517313243e-06,
+      "loss": 0.8003,
+      "step": 7769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.270670281524519e-06,
+      "loss": 0.7523,
+      "step": 7770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2633280805404076e-06,
+      "loss": 0.7444,
+      "step": 7771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.255991915329865e-06,
+      "loss": 0.7717,
+      "step": 7772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.248661786861019e-06,
+      "loss": 0.7307,
+      "step": 7773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.241337696101205e-06,
+      "loss": 0.6817,
+      "step": 7774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.23401964401699e-06,
+      "loss": 0.6906,
+      "step": 7775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2267076315741e-06,
+      "loss": 0.8305,
+      "step": 7776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.219401659737499e-06,
+      "loss": 0.7516,
+      "step": 7777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2121017294713295e-06,
+      "loss": 0.6508,
+      "step": 7778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.204807841738956e-06,
+      "loss": 0.7298,
+      "step": 7779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.197519997502936e-06,
+      "loss": 0.8034,
+      "step": 7780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.190238197725027e-06,
+      "loss": 0.7008,
+      "step": 7781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.182962443366201e-06,
+      "loss": 0.6975,
+      "step": 7782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1756927353866215e-06,
+      "loss": 0.7612,
+      "step": 7783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.168429074745656e-06,
+      "loss": 0.7005,
+      "step": 7784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.161171462401875e-06,
+      "loss": 0.7257,
+      "step": 7785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.153919899313047e-06,
+      "loss": 0.7207,
+      "step": 7786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1466743864361505e-06,
+      "loss": 0.798,
+      "step": 7787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.139434924727359e-06,
+      "loss": 0.7697,
+      "step": 7788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.132201515142037e-06,
+      "loss": 0.6198,
+      "step": 7789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.124974158634787e-06,
+      "loss": 0.7523,
+      "step": 7790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.117752856159373e-06,
+      "loss": 0.789,
+      "step": 7791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.110537608668769e-06,
+      "loss": 0.7597,
+      "step": 7792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.103328417115176e-06,
+      "loss": 0.7277,
+      "step": 7793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.096125282449964e-06,
+      "loss": 0.7349,
+      "step": 7794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.088928205623721e-06,
+      "loss": 0.7267,
+      "step": 7795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.081737187586221e-06,
+      "loss": 0.808,
+      "step": 7796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.074552229286455e-06,
+      "loss": 0.7385,
+      "step": 7797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.067373331672603e-06,
+      "loss": 0.7334,
+      "step": 7798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.060200495692052e-06,
+      "loss": 0.7336,
+      "step": 7799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.053033722291388e-06,
+      "loss": 0.7023,
+      "step": 7800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.045873012416379e-06,
+      "loss": 0.7858,
+      "step": 7801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.038718367012034e-06,
+      "loss": 0.7141,
+      "step": 7802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.031569787022526e-06,
+      "loss": 0.7814,
+      "step": 7803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.02442727339124e-06,
+      "loss": 0.7847,
+      "step": 7804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.017290827060755e-06,
+      "loss": 0.7591,
+      "step": 7805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.010160448972849e-06,
+      "loss": 0.7358,
+      "step": 7806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.003036140068511e-06,
+      "loss": 0.7612,
+      "step": 7807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.99591790128791e-06,
+      "loss": 0.7949,
+      "step": 7808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.988805733570445e-06,
+      "loss": 0.7382,
+      "step": 7809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.98169963785468e-06,
+      "loss": 0.7817,
+      "step": 7810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9745996150783895e-06,
+      "loss": 0.7886,
+      "step": 7811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.967505666178556e-06,
+      "loss": 0.6726,
+      "step": 7812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.960417792091359e-06,
+      "loss": 0.7883,
+      "step": 7813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9533359937521554e-06,
+      "loss": 0.6573,
+      "step": 7814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9462602720955255e-06,
+      "loss": 0.7363,
+      "step": 7815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.939190628055234e-06,
+      "loss": 0.7854,
+      "step": 7816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.93212706256425e-06,
+      "loss": 0.7869,
+      "step": 7817
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.925069576554735e-06,
+      "loss": 0.769,
+      "step": 7818
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9180181709580485e-06,
+      "loss": 0.7366,
+      "step": 7819
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.910972846704752e-06,
+      "loss": 0.6967,
+      "step": 7820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.903933604724596e-06,
+      "loss": 0.7752,
+      "step": 7821
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.89690044594655e-06,
+      "loss": 0.7902,
+      "step": 7822
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.889873371298753e-06,
+      "loss": 0.769,
+      "step": 7823
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.882852381708563e-06,
+      "loss": 0.79,
+      "step": 7824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.875837478102512e-06,
+      "loss": 0.7297,
+      "step": 7825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.868828661406349e-06,
+      "loss": 0.6909,
+      "step": 7826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.861825932545015e-06,
+      "loss": 0.7476,
+      "step": 7827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8548292924426525e-06,
+      "loss": 0.7545,
+      "step": 7828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.84783874202258e-06,
+      "loss": 0.7225,
+      "step": 7829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.840854282207329e-06,
+      "loss": 0.6905,
+      "step": 7830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.833875913918622e-06,
+      "loss": 0.85,
+      "step": 7831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.826903638077395e-06,
+      "loss": 0.7871,
+      "step": 7832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.81993745560375e-06,
+      "loss": 0.8178,
+      "step": 7833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.812977367417003e-06,
+      "loss": 0.7757,
+      "step": 7834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8060233744356633e-06,
+      "loss": 0.6664,
+      "step": 7835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.799075477577435e-06,
+      "loss": 0.7611,
+      "step": 7836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7921336777592187e-06,
+      "loss": 0.7887,
+      "step": 7837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.785197975897109e-06,
+      "loss": 0.8325,
+      "step": 7838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7782683729063904e-06,
+      "loss": 0.7897,
+      "step": 7839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7713448697015485e-06,
+      "loss": 0.7221,
+      "step": 7840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.764427467196274e-06,
+      "loss": 0.7485,
+      "step": 7841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7575161663034318e-06,
+      "loss": 0.7421,
+      "step": 7842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.750610967935103e-06,
+      "loss": 0.6542,
+      "step": 7843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7437118730025424e-06,
+      "loss": 0.7549,
+      "step": 7844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.736818882416204e-06,
+      "loss": 0.7189,
+      "step": 7845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.729931997085756e-06,
+      "loss": 0.7012,
+      "step": 7846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.723051217920043e-06,
+      "loss": 0.7494,
+      "step": 7847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.716176545827105e-06,
+      "loss": 0.8043,
+      "step": 7848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.709307981714172e-06,
+      "loss": 0.7832,
+      "step": 7849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7024455264876747e-06,
+      "loss": 0.8154,
+      "step": 7850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6955891810532494e-06,
+      "loss": 0.7336,
+      "step": 7851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.688738946315706e-06,
+      "loss": 0.8591,
+      "step": 7852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.681894823179055e-06,
+      "loss": 0.7847,
+      "step": 7853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6750568125465012e-06,
+      "loss": 0.6696,
+      "step": 7854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.668224915320445e-06,
+      "loss": 0.7092,
+      "step": 7855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6613991324024766e-06,
+      "loss": 0.7488,
+      "step": 7856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6545794646933805e-06,
+      "loss": 0.7347,
+      "step": 7857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.647765913093132e-06,
+      "loss": 0.7193,
+      "step": 7858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6409584785009055e-06,
+      "loss": 0.7896,
+      "step": 7859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.634157161815049e-06,
+      "loss": 0.6809,
+      "step": 7860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.62736196393314e-06,
+      "loss": 0.6975,
+      "step": 7861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.620572885751916e-06,
+      "loss": 0.9272,
+      "step": 7862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6137899281673105e-06,
+      "loss": 0.6637,
+      "step": 7863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6070130920744737e-06,
+      "loss": 0.7788,
+      "step": 7864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.600242378367724e-06,
+      "loss": 0.7281,
+      "step": 7865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.593477787940569e-06,
+      "loss": 0.7347,
+      "step": 7866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5867193216857333e-06,
+      "loss": 0.7125,
+      "step": 7867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.579966980495103e-06,
+      "loss": 0.7156,
+      "step": 7868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.573220765259777e-06,
+      "loss": 0.7062,
+      "step": 7869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5664806768700264e-06,
+      "loss": 0.767,
+      "step": 7870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.559746716215351e-06,
+      "loss": 0.7577,
+      "step": 7871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.553018884184406e-06,
+      "loss": 0.7508,
+      "step": 7872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.546297181665048e-06,
+      "loss": 0.7052,
+      "step": 7873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539581609544329e-06,
+      "loss": 0.7981,
+      "step": 7874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.532872168708484e-06,
+      "loss": 0.7411,
+      "step": 7875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.52616886004295e-06,
+      "loss": 0.7028,
+      "step": 7876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5194716844323406e-06,
+      "loss": 0.6777,
+      "step": 7877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.512780642760477e-06,
+      "loss": 0.7743,
+      "step": 7878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.506095735910353e-06,
+      "loss": 0.6948,
+      "step": 7879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.499416964764168e-06,
+      "loss": 0.7834,
+      "step": 7880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4927443302033126e-06,
+      "loss": 0.6502,
+      "step": 7881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.486077833108342e-06,
+      "loss": 0.8108,
+      "step": 7882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4794174743590367e-06,
+      "loss": 0.7069,
+      "step": 7883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4727632548343435e-06,
+      "loss": 0.8049,
+      "step": 7884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.46611517541241e-06,
+      "loss": 0.6931,
+      "step": 7885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4594732369705617e-06,
+      "loss": 0.7314,
+      "step": 7886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4528374403853315e-06,
+      "loss": 0.8521,
+      "step": 7887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.446207786532424e-06,
+      "loss": 0.7112,
+      "step": 7888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.439584276286739e-06,
+      "loss": 0.7198,
+      "step": 7889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4329669105223674e-06,
+      "loss": 0.7166,
+      "step": 7890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.426355690112598e-06,
+      "loss": 0.6794,
+      "step": 7891
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4197506159299007e-06,
+      "loss": 0.7379,
+      "step": 7892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4131516888459224e-06,
+      "loss": 0.7691,
+      "step": 7893
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4065589097315163e-06,
+      "loss": 0.7022,
+      "step": 7894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.39997227945672e-06,
+      "loss": 0.7063,
+      "step": 7895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3933917988907493e-06,
+      "loss": 0.79,
+      "step": 7896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.386817468902015e-06,
+      "loss": 0.8027,
+      "step": 7897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.380249290358134e-06,
+      "loss": 0.7645,
+      "step": 7898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.37368726412588e-06,
+      "loss": 0.8352,
+      "step": 7899
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.367131391071232e-06,
+      "loss": 0.7917,
+      "step": 7900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3605816720593652e-06,
+      "loss": 0.8098,
+      "step": 7901
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3540381079546212e-06,
+      "loss": 0.7372,
+      "step": 7902
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3475006996205484e-06,
+      "loss": 0.7451,
+      "step": 7903
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.340969447919873e-06,
+      "loss": 0.759,
+      "step": 7904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3344443537145054e-06,
+      "loss": 0.73,
+      "step": 7905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3279254178655518e-06,
+      "loss": 0.7165,
+      "step": 7906
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.321412641233307e-06,
+      "loss": 0.6766,
+      "step": 7907
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.314906024677239e-06,
+      "loss": 0.7436,
+      "step": 7908
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3084055690560165e-06,
+      "loss": 0.7492,
+      "step": 7909
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3019112752274983e-06,
+      "loss": 0.7353,
+      "step": 7910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.295423144048715e-06,
+      "loss": 0.7495,
+      "step": 7911
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2889411763758927e-06,
+      "loss": 0.7457,
+      "step": 7912
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2824653730644473e-06,
+      "loss": 0.7255,
+      "step": 7913
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2759957349689786e-06,
+      "loss": 0.6855,
+      "step": 7914
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.269532262943259e-06,
+      "loss": 0.6714,
+      "step": 7915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2630749578402675e-06,
+      "loss": 0.7646,
+      "step": 7916
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2566238205121613e-06,
+      "loss": 0.728,
+      "step": 7917
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2501788518102928e-06,
+      "loss": 0.7417,
+      "step": 7918
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2437400525851646e-06,
+      "loss": 0.6934,
+      "step": 7919
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.237307423686525e-06,
+      "loss": 0.7478,
+      "step": 7920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.230880965963251e-06,
+      "loss": 0.7635,
+      "step": 7921
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2244606802634415e-06,
+      "loss": 0.719,
+      "step": 7922
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.218046567434363e-06,
+      "loss": 0.6835,
+      "step": 7923
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.211638628322472e-06,
+      "loss": 0.8352,
+      "step": 7924
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.205236863773414e-06,
+      "loss": 0.7558,
+      "step": 7925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.198841274632014e-06,
+      "loss": 0.7508,
+      "step": 7926
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1924518617422792e-06,
+      "loss": 0.7016,
+      "step": 7927
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1860686259474194e-06,
+      "loss": 0.6949,
+      "step": 7928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1796915680897988e-06,
+      "loss": 0.6185,
+      "step": 7929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1733206890110057e-06,
+      "loss": 0.6661,
+      "step": 7930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.166955989551784e-06,
+      "loss": 0.685,
+      "step": 7931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.160597470552068e-06,
+      "loss": 0.7566,
+      "step": 7932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1542451328509747e-06,
+      "loss": 0.7382,
+      "step": 7933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.147898977286806e-06,
+      "loss": 0.6903,
+      "step": 7934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.14155900469707e-06,
+      "loss": 0.8065,
+      "step": 7935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1352252159184194e-06,
+      "loss": 0.6645,
+      "step": 7936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1288976117867243e-06,
+      "loss": 0.6757,
+      "step": 7937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.122576193137017e-06,
+      "loss": 0.7486,
+      "step": 7938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1162609608035187e-06,
+      "loss": 0.7298,
+      "step": 7939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.109951915619652e-06,
+      "loss": 0.8097,
+      "step": 7940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.103649058418001e-06,
+      "loss": 0.7935,
+      "step": 7941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.097352390030339e-06,
+      "loss": 0.8118,
+      "step": 7942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0910619112876295e-06,
+      "loss": 0.6842,
+      "step": 7943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.084777623020002e-06,
+      "loss": 0.7488,
+      "step": 7944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.078499526056794e-06,
+      "loss": 0.8257,
+      "step": 7945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0722276212265035e-06,
+      "loss": 0.7744,
+      "step": 7946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0659619093568293e-06,
+      "loss": 0.8039,
+      "step": 7947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.059702391274638e-06,
+      "loss": 0.7782,
+      "step": 7948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0534490678059803e-06,
+      "loss": 0.7989,
+      "step": 7949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0472019397761064e-06,
+      "loss": 0.7005,
+      "step": 7950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.04096100800943e-06,
+      "loss": 0.6964,
+      "step": 7951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.034726273329558e-06,
+      "loss": 0.698,
+      "step": 7952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.028497736559266e-06,
+      "loss": 0.7147,
+      "step": 7953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0222753985205354e-06,
+      "loss": 0.766,
+      "step": 7954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0160592600345037e-06,
+      "loss": 0.6643,
+      "step": 7955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0098493219215096e-06,
+      "loss": 0.7252,
+      "step": 7956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.003645585001058e-06,
+      "loss": 0.7703,
+      "step": 7957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9974480500918445e-06,
+      "loss": 0.7225,
+      "step": 7958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9912567180117534e-06,
+      "loss": 0.7329,
+      "step": 7959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9850715895778367e-06,
+      "loss": 0.7821,
+      "step": 7960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.978892665606331e-06,
+      "loss": 0.6948,
+      "step": 7961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9727199469126556e-06,
+      "loss": 0.7172,
+      "step": 7962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.966553434311414e-06,
+      "loss": 0.7591,
+      "step": 7963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9603931286163843e-06,
+      "loss": 0.7361,
+      "step": 7964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9542390306405377e-06,
+      "loss": 0.7855,
+      "step": 7965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.948091141196008e-06,
+      "loss": 0.8056,
+      "step": 7966
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9419494610941235e-06,
+      "loss": 0.8029,
+      "step": 7967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9358139911453864e-06,
+      "loss": 0.7367,
+      "step": 7968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9296847321594876e-06,
+      "loss": 0.7631,
+      "step": 7969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9235616849452864e-06,
+      "loss": 0.7573,
+      "step": 7970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.91744485031083e-06,
+      "loss": 0.7259,
+      "step": 7971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9113342290633504e-06,
+      "loss": 0.6439,
+      "step": 7972
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.905229822009253e-06,
+      "loss": 0.7618,
+      "step": 7973
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.899131629954116e-06,
+      "loss": 0.7547,
+      "step": 7974
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8930396537027115e-06,
+      "loss": 0.6796,
+      "step": 7975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.88695389405898e-06,
+      "loss": 0.7771,
+      "step": 7976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.880874351826046e-06,
+      "loss": 0.6954,
+      "step": 7977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.874801027806218e-06,
+      "loss": 0.7596,
+      "step": 7978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8687339228009812e-06,
+      "loss": 0.718,
+      "step": 7979
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8626730376109967e-06,
+      "loss": 0.77,
+      "step": 7980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8566183730361073e-06,
+      "loss": 0.6672,
+      "step": 7981
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8505699298753352e-06,
+      "loss": 0.7193,
+      "step": 7982
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.844527708926881e-06,
+      "loss": 0.7016,
+      "step": 7983
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8384917109881226e-06,
+      "loss": 0.7742,
+      "step": 7984
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.832461936855618e-06,
+      "loss": 0.7619,
+      "step": 7985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8264383873251033e-06,
+      "loss": 0.7621,
+      "step": 7986
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.820421063191486e-06,
+      "loss": 0.7799,
+      "step": 7987
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8144099652488766e-06,
+      "loss": 0.8146,
+      "step": 7988
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8084050942905447e-06,
+      "loss": 0.7124,
+      "step": 7989
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8024064511089245e-06,
+      "loss": 0.653,
+      "step": 7990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7964140364956592e-06,
+      "loss": 0.7828,
+      "step": 7991
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.790427851241556e-06,
+      "loss": 0.8447,
+      "step": 7992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.784447896136594e-06,
+      "loss": 0.781,
+      "step": 7993
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7784741719699304e-06,
+      "loss": 0.7415,
+      "step": 7994
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.772506679529918e-06,
+      "loss": 0.7838,
+      "step": 7995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7665454196040664e-06,
+      "loss": 0.731,
+      "step": 7996
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7605903929790678e-06,
+      "loss": 0.6834,
+      "step": 7997
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.754641600440794e-06,
+      "loss": 0.7269,
+      "step": 7998
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.748699042774305e-06,
+      "loss": 0.7449,
+      "step": 7999
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.742762720763825e-06,
+      "loss": 0.7242,
+      "step": 8000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.736832635192749e-06,
+      "loss": 0.6831,
+      "step": 8001
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7309087868436722e-06,
+      "loss": 0.7825,
+      "step": 8002
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.724991176498337e-06,
+      "loss": 0.7441,
+      "step": 8003
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.719079804937691e-06,
+      "loss": 0.6281,
+      "step": 8004
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7131746729418373e-06,
+      "loss": 0.7216,
+      "step": 8005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7072757812900584e-06,
+      "loss": 0.5787,
+      "step": 8006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7013831307608373e-06,
+      "loss": 0.6692,
+      "step": 8007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6954967221318018e-06,
+      "loss": 0.68,
+      "step": 8008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6896165561797747e-06,
+      "loss": 0.6671,
+      "step": 8009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.683742633680747e-06,
+      "loss": 0.7538,
+      "step": 8010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.677874955409887e-06,
+      "loss": 0.7601,
+      "step": 8011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6720135221415364e-06,
+      "loss": 0.7532,
+      "step": 8012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.666158334649227e-06,
+      "loss": 0.7745,
+      "step": 8013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6603093937056456e-06,
+      "loss": 0.7148,
+      "step": 8014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.654466700082675e-06,
+      "loss": 0.8267,
+      "step": 8015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6486302545513488e-06,
+      "loss": 0.7404,
+      "step": 8016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6428000578818947e-06,
+      "loss": 0.6659,
+      "step": 8017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.636976110843725e-06,
+      "loss": 0.7411,
+      "step": 8018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6311584142054034e-06,
+      "loss": 0.7646,
+      "step": 8019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6253469687346764e-06,
+      "loss": 0.7358,
+      "step": 8020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6195417751984807e-06,
+      "loss": 0.6221,
+      "step": 8021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.613742834362898e-06,
+      "loss": 0.7378,
+      "step": 8022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6079501469932156e-06,
+      "loss": 0.6599,
+      "step": 8023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6021637138538726e-06,
+      "loss": 0.7456,
+      "step": 8024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5963835357085074e-06,
+      "loss": 0.6724,
+      "step": 8025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5906096133199056e-06,
+      "loss": 0.6877,
+      "step": 8026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.584841947450034e-06,
+      "loss": 0.7336,
+      "step": 8027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5790805388600624e-06,
+      "loss": 0.7461,
+      "step": 8028
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5733253883102938e-06,
+      "loss": 0.6853,
+      "step": 8029
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.567576496560231e-06,
+      "loss": 0.6922,
+      "step": 8030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5618338643685404e-06,
+      "loss": 0.728,
+      "step": 8031
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5560974924930593e-06,
+      "loss": 0.757,
+      "step": 8032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.550367381690816e-06,
+      "loss": 0.7309,
+      "step": 8033
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.544643532717994e-06,
+      "loss": 0.7176,
+      "step": 8034
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.538925946329962e-06,
+      "loss": 0.7296,
+      "step": 8035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5332146232812547e-06,
+      "loss": 0.6359,
+      "step": 8036
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5275095643255752e-06,
+      "loss": 0.7946,
+      "step": 8037
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5218107702158266e-06,
+      "loss": 0.7056,
+      "step": 8038
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5161182417040575e-06,
+      "loss": 0.7092,
+      "step": 8039
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5104319795415e-06,
+      "loss": 0.7157,
+      "step": 8040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5047519844785537e-06,
+      "loss": 0.7679,
+      "step": 8041
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4990782572647975e-06,
+      "loss": 0.6983,
+      "step": 8042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.493410798648993e-06,
+      "loss": 0.7533,
+      "step": 8043
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4877496093790474e-06,
+      "loss": 0.7512,
+      "step": 8044
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4820946902020634e-06,
+      "loss": 0.7027,
+      "step": 8045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.476446041864311e-06,
+      "loss": 0.7797,
+      "step": 8046
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.470803665111221e-06,
+      "loss": 0.7186,
+      "step": 8047
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.465167560687415e-06,
+      "loss": 0.6966,
+      "step": 8048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4595377293366806e-06,
+      "loss": 0.7179,
+      "step": 8049
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.453914171801969e-06,
+      "loss": 0.6917,
+      "step": 8050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.448296888825408e-06,
+      "loss": 0.8172,
+      "step": 8051
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4426858811483043e-06,
+      "loss": 0.8332,
+      "step": 8052
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.437081149511122e-06,
+      "loss": 0.7095,
+      "step": 8053
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.431482694653514e-06,
+      "loss": 0.7359,
+      "step": 8054
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4258905173142998e-06,
+      "loss": 0.7745,
+      "step": 8055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.420304618231456e-06,
+      "loss": 0.6524,
+      "step": 8056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.414724998142143e-06,
+      "loss": 0.8529,
+      "step": 8057
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4091516577827046e-06,
+      "loss": 0.7341,
+      "step": 8058
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.403584597888631e-06,
+      "loss": 0.8588,
+      "step": 8059
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3980238191946057e-06,
+      "loss": 0.6603,
+      "step": 8060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.392469322434454e-06,
+      "loss": 0.7551,
+      "step": 8061
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.386921108341217e-06,
+      "loss": 0.7124,
+      "step": 8062
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.381379177647064e-06,
+      "loss": 0.7771,
+      "step": 8063
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3758435310833606e-06,
+      "loss": 0.6896,
+      "step": 8064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3703141693806274e-06,
+      "loss": 0.7521,
+      "step": 8065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.364791093268559e-06,
+      "loss": 0.7622,
+      "step": 8066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.359274303476039e-06,
+      "loss": 0.715,
+      "step": 8067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.353763800731096e-06,
+      "loss": 0.7366,
+      "step": 8068
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3482595857609425e-06,
+      "loss": 0.8049,
+      "step": 8069
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.342761659291959e-06,
+      "loss": 0.6899,
+      "step": 8070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.337270022049698e-06,
+      "loss": 0.7284,
+      "step": 8071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3317846747588746e-06,
+      "loss": 0.7988,
+      "step": 8072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3263056181433763e-06,
+      "loss": 0.7139,
+      "step": 8073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.320832852926269e-06,
+      "loss": 0.7696,
+      "step": 8074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.315366379829781e-06,
+      "loss": 0.7326,
+      "step": 8075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3099061995753012e-06,
+      "loss": 0.8287,
+      "step": 8076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.304452312883415e-06,
+      "loss": 0.7226,
+      "step": 8077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.299004720473852e-06,
+      "loss": 0.7461,
+      "step": 8078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2935634230655145e-06,
+      "loss": 0.7141,
+      "step": 8079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.28812842137649e-06,
+      "loss": 0.7965,
+      "step": 8080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2826997161240204e-06,
+      "loss": 0.6742,
+      "step": 8081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2772773080245225e-06,
+      "loss": 0.6934,
+      "step": 8082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2718611977935733e-06,
+      "loss": 0.7627,
+      "step": 8083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2664513861459347e-06,
+      "loss": 0.702,
+      "step": 8084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.261047873795519e-06,
+      "loss": 0.7537,
+      "step": 8085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.255650661455416e-06,
+      "loss": 0.9177,
+      "step": 8086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2502597498378963e-06,
+      "loss": 0.7299,
+      "step": 8087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2448751396543787e-06,
+      "loss": 0.7575,
+      "step": 8088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2394968316154617e-06,
+      "loss": 0.7137,
+      "step": 8089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.234124826430911e-06,
+      "loss": 0.7751,
+      "step": 8090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.228759124809654e-06,
+      "loss": 0.7194,
+      "step": 8091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.223399727459791e-06,
+      "loss": 0.6981,
+      "step": 8092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.218046635088594e-06,
+      "loss": 0.7125,
+      "step": 8093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2126998484025053e-06,
+      "loss": 0.8065,
+      "step": 8094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2073593681071092e-06,
+      "loss": 0.7416,
+      "step": 8095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.202025194907198e-06,
+      "loss": 0.7993,
+      "step": 8096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1966973295067083e-06,
+      "loss": 0.6654,
+      "step": 8097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1913757726087335e-06,
+      "loss": 0.709,
+      "step": 8098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1860605249155673e-06,
+      "loss": 0.7125,
+      "step": 8099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1807515871286433e-06,
+      "loss": 0.7713,
+      "step": 8100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.175448959948573e-06,
+      "loss": 0.6678,
+      "step": 8101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.170152644075124e-06,
+      "loss": 0.7401,
+      "step": 8102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.164862640207255e-06,
+      "loss": 0.6428,
+      "step": 8103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1595789490430617e-06,
+      "loss": 0.7408,
+      "step": 8104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1543015712798374e-06,
+      "loss": 0.7894,
+      "step": 8105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.149030507614008e-06,
+      "loss": 0.6493,
+      "step": 8106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1437657587412006e-06,
+      "loss": 0.8032,
+      "step": 8107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1385073253561872e-06,
+      "loss": 0.7125,
+      "step": 8108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1332552081529124e-06,
+      "loss": 0.6945,
+      "step": 8109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1280094078244885e-06,
+      "loss": 0.7331,
+      "step": 8110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.122769925063195e-06,
+      "loss": 0.792,
+      "step": 8111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1175367605604677e-06,
+      "loss": 0.7036,
+      "step": 8112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.112309915006927e-06,
+      "loss": 0.7745,
+      "step": 8113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1070893890923317e-06,
+      "loss": 0.72,
+      "step": 8114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1018751835056474e-06,
+      "loss": 0.7085,
+      "step": 8115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0966672989349566e-06,
+      "loss": 0.7274,
+      "step": 8116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0914657360675606e-06,
+      "loss": 0.7986,
+      "step": 8117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.086270495589876e-06,
+      "loss": 0.7054,
+      "step": 8118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0810815781875214e-06,
+      "loss": 0.8314,
+      "step": 8119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0758989845452605e-06,
+      "loss": 0.7422,
+      "step": 8120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0707227153470343e-06,
+      "loss": 0.7645,
+      "step": 8121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.065552771275936e-06,
+      "loss": 0.7048,
+      "step": 8122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0603891530142417e-06,
+      "loss": 0.7292,
+      "step": 8123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0552318612433785e-06,
+      "loss": 0.7162,
+      "step": 8124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.050080896643941e-06,
+      "loss": 0.7523,
+      "step": 8125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0449362598956967e-06,
+      "loss": 0.7123,
+      "step": 8126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.039797951677569e-06,
+      "loss": 0.767,
+      "step": 8127
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0346659726676542e-06,
+      "loss": 0.6988,
+      "step": 8128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0295403235432053e-06,
+      "loss": 0.7853,
+      "step": 8129
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0244210049806477e-06,
+      "loss": 0.6419,
+      "step": 8130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.019308017655558e-06,
+      "loss": 0.7654,
+      "step": 8131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0142013622426904e-06,
+      "loss": 0.7152,
+      "step": 8132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.009101039415967e-06,
+      "loss": 0.8247,
+      "step": 8133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.004007049848461e-06,
+      "loss": 0.7229,
+      "step": 8134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998919394212412e-06,
+      "loss": 0.7654,
+      "step": 8135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.993838073179233e-06,
+      "loss": 0.7656,
+      "step": 8136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9887630874194985e-06,
+      "loss": 0.6889,
+      "step": 8137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9836944376029344e-06,
+      "loss": 0.6968,
+      "step": 8138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9786321243984494e-06,
+      "loss": 0.7274,
+      "step": 8139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.973576148474099e-06,
+      "loss": 0.765,
+      "step": 8140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9685265104971098e-06,
+      "loss": 0.7195,
+      "step": 8141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.963483211133876e-06,
+      "loss": 0.68,
+      "step": 8142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9584462510499503e-06,
+      "loss": 0.6643,
+      "step": 8143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.953415630910049e-06,
+      "loss": 0.6626,
+      "step": 8144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.948391351378048e-06,
+      "loss": 0.7881,
+      "step": 8145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9433734131170056e-06,
+      "loss": 0.8298,
+      "step": 8146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9383618167891136e-06,
+      "loss": 0.7277,
+      "step": 8147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9333565630557492e-06,
+      "loss": 0.7995,
+      "step": 8148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.928357652577445e-06,
+      "loss": 0.7837,
+      "step": 8149
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.923365086013895e-06,
+      "loss": 0.711,
+      "step": 8150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9183788640239563e-06,
+      "loss": 0.8003,
+      "step": 8151
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.913398987265652e-06,
+      "loss": 0.78,
+      "step": 8152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9084254563961677e-06,
+      "loss": 0.6692,
+      "step": 8153
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9034582720718507e-06,
+      "loss": 0.7445,
+      "step": 8154
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8984974349482044e-06,
+      "loss": 0.574,
+      "step": 8155
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8935429456799047e-06,
+      "loss": 0.7657,
+      "step": 8156
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8885948049207847e-06,
+      "loss": 0.7643,
+      "step": 8157
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.883653013323844e-06,
+      "loss": 0.7265,
+      "step": 8158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878717571541233e-06,
+      "loss": 0.7836,
+      "step": 8159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8737884802242756e-06,
+      "loss": 0.7819,
+      "step": 8160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.868865740023451e-06,
+      "loss": 0.7342,
+      "step": 8161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8639493515884065e-06,
+      "loss": 0.6865,
+      "step": 8162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.859039315567951e-06,
+      "loss": 0.6223,
+      "step": 8163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8541356326100433e-06,
+      "loss": 0.7174,
+      "step": 8164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8492383033618111e-06,
+      "loss": 0.7554,
+      "step": 8165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8443473284695534e-06,
+      "loss": 0.7074,
+      "step": 8166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8394627085787208e-06,
+      "loss": 0.7434,
+      "step": 8167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.83458444433392e-06,
+      "loss": 0.8101,
+      "step": 8168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8297125363789303e-06,
+      "loss": 0.7216,
+      "step": 8169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8248469853566874e-06,
+      "loss": 0.6498,
+      "step": 8170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8199877919092888e-06,
+      "loss": 0.6932,
+      "step": 8171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8151349566779884e-06,
+      "loss": 0.697,
+      "step": 8172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8102884803032073e-06,
+      "loss": 0.7753,
+      "step": 8173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8054483634245222e-06,
+      "loss": 0.7631,
+      "step": 8174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.800614606680673e-06,
+      "loss": 0.6933,
+      "step": 8175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7957872107095707e-06,
+      "loss": 0.6691,
+      "step": 8176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7909661761482621e-06,
+      "loss": 0.7221,
+      "step": 8177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7861515036329823e-06,
+      "loss": 0.7035,
+      "step": 8178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7813431937991064e-06,
+      "loss": 0.6905,
+      "step": 8179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7765412472811771e-06,
+      "loss": 0.6774,
+      "step": 8180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7717456647128927e-06,
+      "loss": 0.7684,
+      "step": 8181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7669564467271305e-06,
+      "loss": 0.7188,
+      "step": 8182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7621735939558958e-06,
+      "loss": 0.8127,
+      "step": 8183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.757397107030384e-06,
+      "loss": 0.7268,
+      "step": 8184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7526269865809352e-06,
+      "loss": 0.7832,
+      "step": 8185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7478632332370514e-06,
+      "loss": 0.7624,
+      "step": 8186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7431058476273965e-06,
+      "loss": 0.7991,
+      "step": 8187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7383548303797903e-06,
+      "loss": 0.8636,
+      "step": 8188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7336101821212203e-06,
+      "loss": 0.7633,
+      "step": 8189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7288719034778188e-06,
+      "loss": 0.7811,
+      "step": 8190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.724139995074897e-06,
+      "loss": 0.7724,
+      "step": 8191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.719414457536911e-06,
+      "loss": 0.7263,
+      "step": 8192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.714695291487478e-06,
+      "loss": 0.7062,
+      "step": 8193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7099824975493727e-06,
+      "loss": 0.843,
+      "step": 8194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7052760763445364e-06,
+      "loss": 0.6957,
+      "step": 8195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7005760284940774e-06,
+      "loss": 0.7824,
+      "step": 8196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6958823546182333e-06,
+      "loss": 0.8112,
+      "step": 8197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6911950553364254e-06,
+      "loss": 0.7511,
+      "step": 8198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6865141312672306e-06,
+      "loss": 0.7365,
+      "step": 8199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6818395830283774e-06,
+      "loss": 0.7713,
+      "step": 8200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6771714112367554e-06,
+      "loss": 0.7931,
+      "step": 8201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6725096165084165e-06,
+      "loss": 0.7974,
+      "step": 8202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6678541994585628e-06,
+      "loss": 0.8122,
+      "step": 8203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6632051607015642e-06,
+      "loss": 0.6075,
+      "step": 8204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.658562500850941e-06,
+      "loss": 0.8063,
+      "step": 8205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6539262205193807e-06,
+      "loss": 0.6672,
+      "step": 8206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6492963203187218e-06,
+      "loss": 0.7273,
+      "step": 8207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6446728008599588e-06,
+      "loss": 0.7499,
+      "step": 8208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6400556627532538e-06,
+      "loss": 0.7976,
+      "step": 8209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6354449066079192e-06,
+      "loss": 0.678,
+      "step": 8210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6308405330324294e-06,
+      "loss": 0.7599,
+      "step": 8211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6262425426344041e-06,
+      "loss": 0.7221,
+      "step": 8212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6216509360206411e-06,
+      "loss": 0.7994,
+      "step": 8213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6170657137970724e-06,
+      "loss": 0.7644,
+      "step": 8214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.612486876568814e-06,
+      "loss": 0.6839,
+      "step": 8215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6079144249401211e-06,
+      "loss": 0.7642,
+      "step": 8216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6033483595144117e-06,
+      "loss": 0.785,
+      "step": 8217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5987886808942532e-06,
+      "loss": 0.8377,
+      "step": 8218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5942353896813866e-06,
+      "loss": 0.6643,
+      "step": 8219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5896884864766925e-06,
+      "loss": 0.7224,
+      "step": 8220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5851479718802243e-06,
+      "loss": 0.7702,
+      "step": 8221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.580613846491169e-06,
+      "loss": 0.6271,
+      "step": 8222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.576086110907904e-06,
+      "loss": 0.6686,
+      "step": 8223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5715647657279286e-06,
+      "loss": 0.7145,
+      "step": 8224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5670498115479326e-06,
+      "loss": 0.7092,
+      "step": 8225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5625412489637337e-06,
+      "loss": 0.7598,
+      "step": 8226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5580390785703226e-06,
+      "loss": 0.7911,
+      "step": 8227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5535433009618406e-06,
+      "loss": 0.7429,
+      "step": 8228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5490539167315799e-06,
+      "loss": 0.8161,
+      "step": 8229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5445709264720055e-06,
+      "loss": 0.7459,
+      "step": 8230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5400943307747163e-06,
+      "loss": 0.7553,
+      "step": 8231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5356241302304898e-06,
+      "loss": 0.8193,
+      "step": 8232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5311603254292429e-06,
+      "loss": 0.6909,
+      "step": 8233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5267029169600544e-06,
+      "loss": 0.7589,
+      "step": 8234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5222519054111595e-06,
+      "loss": 0.7567,
+      "step": 8235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5178072913699605e-06,
+      "loss": 0.7235,
+      "step": 8236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5133690754229879e-06,
+      "loss": 0.7067,
+      "step": 8237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5089372581559514e-06,
+      "loss": 0.6718,
+      "step": 8238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5045118401537106e-06,
+      "loss": 0.7781,
+      "step": 8239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5000928220002652e-06,
+      "loss": 0.807,
+      "step": 8240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4956802042788042e-06,
+      "loss": 0.6808,
+      "step": 8241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4912739875716397e-06,
+      "loss": 0.8271,
+      "step": 8242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4868741724602508e-06,
+      "loss": 0.721,
+      "step": 8243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4824807595252733e-06,
+      "loss": 0.6552,
+      "step": 8244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4780937493465042e-06,
+      "loss": 0.7405,
+      "step": 8245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.473713142502875e-06,
+      "loss": 0.7055,
+      "step": 8246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4693389395724956e-06,
+      "loss": 0.6763,
+      "step": 8247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4649711411326206e-06,
+      "loss": 0.6859,
+      "step": 8248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4606097477596504e-06,
+      "loss": 0.7876,
+      "step": 8249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.456254760029152e-06,
+      "loss": 0.8096,
+      "step": 8250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4519061785158494e-06,
+      "loss": 0.6781,
+      "step": 8251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4475640037936166e-06,
+      "loss": 0.6441,
+      "step": 8252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4432282364354676e-06,
+      "loss": 0.7532,
+      "step": 8253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4388988770136003e-06,
+      "loss": 0.7571,
+      "step": 8254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4345759260993408e-06,
+      "loss": 0.7686,
+      "step": 8255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4302593842631883e-06,
+      "loss": 0.7558,
+      "step": 8256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4259492520747874e-06,
+      "loss": 0.8155,
+      "step": 8257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4216455301029275e-06,
+      "loss": 0.7833,
+      "step": 8258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4173482189155706e-06,
+      "loss": 0.813,
+      "step": 8259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.413057319079819e-06,
+      "loss": 0.7534,
+      "step": 8260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4087728311619419e-06,
+      "loss": 0.7699,
+      "step": 8261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.404494755727348e-06,
+      "loss": 0.753,
+      "step": 8262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4002230933406025e-06,
+      "loss": 0.8244,
+      "step": 8263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.395957844565432e-06,
+      "loss": 0.6671,
+      "step": 8264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3916990099647143e-06,
+      "loss": 0.7164,
+      "step": 8265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.387446590100483e-06,
+      "loss": 0.6903,
+      "step": 8266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3832005855339114e-06,
+      "loss": 0.8064,
+      "step": 8267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3789609968253402e-06,
+      "loss": 0.6619,
+      "step": 8268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3747278245342555e-06,
+      "loss": 0.7279,
+      "step": 8269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.37050106921931e-06,
+      "loss": 0.7034,
+      "step": 8270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3662807314382909e-06,
+      "loss": 0.7511,
+      "step": 8271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3620668117481472e-06,
+      "loss": 0.7329,
+      "step": 8272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3578593107049787e-06,
+      "loss": 0.7966,
+      "step": 8273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3536582288640576e-06,
+      "loss": 0.7426,
+      "step": 8274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.349463566779774e-06,
+      "loss": 0.7543,
+      "step": 8275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.345275325005696e-06,
+      "loss": 0.7541,
+      "step": 8276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3410935040945315e-06,
+      "loss": 0.774,
+      "step": 8277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3369181045981561e-06,
+      "loss": 0.7266,
+      "step": 8278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3327491270675784e-06,
+      "loss": 0.715,
+      "step": 8279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3285865720529811e-06,
+      "loss": 0.6911,
+      "step": 8280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.32443044010368e-06,
+      "loss": 0.7763,
+      "step": 8281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3202807317681532e-06,
+      "loss": 0.7543,
+      "step": 8282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.316137447594029e-06,
+      "loss": 0.7013,
+      "step": 8283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3120005881280872e-06,
+      "loss": 0.741,
+      "step": 8284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.307870153916263e-06,
+      "loss": 0.6826,
+      "step": 8285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.303746145503637e-06,
+      "loss": 0.7685,
+      "step": 8286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2996285634344519e-06,
+      "loss": 0.6619,
+      "step": 8287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2955174082520893e-06,
+      "loss": 0.7694,
+      "step": 8288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2914126804990933e-06,
+      "loss": 0.7592,
+      "step": 8289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2873143807171585e-06,
+      "loss": 0.6861,
+      "step": 8290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2832225094471295e-06,
+      "loss": 0.7952,
+      "step": 8291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2791370672289916e-06,
+      "loss": 0.7465,
+      "step": 8292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2750580546019075e-06,
+      "loss": 0.7977,
+      "step": 8293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2709854721041637e-06,
+      "loss": 0.7755,
+      "step": 8294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2669193202732188e-06,
+      "loss": 0.6933,
+      "step": 8295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.262859599645666e-06,
+      "loss": 0.774,
+      "step": 8296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2588063107572711e-06,
+      "loss": 0.6893,
+      "step": 8297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2547594541429287e-06,
+      "loss": 0.756,
+      "step": 8298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2507190303366945e-06,
+      "loss": 0.7503,
+      "step": 8299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2466850398717755e-06,
+      "loss": 0.723,
+      "step": 8300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2426574832805293e-06,
+      "loss": 0.7602,
+      "step": 8301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2386363610944585e-06,
+      "loss": 0.7191,
+      "step": 8302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2346216738442328e-06,
+      "loss": 0.7162,
+      "step": 8303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2306134220596621e-06,
+      "loss": 0.7857,
+      "step": 8304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2266116062696953e-06,
+      "loss": 0.6934,
+      "step": 8305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2226162270024543e-06,
+      "loss": 0.6828,
+      "step": 8306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.218627284785201e-06,
+      "loss": 0.6638,
+      "step": 8307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2146447801443416e-06,
+      "loss": 0.6802,
+      "step": 8308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2106687136054395e-06,
+      "loss": 0.7327,
+      "step": 8309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2066990856932137e-06,
+      "loss": 0.6668,
+      "step": 8310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2027358969315172e-06,
+      "loss": 0.7808,
+      "step": 8311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1987791478433762e-06,
+      "loss": 0.8197,
+      "step": 8312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.194828838950951e-06,
+      "loss": 0.698,
+      "step": 8313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1908849707755466e-06,
+      "loss": 0.7545,
+      "step": 8314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.186947543837641e-06,
+      "loss": 0.7281,
+      "step": 8315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1830165586568409e-06,
+      "loss": 0.6863,
+      "step": 8316
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1790920157519148e-06,
+      "loss": 0.746,
+      "step": 8317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1751739156407649e-06,
+      "loss": 0.7226,
+      "step": 8318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1712622588404665e-06,
+      "loss": 0.7692,
+      "step": 8319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.167357045867229e-06,
+      "loss": 0.6671,
+      "step": 8320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1634582772364178e-06,
+      "loss": 0.779,
+      "step": 8321
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1595659534625324e-06,
+      "loss": 0.6793,
+      "step": 8322
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.155680075059251e-06,
+      "loss": 0.7409,
+      "step": 8323
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1518006425393801e-06,
+      "loss": 0.707,
+      "step": 8324
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1479276564148822e-06,
+      "loss": 0.6643,
+      "step": 8325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1440611171968596e-06,
+      "loss": 0.7369,
+      "step": 8326
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1402010253955764e-06,
+      "loss": 0.7579,
+      "step": 8327
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1363473815204417e-06,
+      "loss": 0.735,
+      "step": 8328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1325001860800099e-06,
+      "loss": 0.7556,
+      "step": 8329
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.128659439581986e-06,
+      "loss": 0.7612,
+      "step": 8330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1248251425332313e-06,
+      "loss": 0.7447,
+      "step": 8331
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1209972954397462e-06,
+      "loss": 0.7177,
+      "step": 8332
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1171758988066827e-06,
+      "loss": 0.6248,
+      "step": 8333
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1133609531383481e-06,
+      "loss": 0.7487,
+      "step": 8334
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.10955245893819e-06,
+      "loss": 0.7691,
+      "step": 8335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1057504167088063e-06,
+      "loss": 0.7047,
+      "step": 8336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1019548269519452e-06,
+      "loss": 0.71,
+      "step": 8337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.098165690168501e-06,
+      "loss": 0.6907,
+      "step": 8338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0943830068585181e-06,
+      "loss": 0.8154,
+      "step": 8339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.090606777521197e-06,
+      "loss": 0.7528,
+      "step": 8340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0868370026548669e-06,
+      "loss": 0.7485,
+      "step": 8341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0830736827570244e-06,
+      "loss": 0.6968,
+      "step": 8342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0793168183243052e-06,
+      "loss": 0.6984,
+      "step": 8343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0755664098524965e-06,
+      "loss": 0.6818,
+      "step": 8344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0718224578365298e-06,
+      "loss": 0.7812,
+      "step": 8345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0680849627704936e-06,
+      "loss": 0.7223,
+      "step": 8346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0643539251476041e-06,
+      "loss": 0.7119,
+      "step": 8347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0606293454602456e-06,
+      "loss": 0.7012,
+      "step": 8348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0569112241999468e-06,
+      "loss": 0.6549,
+      "step": 8349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.053199561857382e-06,
+      "loss": 0.8629,
+      "step": 8350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0494943589223538e-06,
+      "loss": 0.775,
+      "step": 8351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0457956158838544e-06,
+      "loss": 0.7643,
+      "step": 8352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.042103333229988e-06,
+      "loss": 0.752,
+      "step": 8353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0384175114480143e-06,
+      "loss": 0.7522,
+      "step": 8354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0347381510243448e-06,
+      "loss": 0.7517,
+      "step": 8355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0310652524445408e-06,
+      "loss": 0.7754,
+      "step": 8356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0273988161933036e-06,
+      "loss": 0.7355,
+      "step": 8357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0237388427544903e-06,
+      "loss": 0.7534,
+      "step": 8358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0200853326110927e-06,
+      "loss": 0.811,
+      "step": 8359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0164382862452638e-06,
+      "loss": 0.6718,
+      "step": 8360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0127977041382908e-06,
+      "loss": 0.6813,
+      "step": 8361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0091635867706173e-06,
+      "loss": 0.7389,
+      "step": 8362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0055359346218262e-06,
+      "loss": 0.8076,
+      "step": 8363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0019147481706625e-06,
+      "loss": 0.7646,
+      "step": 8364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.983000278949939e-07,
+      "loss": 0.7534,
+      "step": 8365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.946917742718497e-07,
+      "loss": 0.755,
+      "step": 8366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.910899877774106e-07,
+      "loss": 0.6936,
+      "step": 8367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.874946688869902e-07,
+      "loss": 0.7736,
+      "step": 8368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.839058180750593e-07,
+      "loss": 0.7686,
+      "step": 8369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.80323435815228e-07,
+      "loss": 0.5867,
+      "step": 8370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.767475225802514e-07,
+      "loss": 0.6199,
+      "step": 8371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.731780788420464e-07,
+      "loss": 0.7164,
+      "step": 8372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.69615105071664e-07,
+      "loss": 0.7557,
+      "step": 8373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.660586017392948e-07,
+      "loss": 0.8125,
+      "step": 8374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.625085693142854e-07,
+      "loss": 0.7301,
+      "step": 8375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.58965008265128e-07,
+      "loss": 0.6744,
+      "step": 8376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.554279190594595e-07,
+      "loss": 0.6747,
+      "step": 8377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.51897302164062e-07,
+      "loss": 0.7669,
+      "step": 8378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.483731580448629e-07,
+      "loss": 0.6207,
+      "step": 8379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.448554871669346e-07,
+      "loss": 0.8066,
+      "step": 8380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.413442899944946e-07,
+      "loss": 0.7311,
+      "step": 8381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.378395669909112e-07,
+      "loss": 0.7329,
+      "step": 8382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.343413186187033e-07,
+      "loss": 0.6689,
+      "step": 8383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.308495453395183e-07,
+      "loss": 0.7421,
+      "step": 8384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.273642476141542e-07,
+      "loss": 0.7352,
+      "step": 8385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.238854259025709e-07,
+      "loss": 0.8237,
+      "step": 8386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.204130806638511e-07,
+      "loss": 0.6908,
+      "step": 8387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.169472123562395e-07,
+      "loss": 0.7539,
+      "step": 8388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.134878214371201e-07,
+      "loss": 0.7358,
+      "step": 8389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.100349083630167e-07,
+      "loss": 0.7495,
+      "step": 8390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.06588473589598e-07,
+      "loss": 0.658,
+      "step": 8391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.031485175716947e-07,
+      "loss": 0.8097,
+      "step": 8392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.997150407632659e-07,
+      "loss": 0.76,
+      "step": 8393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.962880436174159e-07,
+      "loss": 0.8151,
+      "step": 8394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.928675265864105e-07,
+      "loss": 0.7324,
+      "step": 8395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.894534901216334e-07,
+      "loss": 0.6972,
+      "step": 8396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.86045934673635e-07,
+      "loss": 0.7357,
+      "step": 8397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.826448606921056e-07,
+      "loss": 0.713,
+      "step": 8398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.792502686258752e-07,
+      "loss": 0.7645,
+      "step": 8399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.758621589229243e-07,
+      "loss": 0.7323,
+      "step": 8400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.724805320303619e-07,
+      "loss": 0.7061,
+      "step": 8401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.691053883944699e-07,
+      "loss": 0.7549,
+      "step": 8402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.657367284606532e-07,
+      "loss": 0.6538,
+      "step": 8403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.623745526734617e-07,
+      "loss": 0.6957,
+      "step": 8404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.590188614766015e-07,
+      "loss": 0.7385,
+      "step": 8405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.556696553129184e-07,
+      "loss": 0.7823,
+      "step": 8406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.523269346243978e-07,
+      "loss": 0.8353,
+      "step": 8407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.489906998521646e-07,
+      "loss": 0.7456,
+      "step": 8408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.456609514365054e-07,
+      "loss": 0.672,
+      "step": 8409
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.423376898168245e-07,
+      "loss": 0.5913,
+      "step": 8410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.390209154317041e-07,
+      "loss": 0.6519,
+      "step": 8411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.357106287188443e-07,
+      "loss": 0.7631,
+      "step": 8412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.324068301150955e-07,
+      "loss": 0.8085,
+      "step": 8413
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.291095200564591e-07,
+      "loss": 0.7136,
+      "step": 8414
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.258186989780648e-07,
+      "loss": 0.7213,
+      "step": 8415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.225343673141983e-07,
+      "loss": 0.6462,
+      "step": 8416
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.192565254982909e-07,
+      "loss": 0.7382,
+      "step": 8417
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.159851739629076e-07,
+      "loss": 0.7191,
+      "step": 8418
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.127203131397587e-07,
+      "loss": 0.7548,
+      "step": 8419
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.094619434597162e-07,
+      "loss": 0.7839,
+      "step": 8420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.062100653527638e-07,
+      "loss": 0.5892,
+      "step": 8421
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.029646792480527e-07,
+      "loss": 0.8087,
+      "step": 8422
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.997257855738682e-07,
+      "loss": 0.6959,
+      "step": 8423
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.964933847576405e-07,
+      "loss": 0.7165,
+      "step": 8424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.932674772259452e-07,
+      "loss": 0.8601,
+      "step": 8425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.900480634044971e-07,
+      "loss": 0.7201,
+      "step": 8426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.868351437181565e-07,
+      "loss": 0.8096,
+      "step": 8427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.83628718590923e-07,
+      "loss": 0.6462,
+      "step": 8428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.804287884459416e-07,
+      "loss": 0.7222,
+      "step": 8429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.772353537054966e-07,
+      "loss": 0.6691,
+      "step": 8430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.740484147910287e-07,
+      "loss": 0.8142,
+      "step": 8431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.708679721231071e-07,
+      "loss": 0.7501,
+      "step": 8432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.676940261214516e-07,
+      "loss": 0.7811,
+      "step": 8433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.645265772049159e-07,
+      "loss": 0.6881,
+      "step": 8434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.613656257915047e-07,
+      "loss": 0.6638,
+      "step": 8435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.58211172298362e-07,
+      "loss": 0.7098,
+      "step": 8436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.550632171417716e-07,
+      "loss": 0.7713,
+      "step": 8437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.51921760737162e-07,
+      "loss": 0.7281,
+      "step": 8438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.487868034991074e-07,
+      "loss": 0.673,
+      "step": 8439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.456583458413269e-07,
+      "loss": 0.6444,
+      "step": 8440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.425363881766622e-07,
+      "loss": 0.7355,
+      "step": 8441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.394209309171285e-07,
+      "loss": 0.7464,
+      "step": 8442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.363119744738578e-07,
+      "loss": 0.7764,
+      "step": 8443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.332095192571331e-07,
+      "loss": 0.7559,
+      "step": 8444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.301135656763769e-07,
+      "loss": 0.6749,
+      "step": 8445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.270241141401569e-07,
+      "loss": 0.7576,
+      "step": 8446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.239411650561856e-07,
+      "loss": 0.6708,
+      "step": 8447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.2086471883131e-07,
+      "loss": 0.6653,
+      "step": 8448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.17794775871522e-07,
+      "loss": 0.7445,
+      "step": 8449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.14731336581953e-07,
+      "loss": 0.7496,
+      "step": 8450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.116744013668852e-07,
+      "loss": 0.8073,
+      "step": 8451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.086239706297293e-07,
+      "loss": 0.737,
+      "step": 8452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.05580044773052e-07,
+      "loss": 0.7511,
+      "step": 8453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.025426241985488e-07,
+      "loss": 0.6591,
+      "step": 8454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.995117093070602e-07,
+      "loss": 0.7745,
+      "step": 8455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.964873004985717e-07,
+      "loss": 0.7011,
+      "step": 8456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.93469398172214e-07,
+      "loss": 0.7799,
+      "step": 8457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.904580027262409e-07,
+      "loss": 0.7687,
+      "step": 8458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.874531145580731e-07,
+      "loss": 0.7126,
+      "step": 8459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.844547340642493e-07,
+      "loss": 0.7828,
+      "step": 8460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.814628616404639e-07,
+      "loss": 0.7696,
+      "step": 8461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.784774976815511e-07,
+      "loss": 0.73,
+      "step": 8462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.754986425814791e-07,
+      "loss": 0.6776,
+      "step": 8463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.725262967333667e-07,
+      "loss": 0.7116,
+      "step": 8464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.695604605294559e-07,
+      "loss": 0.6525,
+      "step": 8465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.666011343611556e-07,
+      "loss": 0.6572,
+      "step": 8466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.636483186189923e-07,
+      "loss": 0.7309,
+      "step": 8467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.60702013692649e-07,
+      "loss": 0.6803,
+      "step": 8468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.577622199709365e-07,
+      "loss": 0.6845,
+      "step": 8469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.548289378418171e-07,
+      "loss": 0.7422,
+      "step": 8470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.519021676923975e-07,
+      "loss": 0.7458,
+      "step": 8471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.489819099089078e-07,
+      "loss": 0.7698,
+      "step": 8472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.460681648767341e-07,
+      "loss": 0.7382,
+      "step": 8473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.431609329803912e-07,
+      "loss": 0.7078,
+      "step": 8474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.402602146035385e-07,
+      "loss": 0.6475,
+      "step": 8475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.373660101289869e-07,
+      "loss": 0.7373,
+      "step": 8476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.344783199386806e-07,
+      "loss": 0.7273,
+      "step": 8477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.315971444136925e-07,
+      "loss": 0.7344,
+      "step": 8478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.28722483934241e-07,
+      "loss": 0.7094,
+      "step": 8479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.258543388797055e-07,
+      "loss": 0.7034,
+      "step": 8480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.229927096285781e-07,
+      "loss": 0.7358,
+      "step": 8481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.201375965585065e-07,
+      "loss": 0.7227,
+      "step": 8482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.172890000462728e-07,
+      "loss": 0.6957,
+      "step": 8483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.144469204677982e-07,
+      "loss": 0.7936,
+      "step": 8484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.116113581981497e-07,
+      "loss": 0.7113,
+      "step": 8485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.087823136115278e-07,
+      "loss": 0.7713,
+      "step": 8486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.059597870812727e-07,
+      "loss": 0.8319,
+      "step": 8487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.031437789798755e-07,
+      "loss": 0.7913,
+      "step": 8488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.003342896789499e-07,
+      "loss": 0.6328,
+      "step": 8489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.975313195492604e-07,
+      "loss": 0.7546,
+      "step": 8490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.947348689607168e-07,
+      "loss": 0.7055,
+      "step": 8491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.919449382823572e-07,
+      "loss": 0.7715,
+      "step": 8492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.891615278823537e-07,
+      "loss": 0.7416,
+      "step": 8493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.863846381280402e-07,
+      "loss": 0.7457,
+      "step": 8494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.836142693858737e-07,
+      "loss": 0.7317,
+      "step": 8495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.808504220214505e-07,
+      "loss": 0.7008,
+      "step": 8496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.780930963995124e-07,
+      "loss": 0.7032,
+      "step": 8497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.753422928839347e-07,
+      "loss": 0.6823,
+      "step": 8498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.725980118377328e-07,
+      "loss": 0.7321,
+      "step": 8499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.698602536230779e-07,
+      "loss": 0.7899,
+      "step": 8500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.671290186012479e-07,
+      "loss": 0.7428,
+      "step": 8501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.644043071326932e-07,
+      "loss": 0.7301,
+      "step": 8502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.616861195769818e-07,
+      "loss": 0.6718,
+      "step": 8503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.589744562928268e-07,
+      "loss": 0.6707,
+      "step": 8504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.562693176380807e-07,
+      "loss": 0.6929,
+      "step": 8505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.535707039697357e-07,
+      "loss": 0.7718,
+      "step": 8506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.508786156439239e-07,
+      "loss": 0.6237,
+      "step": 8507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.481930530159163e-07,
+      "loss": 0.795,
+      "step": 8508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.455140164401073e-07,
+      "loss": 0.7464,
+      "step": 8509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.428415062700643e-07,
+      "loss": 0.7496,
+      "step": 8510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.401755228584604e-07,
+      "loss": 0.7723,
+      "step": 8511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.375160665571144e-07,
+      "loss": 0.6945,
+      "step": 8512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.348631377170066e-07,
+      "loss": 0.6845,
+      "step": 8513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.322167366882291e-07,
+      "loss": 0.725,
+      "step": 8514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.295768638200194e-07,
+      "loss": 0.635,
+      "step": 8515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.269435194607598e-07,
+      "loss": 0.5908,
+      "step": 8516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.243167039579667e-07,
+      "loss": 0.7924,
+      "step": 8517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.216964176582905e-07,
+      "loss": 0.7594,
+      "step": 8518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.190826609075327e-07,
+      "loss": 0.8779,
+      "step": 8519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.164754340506228e-07,
+      "loss": 0.7722,
+      "step": 8520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.138747374316299e-07,
+      "loss": 0.736,
+      "step": 8521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.11280571393763e-07,
+      "loss": 0.7208,
+      "step": 8522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.086929362793702e-07,
+      "loss": 0.8592,
+      "step": 8523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.061118324299286e-07,
+      "loss": 0.7551,
+      "step": 8524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.035372601860766e-07,
+      "loss": 0.6848,
+      "step": 8525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.009692198875593e-07,
+      "loss": 0.7312,
+      "step": 8526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.984077118732833e-07,
+      "loss": 0.7137,
+      "step": 8527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.958527364812838e-07,
+      "loss": 0.7959,
+      "step": 8528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.933042940487409e-07,
+      "loss": 0.788,
+      "step": 8529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.907623849119525e-07,
+      "loss": 0.6723,
+      "step": 8530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.88227009406389e-07,
+      "loss": 0.764,
+      "step": 8531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.856981678666272e-07,
+      "loss": 0.7754,
+      "step": 8532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.831758606263947e-07,
+      "loss": 0.6501,
+      "step": 8533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.80660088018553e-07,
+      "loss": 0.6859,
+      "step": 8534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.781508503751087e-07,
+      "loss": 0.7048,
+      "step": 8535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.756481480271968e-07,
+      "loss": 0.8057,
+      "step": 8536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7315198130509217e-07,
+      "loss": 0.7085,
+      "step": 8537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.706623505382091e-07,
+      "loss": 0.7555,
+      "step": 8538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6817925605510124e-07,
+      "loss": 0.7432,
+      "step": 8539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6570269818346224e-07,
+      "loss": 0.7422,
+      "step": 8540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6323267725011387e-07,
+      "loss": 0.6936,
+      "step": 8541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6076919358101214e-07,
+      "loss": 0.6357,
+      "step": 8542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.583122475012691e-07,
+      "loss": 0.736,
+      "step": 8543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5586183933511996e-07,
+      "loss": 0.6793,
+      "step": 8544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5341796940593375e-07,
+      "loss": 0.5914,
+      "step": 8545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5098063803623026e-07,
+      "loss": 0.7556,
+      "step": 8546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.485498455476522e-07,
+      "loss": 0.7908,
+      "step": 8547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.461255922609986e-07,
+      "loss": 0.6237,
+      "step": 8548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4370787849617456e-07,
+      "loss": 0.7603,
+      "step": 8549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4129670457225823e-07,
+      "loss": 0.8239,
+      "step": 8550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3889207080743953e-07,
+      "loss": 0.6632,
+      "step": 8551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.364939775190535e-07,
+      "loss": 0.7785,
+      "step": 8552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3410242502356923e-07,
+      "loss": 0.641,
+      "step": 8553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.317174136366009e-07,
+      "loss": 0.7727,
+      "step": 8554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2933894367288566e-07,
+      "loss": 0.7467,
+      "step": 8555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2696701544631123e-07,
+      "loss": 0.747,
+      "step": 8556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.24601629269894e-07,
+      "loss": 0.7711,
+      "step": 8557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2224278545578974e-07,
+      "loss": 0.6878,
+      "step": 8558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.198904843152884e-07,
+      "loss": 0.8222,
+      "step": 8559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1754472615881943e-07,
+      "loss": 0.6551,
+      "step": 8560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.152055112959519e-07,
+      "loss": 0.7383,
+      "step": 8561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1287284003538316e-07,
+      "loss": 0.7418,
+      "step": 8562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1054671268494495e-07,
+      "loss": 0.6632,
+      "step": 8563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0822712955161937e-07,
+      "loss": 0.7858,
+      "step": 8564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.059140909415171e-07,
+      "loss": 0.7193,
+      "step": 8565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0360759715988294e-07,
+      "loss": 0.6886,
+      "step": 8566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0130764851110114e-07,
+      "loss": 0.8179,
+      "step": 8567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.990142452986845e-07,
+      "loss": 0.6813,
+      "step": 8568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.967273878252964e-07,
+      "loss": 0.8041,
+      "step": 8569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.944470763927288e-07,
+      "loss": 0.7549,
+      "step": 8570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9217331130190773e-07,
+      "loss": 0.6287,
+      "step": 8571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.899060928528986e-07,
+      "loss": 0.7089,
+      "step": 8572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.876454213448955e-07,
+      "loss": 0.6718,
+      "step": 8573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.85391297076243e-07,
+      "loss": 0.7038,
+      "step": 8574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.831437203444088e-07,
+      "loss": 0.7735,
+      "step": 8575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.809026914460001e-07,
+      "loss": 0.7539,
+      "step": 8576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7866821067675805e-07,
+      "loss": 0.6519,
+      "step": 8577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7644027833156904e-07,
+      "loss": 0.7894,
+      "step": 8578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.742188947044478e-07,
+      "loss": 0.703,
+      "step": 8579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7200406008854327e-07,
+      "loss": 0.7847,
+      "step": 8580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.697957747761438e-07,
+      "loss": 0.6273,
+      "step": 8581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6759403905867187e-07,
+      "loss": 0.6578,
+      "step": 8582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6539885322668386e-07,
+      "loss": 0.7385,
+      "step": 8583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6321021756987596e-07,
+      "loss": 0.638,
+      "step": 8584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.610281323770781e-07,
+      "loss": 0.7732,
+      "step": 8585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.588525979362545e-07,
+      "loss": 0.8084,
+      "step": 8586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5668361453450317e-07,
+      "loss": 0.7352,
+      "step": 8587
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5452118245806745e-07,
+      "loss": 0.669,
+      "step": 8588
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.52365301992319e-07,
+      "loss": 0.7404,
+      "step": 8589
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5021597342175806e-07,
+      "loss": 0.8092,
+      "step": 8590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4807319703002993e-07,
+      "loss": 0.7795,
+      "step": 8591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4593697309990845e-07,
+      "loss": 0.8191,
+      "step": 8592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.438073019133181e-07,
+      "loss": 0.7416,
+      "step": 8593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.416841837512952e-07,
+      "loss": 0.8121,
+      "step": 8594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3956761889402666e-07,
+      "loss": 0.5882,
+      "step": 8595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.374576076208391e-07,
+      "loss": 0.7524,
+      "step": 8596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.353541502101709e-07,
+      "loss": 0.7173,
+      "step": 8597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3325724693962225e-07,
+      "loss": 0.834,
+      "step": 8598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.311668980859162e-07,
+      "loss": 0.7025,
+      "step": 8599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.290831039249098e-07,
+      "loss": 0.6675,
+      "step": 8600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.270058647315943e-07,
+      "loss": 0.7319,
+      "step": 8601
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.24935180780106e-07,
+      "loss": 0.831,
+      "step": 8602
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.22871052343704e-07,
+      "loss": 0.7334,
+      "step": 8603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.20813479694787e-07,
+      "loss": 0.7111,
+      "step": 8604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.18762463104888e-07,
+      "loss": 0.7081,
+      "step": 8605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1671800284467925e-07,
+      "loss": 0.7473,
+      "step": 8606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.14680099183956e-07,
+      "loss": 0.7178,
+      "step": 8607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1264875239166434e-07,
+      "loss": 0.6432,
+      "step": 8608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1062396273587314e-07,
+      "loss": 0.6955,
+      "step": 8609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0860573048379084e-07,
+      "loss": 0.6679,
+      "step": 8610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0659405590175994e-07,
+      "loss": 0.8148,
+      "step": 8611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0458893925525144e-07,
+      "loss": 0.7375,
+      "step": 8612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0259038080888704e-07,
+      "loss": 0.7027,
+      "step": 8613
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.005983808264001e-07,
+      "loss": 0.8502,
+      "step": 8614
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9861293957068047e-07,
+      "loss": 0.7302,
+      "step": 8615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.966340573037407e-07,
+      "loss": 0.7781,
+      "step": 8616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9466173428672193e-07,
+      "loss": 0.7413,
+      "step": 8617
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.92695970779916e-07,
+      "loss": 0.8379,
+      "step": 8618
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9073676704273746e-07,
+      "loss": 0.7114,
+      "step": 8619
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.887841233337407e-07,
+      "loss": 0.7512,
+      "step": 8620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8683803991061407e-07,
+      "loss": 0.7454,
+      "step": 8621
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.848985170301688e-07,
+      "loss": 0.6792,
+      "step": 8622
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8296555494836675e-07,
+      "loss": 0.75,
+      "step": 8623
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.810391539202983e-07,
+      "loss": 0.7858,
+      "step": 8624
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7911931420018245e-07,
+      "loss": 0.7675,
+      "step": 8625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7720603604138306e-07,
+      "loss": 0.7846,
+      "step": 8626
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.75299319696376e-07,
+      "loss": 0.6809,
+      "step": 8627
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.733991654168044e-07,
+      "loss": 0.7067,
+      "step": 8628
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7150557345341775e-07,
+      "loss": 0.7137,
+      "step": 8629
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.696185440561161e-07,
+      "loss": 0.7845,
+      "step": 8630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.677380774739169e-07,
+      "loss": 0.6924,
+      "step": 8631
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.658641739549883e-07,
+      "loss": 0.719,
+      "step": 8632
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6399683374662697e-07,
+      "loss": 0.7417,
+      "step": 8633
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.62136057095258e-07,
+      "loss": 0.7638,
+      "step": 8634
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.602818442464516e-07,
+      "loss": 0.6817,
+      "step": 8635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.584341954448899e-07,
+      "loss": 0.7518,
+      "step": 8636
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5659311093441107e-07,
+      "loss": 0.7123,
+      "step": 8637
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5475859095798194e-07,
+      "loss": 0.7286,
+      "step": 8638
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5293063575770327e-07,
+      "loss": 0.6555,
+      "step": 8639
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.511092455747932e-07,
+      "loss": 0.6504,
+      "step": 8640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4929442064963173e-07,
+      "loss": 0.791,
+      "step": 8641
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4748616122171055e-07,
+      "loss": 0.8045,
+      "step": 8642
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.45684467529661e-07,
+      "loss": 0.832,
+      "step": 8643
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.438893398112485e-07,
+      "loss": 0.7181,
+      "step": 8644
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4210077830337797e-07,
+      "loss": 0.6567,
+      "step": 8645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.403187832420717e-07,
+      "loss": 0.73,
+      "step": 8646
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.385433548625027e-07,
+      "loss": 0.6961,
+      "step": 8647
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3677449339897794e-07,
+      "loss": 0.7488,
+      "step": 8648
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3501219908491634e-07,
+      "loss": 0.8037,
+      "step": 8649
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3325647215289291e-07,
+      "loss": 0.6511,
+      "step": 8650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3150731283460015e-07,
+      "loss": 0.7537,
+      "step": 8651
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2976472136088113e-07,
+      "loss": 0.6713,
+      "step": 8652
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2802869796169635e-07,
+      "loss": 0.7724,
+      "step": 8653
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2629924286614035e-07,
+      "loss": 0.6538,
+      "step": 8654
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2457635630245832e-07,
+      "loss": 0.7231,
+      "step": 8655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2286003849800175e-07,
+      "loss": 0.8018,
+      "step": 8656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2115028967927832e-07,
+      "loss": 0.8273,
+      "step": 8657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1944711007191865e-07,
+      "loss": 0.7583,
+      "step": 8658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1775049990068741e-07,
+      "loss": 0.674,
+      "step": 8659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1606045938948326e-07,
+      "loss": 0.7303,
+      "step": 8660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1437698876133338e-07,
+      "loss": 0.6887,
+      "step": 8661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1270008823841004e-07,
+      "loss": 0.6988,
+      "step": 8662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1102975804200286e-07,
+      "loss": 0.7226,
+      "step": 8663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0936599839254113e-07,
+      "loss": 0.7194,
+      "step": 8664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.077088095095936e-07,
+      "loss": 0.7391,
+      "step": 8665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0605819161185204e-07,
+      "loss": 0.7484,
+      "step": 8666
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0441414491714773e-07,
+      "loss": 0.7739,
+      "step": 8667
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0277666964244046e-07,
+      "loss": 0.8153,
+      "step": 8668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0114576600382405e-07,
+      "loss": 0.7968,
+      "step": 8669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9952143421653192e-07,
+      "loss": 0.8064,
+      "step": 8670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.979036744949092e-07,
+      "loss": 0.6391,
+      "step": 8671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9629248705246296e-07,
+      "loss": 0.6841,
+      "step": 8672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9468787210181195e-07,
+      "loss": 0.7781,
+      "step": 8673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9308982985470903e-07,
+      "loss": 0.7358,
+      "step": 8674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9149836052205771e-07,
+      "loss": 0.6767,
+      "step": 8675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.899134643138678e-07,
+      "loss": 0.8896,
+      "step": 8676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8833514143929975e-07,
+      "loss": 0.7848,
+      "step": 8677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8676339210664252e-07,
+      "loss": 0.6551,
+      "step": 8678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8519821652331904e-07,
+      "loss": 0.7332,
+      "step": 8679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8363961489587522e-07,
+      "loss": 0.7822,
+      "step": 8680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.820875874300021e-07,
+      "loss": 0.7417,
+      "step": 8681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.805421343305136e-07,
+      "loss": 0.7414,
+      "step": 8682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7900325580136324e-07,
+      "loss": 0.8207,
+      "step": 8683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7747095204563303e-07,
+      "loss": 0.731,
+      "step": 8684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7594522326553897e-07,
+      "loss": 0.7184,
+      "step": 8685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7442606966242004e-07,
+      "loss": 0.7355,
+      "step": 8686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.729134914367714e-07,
+      "loss": 0.7637,
+      "step": 8687
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.714074887881889e-07,
+      "loss": 0.7589,
+      "step": 8688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.699080619154303e-07,
+      "loss": 0.8009,
+      "step": 8689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6841521101635948e-07,
+      "loss": 0.7965,
+      "step": 8690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.669289362879911e-07,
+      "loss": 0.6795,
+      "step": 8691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6544923792646828e-07,
+      "loss": 0.7291,
+      "step": 8692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.63976116127057e-07,
+      "loss": 0.7616,
+      "step": 8693
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6250957108416842e-07,
+      "loss": 0.7867,
+      "step": 8694
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6104960299133109e-07,
+      "loss": 0.768,
+      "step": 8695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5959621204122422e-07,
+      "loss": 0.727,
+      "step": 8696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5814939842564436e-07,
+      "loss": 0.6764,
+      "step": 8697
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5670916233552768e-07,
+      "loss": 0.7735,
+      "step": 8698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5527550396093327e-07,
+      "loss": 0.8153,
+      "step": 8699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.538484234910542e-07,
+      "loss": 0.8257,
+      "step": 8700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5242792111423433e-07,
+      "loss": 0.696,
+      "step": 8701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5101399701792363e-07,
+      "loss": 0.7251,
+      "step": 8702
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4960665138871176e-07,
+      "loss": 0.742,
+      "step": 8703
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4820588441233352e-07,
+      "loss": 0.6466,
+      "step": 8704
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.46811696273641e-07,
+      "loss": 0.7531,
+      "step": 8705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4542408715661482e-07,
+      "loss": 0.7505,
+      "step": 8706
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4404305724438627e-07,
+      "loss": 0.7078,
+      "step": 8707
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4266860671920402e-07,
+      "loss": 0.6969,
+      "step": 8708
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.413007357624452e-07,
+      "loss": 0.7945,
+      "step": 8709
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.399394445546265e-07,
+      "loss": 0.6624,
+      "step": 8710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3858473327540422e-07,
+      "loss": 0.652,
+      "step": 8711
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.372366021035465e-07,
+      "loss": 0.7697,
+      "step": 8712
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3589505121696655e-07,
+      "loss": 0.7592,
+      "step": 8713
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3456008079271165e-07,
+      "loss": 0.7474,
+      "step": 8714
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3323169100694088e-07,
+      "loss": 0.72,
+      "step": 8715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3190988203497512e-07,
+      "loss": 0.6999,
+      "step": 8716
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.305946540512415e-07,
+      "loss": 0.7252,
+      "step": 8717
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2928600722930673e-07,
+      "loss": 0.6999,
+      "step": 8718
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.279839417418771e-07,
+      "loss": 0.7865,
+      "step": 8719
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2668845776078185e-07,
+      "loss": 0.692,
+      "step": 8720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2539955545697867e-07,
+      "loss": 0.7948,
+      "step": 8721
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2411723500056482e-07,
+      "loss": 0.7209,
+      "step": 8722
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2284149656076605e-07,
+      "loss": 0.678,
+      "step": 8723
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.21572340305931e-07,
+      "loss": 0.7838,
+      "step": 8724
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2030976640355908e-07,
+      "loss": 0.6784,
+      "step": 8725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1905377502026694e-07,
+      "loss": 0.7688,
+      "step": 8726
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1780436632179981e-07,
+      "loss": 0.5764,
+      "step": 8727
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1656154047303691e-07,
+      "loss": 0.7802,
+      "step": 8728
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1532529763800814e-07,
+      "loss": 0.72,
+      "step": 8729
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1409563797983858e-07,
+      "loss": 0.7436,
+      "step": 8730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1287256166081506e-07,
+      "loss": 0.6564,
+      "step": 8731
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1165606884234181e-07,
+      "loss": 0.787,
+      "step": 8732
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1044615968495709e-07,
+      "loss": 0.7122,
+      "step": 8733
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0924283434832761e-07,
+      "loss": 0.6766,
+      "step": 8734
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0804609299125413e-07,
+      "loss": 0.6723,
+      "step": 8735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0685593577167141e-07,
+      "loss": 0.7642,
+      "step": 8736
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0567236284664273e-07,
+      "loss": 0.758,
+      "step": 8737
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0449537437235979e-07,
+      "loss": 0.7671,
+      "step": 8738
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.033249705041428e-07,
+      "loss": 0.7192,
+      "step": 8739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0216115139645155e-07,
+      "loss": 0.8354,
+      "step": 8740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0100391720287428e-07,
+      "loss": 0.7618,
+      "step": 8741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.985326807612772e-08,
+      "loss": 0.8083,
+      "step": 8742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.870920416805707e-08,
+      "loss": 0.7556,
+      "step": 8743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.757172562964712e-08,
+      "loss": 0.672,
+      "step": 8744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.644083261100556e-08,
+      "loss": 0.7426,
+      "step": 8745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.531652526137968e-08,
+      "loss": 0.7019,
+      "step": 8746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.419880372913415e-08,
+      "loss": 0.6991,
+      "step": 8747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.308766816177316e-08,
+      "loss": 0.7421,
+      "step": 8748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.198311870593501e-08,
+      "loss": 0.8054,
+      "step": 8749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.088515550738641e-08,
+      "loss": 0.7111,
+      "step": 8750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.979377871101702e-08,
+      "loss": 0.6842,
+      "step": 8751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.870898846085607e-08,
+      "loss": 0.8482,
+      "step": 8752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.763078490006127e-08,
+      "loss": 0.7596,
+      "step": 8753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.65591681709188e-08,
+      "loss": 0.7526,
+      "step": 8754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.549413841485443e-08,
+      "loss": 0.6905,
+      "step": 8755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.443569577241683e-08,
+      "loss": 0.7011,
+      "step": 8756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.338384038327762e-08,
+      "loss": 0.8424,
+      "step": 8757
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.233857238625354e-08,
+      "loss": 0.7709,
+      "step": 8758
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.129989191928978e-08,
+      "loss": 0.7432,
+      "step": 8759
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.026779911945448e-08,
+      "loss": 0.7064,
+      "step": 8760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.924229412294981e-08,
+      "loss": 0.8183,
+      "step": 8761
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.822337706511195e-08,
+      "loss": 0.7553,
+      "step": 8762
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.721104808040557e-08,
+      "loss": 0.7274,
+      "step": 8763
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.620530730242381e-08,
+      "loss": 0.7676,
+      "step": 8764
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.520615486389382e-08,
+      "loss": 0.7245,
+      "step": 8765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.421359089667124e-08,
+      "loss": 0.7566,
+      "step": 8766
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.322761553174573e-08,
+      "loss": 0.6546,
+      "step": 8767
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.224822889922433e-08,
+      "loss": 0.8132,
+      "step": 8768
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.127543112837031e-08,
+      "loss": 0.7276,
+      "step": 8769
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.030922234754766e-08,
+      "loss": 0.7173,
+      "step": 8770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.934960268427105e-08,
+      "loss": 0.7193,
+      "step": 8771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.839657226517803e-08,
+      "loss": 0.7846,
+      "step": 8772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.745013121604026e-08,
+      "loss": 0.7472,
+      "step": 8773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.651027966175227e-08,
+      "loss": 0.7612,
+      "step": 8774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.557701772635372e-08,
+      "loss": 0.7286,
+      "step": 8775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.46503455329961e-08,
+      "loss": 0.7844,
+      "step": 8776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.373026320397602e-08,
+      "loss": 0.8083,
+      "step": 8777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.281677086071303e-08,
+      "loss": 0.784,
+      "step": 8778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.190986862376069e-08,
+      "loss": 0.8011,
+      "step": 8779
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.100955661279551e-08,
+      "loss": 0.6986,
+      "step": 8780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.011583494663353e-08,
+      "loss": 0.6599,
+      "step": 8781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.922870374321376e-08,
+      "loss": 0.6574,
+      "step": 8782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.834816311961477e-08,
+      "loss": 0.7761,
+      "step": 8783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.747421319203805e-08,
+      "loss": 0.7893,
+      "step": 8784
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6606854075819115e-08,
+      "loss": 0.6166,
+      "step": 8785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.574608588541641e-08,
+      "loss": 0.655,
+      "step": 8786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.48919087344224e-08,
+      "loss": 0.712,
+      "step": 8787
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.404432273556914e-08,
+      "loss": 0.6465,
+      "step": 8788
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3203328000706044e-08,
+      "loss": 0.6955,
+      "step": 8789
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2368924640816555e-08,
+      "loss": 0.6688,
+      "step": 8790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.154111276601259e-08,
+      "loss": 0.6973,
+      "step": 8791
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.07198924855512e-08,
+      "loss": 0.7888,
+      "step": 8792
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9905263907790154e-08,
+      "loss": 0.7593,
+      "step": 8793
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.909722714024345e-08,
+      "loss": 0.6985,
+      "step": 8794
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.829578228954801e-08,
+      "loss": 0.6362,
+      "step": 8795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.750092946146367e-08,
+      "loss": 0.7986,
+      "step": 8796
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6712668760889865e-08,
+      "loss": 0.7736,
+      "step": 8797
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.593100029184894e-08,
+      "loss": 0.753,
+      "step": 8798
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.51559241574917e-08,
+      "loss": 0.7814,
+      "step": 8799
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4387440460114114e-08,
+      "loss": 0.6707,
+      "step": 8800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3625549301123946e-08,
+      "loss": 0.7679,
+      "step": 8801
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.287025078106854e-08,
+      "loss": 0.7574,
+      "step": 8802
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2121544999618176e-08,
+      "loss": 0.7592,
+      "step": 8803
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.137943205558825e-08,
+      "loss": 0.6303,
+      "step": 8804
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0643912046905986e-08,
+      "loss": 0.6727,
+      "step": 8805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9914985070638176e-08,
+      "loss": 0.7252,
+      "step": 8806
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9192651222980103e-08,
+      "loss": 0.7948,
+      "step": 8807
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.847691059925551e-08,
+      "loss": 0.7568,
+      "step": 8808
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.776776329392218e-08,
+      "loss": 0.6545,
+      "step": 8809
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7065209400560794e-08,
+      "loss": 0.6992,
+      "step": 8810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.636924901189165e-08,
+      "loss": 0.8038,
+      "step": 8811
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.567988221975238e-08,
+      "loss": 0.7411,
+      "step": 8812
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.499710911512022e-08,
+      "loss": 0.7007,
+      "step": 8813
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.432092978810086e-08,
+      "loss": 0.7535,
+      "step": 8814
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.365134432792849e-08,
+      "loss": 0.7237,
+      "step": 8815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2988352822965754e-08,
+      "loss": 0.7597,
+      "step": 8816
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.233195536070377e-08,
+      "loss": 0.7209,
+      "step": 8817
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.16821520277677e-08,
+      "loss": 0.9114,
+      "step": 8818
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.103894290991116e-08,
+      "loss": 0.6699,
+      "step": 8819
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.040232809201626e-08,
+      "loss": 0.7187,
+      "step": 8820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.977230765809913e-08,
+      "loss": 0.8048,
+      "step": 8821
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9148881691298813e-08,
+      "loss": 0.6242,
+      "step": 8822
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8532050273888388e-08,
+      "loss": 0.6943,
+      "step": 8823
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.792181348726941e-08,
+      "loss": 0.7563,
+      "step": 8824
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7318171411977455e-08,
+      "loss": 0.7781,
+      "step": 8825
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6721124127665476e-08,
+      "loss": 0.7792,
+      "step": 8826
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.613067171313155e-08,
+      "loss": 0.7228,
+      "step": 8827
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.554681424629668e-08,
+      "loss": 0.7131,
+      "step": 8828
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4969551804204795e-08,
+      "loss": 0.7195,
+      "step": 8829
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4398884463044945e-08,
+      "loss": 0.7628,
+      "step": 8830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3834812298118015e-08,
+      "loss": 0.7998,
+      "step": 8831
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3277335383870003e-08,
+      "loss": 0.7851,
+      "step": 8832
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.272645379386429e-08,
+      "loss": 0.8086,
+      "step": 8833
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2182167600809377e-08,
+      "loss": 0.8234,
+      "step": 8834
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.16444768765256e-08,
+      "loss": 0.7109,
+      "step": 8835
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1113381691967305e-08,
+      "loss": 0.641,
+      "step": 8836
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0588882117233978e-08,
+      "loss": 0.8163,
+      "step": 8837
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0070978221531367e-08,
+      "loss": 0.7579,
+      "step": 8838
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9559670073210356e-08,
+      "loss": 0.6993,
+      "step": 8839
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9054957739750302e-08,
+      "loss": 0.6846,
+      "step": 8840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8556841287753478e-08,
+      "loss": 0.8028,
+      "step": 8841
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8065320782956198e-08,
+      "loss": 0.7694,
+      "step": 8842
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7580396290223233e-08,
+      "loss": 0.803,
+      "step": 8843
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7102067873547846e-08,
+      "loss": 0.7165,
+      "step": 8844
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.663033559605731e-08,
+      "loss": 0.744,
+      "step": 8845
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.616519952000739e-08,
+      "loss": 0.69,
+      "step": 8846
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.570665970677121e-08,
+      "loss": 0.865,
+      "step": 8847
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5254716216867025e-08,
+      "loss": 0.7991,
+      "step": 8848
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4809369109936021e-08,
+      "loss": 0.8319,
+      "step": 8849
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4370618444753404e-08,
+      "loss": 0.6703,
+      "step": 8850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.39384642792173e-08,
+      "loss": 0.6693,
+      "step": 8851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.351290667035987e-08,
+      "loss": 0.7406,
+      "step": 8852
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3093945674336194e-08,
+      "loss": 0.7019,
+      "step": 8853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.268158134644093e-08,
+      "loss": 0.7465,
+      "step": 8854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2275813741091658e-08,
+      "loss": 0.8265,
+      "step": 8855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1876642911834435e-08,
+      "loss": 0.6867,
+      "step": 8856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1484068911349344e-08,
+      "loss": 0.6716,
+      "step": 8857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1098091791444943e-08,
+      "loss": 0.847,
+      "step": 8858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.071871160305271e-08,
+      "loss": 0.6903,
+      "step": 8859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0345928396243709e-08,
+      "loss": 0.7204,
+      "step": 8860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.979742220206368e-09,
+      "loss": 0.7892,
+      "step": 8861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.620153123274246e-09,
+      "loss": 0.7239,
+      "step": 8862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.267161152898274e-09,
+      "loss": 0.8055,
+      "step": 8863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.920766355663413e-09,
+      "loss": 0.8311,
+      "step": 8864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.580968777277543e-09,
+      "loss": 0.6841,
+      "step": 8865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.247768462588124e-09,
+      "loss": 0.7296,
+      "step": 8866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.921165455565538e-09,
+      "loss": 0.7395,
+      "step": 8867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.601159799308643e-09,
+      "loss": 0.6509,
+      "step": 8868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.287751536050324e-09,
+      "loss": 0.696,
+      "step": 8869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.980940707146389e-09,
+      "loss": 0.8364,
+      "step": 8870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.680727353097771e-09,
+      "loss": 0.703,
+      "step": 8871
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.387111513506128e-09,
+      "loss": 0.6984,
+      "step": 8872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.100093227134896e-09,
+      "loss": 0.6579,
+      "step": 8873
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.81967253184823e-09,
+      "loss": 0.8363,
+      "step": 8874
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.545849464666519e-09,
+      "loss": 0.7457,
+      "step": 8875
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.278624061710869e-09,
+      "loss": 0.7852,
+      "step": 8876
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0179963582586146e-09,
+      "loss": 0.7469,
+      "step": 8877
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.763966388693364e-09,
+      "loss": 0.6235,
+      "step": 8878
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5165341865494034e-09,
+      "loss": 0.6962,
+      "step": 8879
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.275699784472842e-09,
+      "loss": 0.6769,
+      "step": 8880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.041463214249364e-09,
+      "loss": 0.6026,
+      "step": 8881
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.813824506793129e-09,
+      "loss": 0.807,
+      "step": 8882
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5927836921356705e-09,
+      "loss": 0.6176,
+      "step": 8883
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3783407994536497e-09,
+      "loss": 0.8086,
+      "step": 8884
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1704958570522024e-09,
+      "loss": 0.5935,
+      "step": 8885
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9692488923482864e-09,
+      "loss": 0.8133,
+      "step": 8886
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7745999319095384e-09,
+      "loss": 0.7187,
+      "step": 8887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.586549001415417e-09,
+      "loss": 0.7542,
+      "step": 8888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.405096125690509e-09,
+      "loss": 0.809,
+      "step": 8889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2302413286767743e-09,
+      "loss": 0.812,
+      "step": 8890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.061984633455749e-09,
+      "loss": 0.6773,
+      "step": 8891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9003260622207918e-09,
+      "loss": 0.6849,
+      "step": 8892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7452656363103891e-09,
+      "loss": 0.666,
+      "step": 8893
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5968033761859513e-09,
+      "loss": 0.6922,
+      "step": 8894
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4549393014484658e-09,
+      "loss": 0.6899,
+      "step": 8895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3196734308051907e-09,
+      "loss": 0.6954,
+      "step": 8896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1910057821196142e-09,
+      "loss": 0.7132,
+      "step": 8897
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0689363723670464e-09,
+      "loss": 0.7642,
+      "step": 8898
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.534652176568238e-10,
+      "loss": 0.7586,
+      "step": 8899
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.445923332267569e-10,
+      "loss": 0.7976,
+      "step": 8900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.42317733443132e-10,
+      "loss": 0.7333,
+      "step": 8901
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.466414318062608e-10,
+      "loss": 0.8183,
+      "step": 8902
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.575634409393793e-10,
+      "loss": 0.7346,
+      "step": 8903
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.750837725997493e-10,
+      "loss": 0.7463,
+      "step": 8904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9920243767310737e-10,
+      "loss": 0.8051,
+      "step": 8905
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.299194461681143e-10,
+      "loss": 0.7394,
+      "step": 8906
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.672348072385589e-10,
+      "loss": 0.7686,
+      "step": 8907
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1114852914450035e-10,
+      "loss": 0.7387,
+      "step": 8908
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.616606192911263e-10,
+      "loss": 0.7316,
+      "step": 8909
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1877108421765037e-10,
+      "loss": 0.7905,
+      "step": 8910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.247992956955663e-11,
+      "loss": 0.8211,
+      "step": 8911
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.278716014855967e-11,
+      "loss": 0.6822,
+      "step": 8912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.969277986819563e-11,
+      "loss": 0.6807,
+      "step": 8913
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3196791781577844e-11,
+      "loss": 0.7324,
+      "step": 8914
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.299198053641206e-12,
+      "loss": 0.5719,
+      "step": 8915
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.8231,
+      "step": 8916
+    },
+    {
+      "epoch": 1.0,
+      "step": 8916,
+      "total_flos": 6.019905031346586e+16,
+      "train_loss": 0.8012945529937531,
+      "train_runtime": 94006.866,
+      "train_samples_per_second": 12.14,
+      "train_steps_per_second": 0.095
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 8916,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 10000,
+  "total_flos": 6.019905031346586e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}