diff --git "a/VQA_lora_pmcclip/vqa/trainer_state.json" "b/VQA_lora_pmcclip/vqa/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/VQA_lora_pmcclip/vqa/trainer_state.json"
@@ -0,0 +1,82975 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "global_step": 13825,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.819277108433736e-08,
+      "loss": 11.3687,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.638554216867472e-08,
+      "loss": 11.3743,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4457831325301206e-07,
+      "loss": 11.35,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9277108433734944e-07,
+      "loss": 11.3505,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.409638554216868e-07,
+      "loss": 11.3543,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.891566265060241e-07,
+      "loss": 11.3328,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.373493975903615e-07,
+      "loss": 11.3193,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.855421686746989e-07,
+      "loss": 11.3109,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3373493975903615e-07,
+      "loss": 11.2966,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.819277108433736e-07,
+      "loss": 11.2477,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.301204819277109e-07,
+      "loss": 11.2207,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.783132530120482e-07,
+      "loss": 11.217,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.265060240963857e-07,
+      "loss": 11.177,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.74698795180723e-07,
+      "loss": 11.1146,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.228915662650602e-07,
+      "loss": 11.0793,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.710843373493978e-07,
+      "loss": 11.0396,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.192771084337351e-07,
+      "loss": 10.9766,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.674698795180723e-07,
+      "loss": 10.9396,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.156626506024096e-07,
+      "loss": 10.9137,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.638554216867472e-07,
+      "loss": 10.8295,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0120481927710845e-06,
+      "loss": 10.7704,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0602409638554218e-06,
+      "loss": 10.7381,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1084337349397592e-06,
+      "loss": 10.6351,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1566265060240965e-06,
+      "loss": 10.5875,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2048192771084338e-06,
+      "loss": 10.5357,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2530120481927713e-06,
+      "loss": 10.4665,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3012048192771085e-06,
+      "loss": 10.3655,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.349397590361446e-06,
+      "loss": 10.2674,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3975903614457833e-06,
+      "loss": 10.2001,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4457831325301204e-06,
+      "loss": 10.0931,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.493975903614458e-06,
+      "loss": 10.0301,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5421686746987955e-06,
+      "loss": 9.9049,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5903614457831326e-06,
+      "loss": 9.8376,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6385542168674702e-06,
+      "loss": 9.7196,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6867469879518073e-06,
+      "loss": 9.6371,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7349397590361446e-06,
+      "loss": 9.5014,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7831325301204822e-06,
+      "loss": 9.4183,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8313253012048193e-06,
+      "loss": 9.2784,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8795180722891568e-06,
+      "loss": 9.1529,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9277108433734943e-06,
+      "loss": 9.0348,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9759036144578312e-06,
+      "loss": 8.92,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.024096385542169e-06,
+      "loss": 8.7639,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0722891566265063e-06,
+      "loss": 8.6552,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1204819277108437e-06,
+      "loss": 8.5065,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.168674698795181e-06,
+      "loss": 8.3622,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2168674698795183e-06,
+      "loss": 8.2433,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2650602409638556e-06,
+      "loss": 8.1107,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.313253012048193e-06,
+      "loss": 7.9215,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3614457831325303e-06,
+      "loss": 7.7863,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4096385542168676e-06,
+      "loss": 7.5941,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.457831325301205e-06,
+      "loss": 7.4377,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5060240963855427e-06,
+      "loss": 7.2862,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5542168674698796e-06,
+      "loss": 7.0833,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.602409638554217e-06,
+      "loss": 6.9188,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6506024096385547e-06,
+      "loss": 6.7637,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.698795180722892e-06,
+      "loss": 6.5879,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.746987951807229e-06,
+      "loss": 6.3744,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7951807228915666e-06,
+      "loss": 6.1988,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.843373493975904e-06,
+      "loss": 6.013,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.891566265060241e-06,
+      "loss": 5.8462,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9397590361445786e-06,
+      "loss": 5.6436,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.987951807228916e-06,
+      "loss": 5.5071,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0361445783132533e-06,
+      "loss": 5.334,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.084337349397591e-06,
+      "loss": 5.1609,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.132530120481928e-06,
+      "loss": 5.012,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1807228915662653e-06,
+      "loss": 4.8959,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2289156626506026e-06,
+      "loss": 4.785,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2771084337349403e-06,
+      "loss": 4.6499,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3253012048192772e-06,
+      "loss": 4.5113,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3734939759036146e-06,
+      "loss": 4.4022,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4216867469879523e-06,
+      "loss": 4.3412,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4698795180722892e-06,
+      "loss": 4.1839,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5180722891566266e-06,
+      "loss": 4.0896,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5662650602409643e-06,
+      "loss": 3.9756,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6144578313253016e-06,
+      "loss": 3.8625,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6626506024096385e-06,
+      "loss": 3.7413,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7108433734939763e-06,
+      "loss": 3.6755,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7590361445783136e-06,
+      "loss": 3.5984,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.807228915662651e-06,
+      "loss": 3.4659,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.855421686746989e-06,
+      "loss": 3.3896,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.903614457831326e-06,
+      "loss": 3.3546,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9518072289156625e-06,
+      "loss": 3.2607,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 3.2087,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.048192771084338e-06,
+      "loss": 3.1308,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.096385542168675e-06,
+      "loss": 3.1089,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.144578313253013e-06,
+      "loss": 3.0622,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.19277108433735e-06,
+      "loss": 3.0768,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.240963855421687e-06,
+      "loss": 2.8869,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.289156626506025e-06,
+      "loss": 2.8481,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.337349397590362e-06,
+      "loss": 2.7107,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.385542168674699e-06,
+      "loss": 2.6087,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.433734939759037e-06,
+      "loss": 2.6974,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.481927710843374e-06,
+      "loss": 2.5069,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.530120481927711e-06,
+      "loss": 2.4299,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.578313253012049e-06,
+      "loss": 2.3638,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.626506024096386e-06,
+      "loss": 2.2947,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.674698795180723e-06,
+      "loss": 2.2129,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7228915662650606e-06,
+      "loss": 2.1521,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.771084337349398e-06,
+      "loss": 2.1534,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.819277108433735e-06,
+      "loss": 2.0375,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8674698795180725e-06,
+      "loss": 1.9759,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.91566265060241e-06,
+      "loss": 1.8695,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.963855421686747e-06,
+      "loss": 1.7393,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.012048192771085e-06,
+      "loss": 1.7256,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.060240963855422e-06,
+      "loss": 1.5873,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.108433734939759e-06,
+      "loss": 1.6305,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.156626506024097e-06,
+      "loss": 1.5346,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.204819277108434e-06,
+      "loss": 1.5015,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.253012048192771e-06,
+      "loss": 1.5324,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.301204819277109e-06,
+      "loss": 1.4469,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.349397590361446e-06,
+      "loss": 1.2841,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.397590361445784e-06,
+      "loss": 1.4173,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.445783132530121e-06,
+      "loss": 1.4349,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.493975903614458e-06,
+      "loss": 1.3718,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.542168674698796e-06,
+      "loss": 1.1732,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.590361445783133e-06,
+      "loss": 1.1884,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.63855421686747e-06,
+      "loss": 1.1581,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.686746987951808e-06,
+      "loss": 1.168,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.734939759036145e-06,
+      "loss": 1.0928,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.783132530120482e-06,
+      "loss": 1.1126,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.83132530120482e-06,
+      "loss": 1.0596,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.879518072289157e-06,
+      "loss": 1.0166,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.927710843373495e-06,
+      "loss": 1.0827,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.975903614457832e-06,
+      "loss": 0.9825,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.02409638554217e-06,
+      "loss": 0.9547,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.0722891566265066e-06,
+      "loss": 0.959,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.120481927710844e-06,
+      "loss": 0.9747,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.168674698795182e-06,
+      "loss": 0.9697,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.2168674698795185e-06,
+      "loss": 0.9601,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.265060240963856e-06,
+      "loss": 0.9763,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.313253012048192e-06,
+      "loss": 0.9774,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.3614457831325305e-06,
+      "loss": 0.9581,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.409638554216868e-06,
+      "loss": 0.9451,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.457831325301205e-06,
+      "loss": 0.9367,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.5060240963855425e-06,
+      "loss": 0.9496,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.554216867469881e-06,
+      "loss": 0.9413,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.602409638554217e-06,
+      "loss": 0.9419,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.6506024096385545e-06,
+      "loss": 0.917,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.698795180722893e-06,
+      "loss": 0.9281,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.746987951807229e-06,
+      "loss": 0.9088,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.7951807228915665e-06,
+      "loss": 0.9457,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.843373493975905e-06,
+      "loss": 0.9055,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.891566265060241e-06,
+      "loss": 0.9426,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.9397590361445784e-06,
+      "loss": 0.9274,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.987951807228917e-06,
+      "loss": 0.9186,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.036144578313253e-06,
+      "loss": 0.906,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.084337349397591e-06,
+      "loss": 0.9205,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.132530120481929e-06,
+      "loss": 0.9136,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.180722891566265e-06,
+      "loss": 0.9193,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.228915662650603e-06,
+      "loss": 0.9259,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.277108433734941e-06,
+      "loss": 0.9023,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.325301204819277e-06,
+      "loss": 0.9067,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.373493975903615e-06,
+      "loss": 0.9193,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.4216867469879526e-06,
+      "loss": 0.8937,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.469879518072289e-06,
+      "loss": 0.8929,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.518072289156627e-06,
+      "loss": 0.9067,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.5662650602409645e-06,
+      "loss": 0.8839,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.614457831325302e-06,
+      "loss": 0.8939,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.66265060240964e-06,
+      "loss": 0.9078,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.710843373493977e-06,
+      "loss": 0.9234,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.759036144578314e-06,
+      "loss": 0.8968,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.807228915662652e-06,
+      "loss": 0.8979,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.855421686746989e-06,
+      "loss": 0.9055,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.903614457831325e-06,
+      "loss": 0.9065,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.951807228915663e-06,
+      "loss": 0.9014,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.9021,
+      "step": 166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.048192771084338e-06,
+      "loss": 0.8802,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.096385542168676e-06,
+      "loss": 0.8832,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.144578313253012e-06,
+      "loss": 0.8787,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.19277108433735e-06,
+      "loss": 0.8965,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.240963855421687e-06,
+      "loss": 0.8719,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.289156626506025e-06,
+      "loss": 0.8621,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.337349397590362e-06,
+      "loss": 0.888,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.3855421686747e-06,
+      "loss": 0.9058,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.433734939759038e-06,
+      "loss": 0.9018,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.481927710843375e-06,
+      "loss": 0.8968,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.530120481927711e-06,
+      "loss": 0.8896,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.57831325301205e-06,
+      "loss": 0.8726,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.626506024096386e-06,
+      "loss": 0.8787,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.674698795180724e-06,
+      "loss": 0.8772,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.722891566265062e-06,
+      "loss": 0.8933,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.771084337349399e-06,
+      "loss": 0.9083,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.819277108433735e-06,
+      "loss": 0.89,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.867469879518073e-06,
+      "loss": 0.8838,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.91566265060241e-06,
+      "loss": 0.8748,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.963855421686748e-06,
+      "loss": 0.8607,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.012048192771084e-06,
+      "loss": 0.8647,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.060240963855423e-06,
+      "loss": 0.8687,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.10843373493976e-06,
+      "loss": 0.8664,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.156626506024097e-06,
+      "loss": 0.881,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.204819277108434e-06,
+      "loss": 0.8537,
+      "step": 191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.253012048192772e-06,
+      "loss": 0.8412,
+      "step": 192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.301204819277108e-06,
+      "loss": 0.8504,
+      "step": 193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.349397590361446e-06,
+      "loss": 0.8783,
+      "step": 194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.397590361445785e-06,
+      "loss": 0.8575,
+      "step": 195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.445783132530121e-06,
+      "loss": 0.8534,
+      "step": 196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.49397590361446e-06,
+      "loss": 0.872,
+      "step": 197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.542168674698796e-06,
+      "loss": 0.8626,
+      "step": 198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.590361445783132e-06,
+      "loss": 0.8561,
+      "step": 199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.63855421686747e-06,
+      "loss": 0.8604,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.686746987951809e-06,
+      "loss": 0.8601,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.734939759036145e-06,
+      "loss": 0.8662,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.783132530120483e-06,
+      "loss": 0.8445,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.83132530120482e-06,
+      "loss": 0.8427,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879518072289156e-06,
+      "loss": 0.8191,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.927710843373494e-06,
+      "loss": 0.8372,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.975903614457833e-06,
+      "loss": 0.8382,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.002409638554217e-05,
+      "loss": 0.8462,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0072289156626507e-05,
+      "loss": 0.7932,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0120481927710844e-05,
+      "loss": 0.7994,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0168674698795182e-05,
+      "loss": 0.8037,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0216867469879518e-05,
+      "loss": 0.7943,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0265060240963855e-05,
+      "loss": 0.8023,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0313253012048195e-05,
+      "loss": 0.8264,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0361445783132531e-05,
+      "loss": 0.7721,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0409638554216868e-05,
+      "loss": 0.7861,
+      "step": 216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0457831325301206e-05,
+      "loss": 0.7665,
+      "step": 217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0506024096385542e-05,
+      "loss": 0.7777,
+      "step": 218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.055421686746988e-05,
+      "loss": 0.7907,
+      "step": 219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0602409638554219e-05,
+      "loss": 0.7862,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0650602409638555e-05,
+      "loss": 0.7541,
+      "step": 221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0698795180722892e-05,
+      "loss": 0.7775,
+      "step": 222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.074698795180723e-05,
+      "loss": 0.7448,
+      "step": 223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0795180722891568e-05,
+      "loss": 0.7355,
+      "step": 224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0843373493975904e-05,
+      "loss": 0.7273,
+      "step": 225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0891566265060243e-05,
+      "loss": 0.7286,
+      "step": 226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0939759036144579e-05,
+      "loss": 0.7284,
+      "step": 227
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0987951807228916e-05,
+      "loss": 0.7062,
+      "step": 228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1036144578313255e-05,
+      "loss": 0.6925,
+      "step": 229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1084337349397592e-05,
+      "loss": 0.6799,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1132530120481928e-05,
+      "loss": 0.6673,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1180722891566267e-05,
+      "loss": 0.697,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1228915662650603e-05,
+      "loss": 0.6537,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.127710843373494e-05,
+      "loss": 0.6479,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.132530120481928e-05,
+      "loss": 0.6668,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1373493975903616e-05,
+      "loss": 0.6407,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1421686746987952e-05,
+      "loss": 0.6422,
+      "step": 237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.146987951807229e-05,
+      "loss": 0.6186,
+      "step": 238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1518072289156627e-05,
+      "loss": 0.6152,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1566265060240964e-05,
+      "loss": 0.5843,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1614457831325303e-05,
+      "loss": 0.5821,
+      "step": 241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.166265060240964e-05,
+      "loss": 0.5889,
+      "step": 242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1710843373493976e-05,
+      "loss": 0.5806,
+      "step": 243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1759036144578315e-05,
+      "loss": 0.5824,
+      "step": 244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1807228915662651e-05,
+      "loss": 0.5889,
+      "step": 245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.185542168674699e-05,
+      "loss": 0.57,
+      "step": 246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1903614457831327e-05,
+      "loss": 0.5521,
+      "step": 247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1951807228915664e-05,
+      "loss": 0.5468,
+      "step": 248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2e-05,
+      "loss": 0.546,
+      "step": 249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.204819277108434e-05,
+      "loss": 0.5667,
+      "step": 250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2096385542168677e-05,
+      "loss": 0.5666,
+      "step": 251
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2144578313253013e-05,
+      "loss": 0.5181,
+      "step": 252
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2192771084337351e-05,
+      "loss": 0.5023,
+      "step": 253
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2240963855421688e-05,
+      "loss": 0.5308,
+      "step": 254
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2289156626506024e-05,
+      "loss": 0.4984,
+      "step": 255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2337349397590364e-05,
+      "loss": 0.5142,
+      "step": 256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.23855421686747e-05,
+      "loss": 0.4944,
+      "step": 257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2433734939759037e-05,
+      "loss": 0.4887,
+      "step": 258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2481927710843375e-05,
+      "loss": 0.4804,
+      "step": 259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2530120481927712e-05,
+      "loss": 0.4727,
+      "step": 260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2578313253012048e-05,
+      "loss": 0.4649,
+      "step": 261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2626506024096385e-05,
+      "loss": 0.4921,
+      "step": 262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2674698795180725e-05,
+      "loss": 0.4645,
+      "step": 263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2722891566265061e-05,
+      "loss": 0.4781,
+      "step": 264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2771084337349398e-05,
+      "loss": 0.4615,
+      "step": 265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2819277108433736e-05,
+      "loss": 0.501,
+      "step": 266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2867469879518072e-05,
+      "loss": 0.4824,
+      "step": 267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.291566265060241e-05,
+      "loss": 0.4844,
+      "step": 268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2963855421686749e-05,
+      "loss": 0.4806,
+      "step": 269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3012048192771085e-05,
+      "loss": 0.4713,
+      "step": 270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3060240963855421e-05,
+      "loss": 0.4639,
+      "step": 271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3108433734939761e-05,
+      "loss": 0.4653,
+      "step": 272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3156626506024098e-05,
+      "loss": 0.4724,
+      "step": 273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3204819277108434e-05,
+      "loss": 0.4635,
+      "step": 274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3253012048192772e-05,
+      "loss": 0.4731,
+      "step": 275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3301204819277109e-05,
+      "loss": 0.4879,
+      "step": 276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3349397590361445e-05,
+      "loss": 0.4385,
+      "step": 277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3397590361445785e-05,
+      "loss": 0.4449,
+      "step": 278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3445783132530122e-05,
+      "loss": 0.4967,
+      "step": 279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3493975903614458e-05,
+      "loss": 0.5081,
+      "step": 280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3542168674698796e-05,
+      "loss": 0.5619,
+      "step": 281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3590361445783133e-05,
+      "loss": 0.552,
+      "step": 282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.363855421686747e-05,
+      "loss": 0.4949,
+      "step": 283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.368674698795181e-05,
+      "loss": 0.4727,
+      "step": 284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3734939759036146e-05,
+      "loss": 0.4719,
+      "step": 285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3783132530120482e-05,
+      "loss": 0.4755,
+      "step": 286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.383132530120482e-05,
+      "loss": 0.4804,
+      "step": 287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3879518072289157e-05,
+      "loss": 0.4838,
+      "step": 288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3927710843373493e-05,
+      "loss": 0.497,
+      "step": 289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3975903614457833e-05,
+      "loss": 0.4722,
+      "step": 290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.402409638554217e-05,
+      "loss": 0.4762,
+      "step": 291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4072289156626506e-05,
+      "loss": 0.4535,
+      "step": 292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4120481927710844e-05,
+      "loss": 0.4507,
+      "step": 293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4168674698795183e-05,
+      "loss": 0.4684,
+      "step": 294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4216867469879519e-05,
+      "loss": 0.4796,
+      "step": 295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4265060240963857e-05,
+      "loss": 0.4506,
+      "step": 296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4313253012048194e-05,
+      "loss": 0.4584,
+      "step": 297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.436144578313253e-05,
+      "loss": 0.4574,
+      "step": 298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.440963855421687e-05,
+      "loss": 0.464,
+      "step": 299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4457831325301207e-05,
+      "loss": 0.4594,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4506024096385543e-05,
+      "loss": 0.4798,
+      "step": 301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4554216867469881e-05,
+      "loss": 0.4504,
+      "step": 302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4602409638554218e-05,
+      "loss": 0.484,
+      "step": 303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4650602409638554e-05,
+      "loss": 0.4416,
+      "step": 304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4698795180722894e-05,
+      "loss": 0.481,
+      "step": 305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.474698795180723e-05,
+      "loss": 0.4726,
+      "step": 306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4795180722891567e-05,
+      "loss": 0.4754,
+      "step": 307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4843373493975905e-05,
+      "loss": 0.4734,
+      "step": 308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4891566265060242e-05,
+      "loss": 0.424,
+      "step": 309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4939759036144578e-05,
+      "loss": 0.4654,
+      "step": 310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4987951807228918e-05,
+      "loss": 0.5109,
+      "step": 311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5036144578313254e-05,
+      "loss": 0.4789,
+      "step": 312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5084337349397591e-05,
+      "loss": 0.449,
+      "step": 313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5132530120481929e-05,
+      "loss": 0.4834,
+      "step": 314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5180722891566266e-05,
+      "loss": 0.4524,
+      "step": 315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5228915662650604e-05,
+      "loss": 0.458,
+      "step": 316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5277108433734942e-05,
+      "loss": 0.48,
+      "step": 317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.532530120481928e-05,
+      "loss": 0.4723,
+      "step": 318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5373493975903615e-05,
+      "loss": 0.4742,
+      "step": 319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5421686746987955e-05,
+      "loss": 0.4704,
+      "step": 320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.546987951807229e-05,
+      "loss": 0.4515,
+      "step": 321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5518072289156628e-05,
+      "loss": 0.4576,
+      "step": 322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5566265060240968e-05,
+      "loss": 0.4504,
+      "step": 323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5614457831325304e-05,
+      "loss": 0.4499,
+      "step": 324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.566265060240964e-05,
+      "loss": 0.434,
+      "step": 325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5710843373493977e-05,
+      "loss": 0.4606,
+      "step": 326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5759036144578313e-05,
+      "loss": 0.4754,
+      "step": 327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.580722891566265e-05,
+      "loss": 0.4514,
+      "step": 328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.585542168674699e-05,
+      "loss": 0.474,
+      "step": 329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5903614457831326e-05,
+      "loss": 0.4674,
+      "step": 330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5951807228915663e-05,
+      "loss": 0.4526,
+      "step": 331
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.4493,
+      "step": 332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.604819277108434e-05,
+      "loss": 0.4495,
+      "step": 333
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6096385542168676e-05,
+      "loss": 0.4923,
+      "step": 334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6144578313253015e-05,
+      "loss": 0.4567,
+      "step": 335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6192771084337352e-05,
+      "loss": 0.5321,
+      "step": 336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.624096385542169e-05,
+      "loss": 0.5038,
+      "step": 337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6289156626506025e-05,
+      "loss": 0.4775,
+      "step": 338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.633734939759036e-05,
+      "loss": 0.476,
+      "step": 339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.63855421686747e-05,
+      "loss": 0.4817,
+      "step": 340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6433734939759038e-05,
+      "loss": 0.4736,
+      "step": 341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6481927710843374e-05,
+      "loss": 0.4398,
+      "step": 342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.653012048192771e-05,
+      "loss": 0.4478,
+      "step": 343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.657831325301205e-05,
+      "loss": 0.4775,
+      "step": 344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6626506024096387e-05,
+      "loss": 0.4749,
+      "step": 345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6674698795180724e-05,
+      "loss": 0.4351,
+      "step": 346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6722891566265063e-05,
+      "loss": 0.4503,
+      "step": 347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.67710843373494e-05,
+      "loss": 0.4393,
+      "step": 348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6819277108433736e-05,
+      "loss": 0.4487,
+      "step": 349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6867469879518076e-05,
+      "loss": 0.4904,
+      "step": 350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6915662650602413e-05,
+      "loss": 0.4618,
+      "step": 351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.696385542168675e-05,
+      "loss": 0.4705,
+      "step": 352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7012048192771086e-05,
+      "loss": 0.4571,
+      "step": 353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7060240963855422e-05,
+      "loss": 0.4789,
+      "step": 354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.710843373493976e-05,
+      "loss": 0.4852,
+      "step": 355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.71566265060241e-05,
+      "loss": 0.4524,
+      "step": 356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7204819277108435e-05,
+      "loss": 0.4382,
+      "step": 357
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.725301204819277e-05,
+      "loss": 0.4557,
+      "step": 358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.730120481927711e-05,
+      "loss": 0.4509,
+      "step": 359
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7349397590361448e-05,
+      "loss": 0.4327,
+      "step": 360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7397590361445784e-05,
+      "loss": 0.4849,
+      "step": 361
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7445783132530124e-05,
+      "loss": 0.4985,
+      "step": 362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.749397590361446e-05,
+      "loss": 0.4614,
+      "step": 363
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7542168674698797e-05,
+      "loss": 0.4573,
+      "step": 364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7590361445783134e-05,
+      "loss": 0.4565,
+      "step": 365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.763855421686747e-05,
+      "loss": 0.4704,
+      "step": 366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.768674698795181e-05,
+      "loss": 0.4762,
+      "step": 367
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7734939759036146e-05,
+      "loss": 0.4683,
+      "step": 368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7783132530120483e-05,
+      "loss": 0.4652,
+      "step": 369
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.783132530120482e-05,
+      "loss": 0.4639,
+      "step": 370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7879518072289156e-05,
+      "loss": 0.4618,
+      "step": 371
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7927710843373496e-05,
+      "loss": 0.4722,
+      "step": 372
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7975903614457832e-05,
+      "loss": 0.4709,
+      "step": 373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.802409638554217e-05,
+      "loss": 0.4637,
+      "step": 374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.807228915662651e-05,
+      "loss": 0.4505,
+      "step": 375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8120481927710845e-05,
+      "loss": 0.4612,
+      "step": 376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.816867469879518e-05,
+      "loss": 0.499,
+      "step": 377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.821686746987952e-05,
+      "loss": 0.4833,
+      "step": 378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8265060240963858e-05,
+      "loss": 0.4563,
+      "step": 379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8313253012048194e-05,
+      "loss": 0.4622,
+      "step": 380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.836144578313253e-05,
+      "loss": 0.45,
+      "step": 381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8409638554216867e-05,
+      "loss": 0.4628,
+      "step": 382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8457831325301204e-05,
+      "loss": 0.4625,
+      "step": 383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8506024096385544e-05,
+      "loss": 0.4775,
+      "step": 384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.855421686746988e-05,
+      "loss": 0.4668,
+      "step": 385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8602409638554217e-05,
+      "loss": 0.4541,
+      "step": 386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8650602409638556e-05,
+      "loss": 0.4573,
+      "step": 387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8698795180722893e-05,
+      "loss": 0.4815,
+      "step": 388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.874698795180723e-05,
+      "loss": 0.4597,
+      "step": 389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.879518072289157e-05,
+      "loss": 0.4895,
+      "step": 390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8843373493975906e-05,
+      "loss": 0.4581,
+      "step": 391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8891566265060242e-05,
+      "loss": 0.4419,
+      "step": 392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.893975903614458e-05,
+      "loss": 0.4607,
+      "step": 393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.898795180722892e-05,
+      "loss": 0.548,
+      "step": 394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9036144578313255e-05,
+      "loss": 0.5068,
+      "step": 395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908433734939759e-05,
+      "loss": 0.4574,
+      "step": 396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9132530120481928e-05,
+      "loss": 0.4434,
+      "step": 397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9180722891566265e-05,
+      "loss": 0.4587,
+      "step": 398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9228915662650604e-05,
+      "loss": 0.4583,
+      "step": 399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.927710843373494e-05,
+      "loss": 0.4402,
+      "step": 400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9325301204819277e-05,
+      "loss": 0.4731,
+      "step": 401
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9373493975903617e-05,
+      "loss": 0.493,
+      "step": 402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9421686746987954e-05,
+      "loss": 0.4602,
+      "step": 403
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.946987951807229e-05,
+      "loss": 0.4469,
+      "step": 404
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.951807228915663e-05,
+      "loss": 0.4596,
+      "step": 405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9566265060240967e-05,
+      "loss": 0.461,
+      "step": 406
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9614457831325303e-05,
+      "loss": 0.4672,
+      "step": 407
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.966265060240964e-05,
+      "loss": 0.4675,
+      "step": 408
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9710843373493976e-05,
+      "loss": 0.4534,
+      "step": 409
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9759036144578312e-05,
+      "loss": 0.4655,
+      "step": 410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9807228915662652e-05,
+      "loss": 0.4879,
+      "step": 411
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.985542168674699e-05,
+      "loss": 0.4461,
+      "step": 412
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9903614457831325e-05,
+      "loss": 0.4621,
+      "step": 413
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9951807228915665e-05,
+      "loss": 0.4724,
+      "step": 414
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.4718,
+      "step": 415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999999725582255e-05,
+      "loss": 0.4662,
+      "step": 416
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999998902329033e-05,
+      "loss": 0.4491,
+      "step": 417
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999997530240382e-05,
+      "loss": 0.4386,
+      "step": 418
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999995609316377e-05,
+      "loss": 0.4513,
+      "step": 419
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999993139557117e-05,
+      "loss": 0.4998,
+      "step": 420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999990120962743e-05,
+      "loss": 0.4867,
+      "step": 421
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999986553533422e-05,
+      "loss": 0.4503,
+      "step": 422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999998243726935e-05,
+      "loss": 0.4663,
+      "step": 423
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999977772170748e-05,
+      "loss": 0.4629,
+      "step": 424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999972558237875e-05,
+      "loss": 0.4566,
+      "step": 425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999996679547102e-05,
+      "loss": 0.4536,
+      "step": 426
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999960483870494e-05,
+      "loss": 0.4504,
+      "step": 427
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999995362343665e-05,
+      "loss": 0.4954,
+      "step": 428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999946214169852e-05,
+      "loss": 0.4437,
+      "step": 429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999993825607052e-05,
+      "loss": 0.4975,
+      "step": 430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999929749139087e-05,
+      "loss": 0.4663,
+      "step": 431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999920693376013e-05,
+      "loss": 0.4716,
+      "step": 432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999911088781804e-05,
+      "loss": 0.4685,
+      "step": 433
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999900935356987e-05,
+      "loss": 0.486,
+      "step": 434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999890233102113e-05,
+      "loss": 0.4666,
+      "step": 435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999878982017772e-05,
+      "loss": 0.4568,
+      "step": 436
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999867182104583e-05,
+      "loss": 0.4562,
+      "step": 437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999854833363193e-05,
+      "loss": 0.4777,
+      "step": 438
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999984193579428e-05,
+      "loss": 0.4438,
+      "step": 439
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999982848939855e-05,
+      "loss": 0.4682,
+      "step": 440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999814494176737e-05,
+      "loss": 0.4512,
+      "step": 441
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999979995012962e-05,
+      "loss": 0.4754,
+      "step": 442
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999784857257988e-05,
+      "loss": 0.4559,
+      "step": 443
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999769215562678e-05,
+      "loss": 0.5196,
+      "step": 444
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999753025044538e-05,
+      "loss": 0.4786,
+      "step": 445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999736285704463e-05,
+      "loss": 0.4782,
+      "step": 446
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999718997543374e-05,
+      "loss": 0.4958,
+      "step": 447
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999701160562214e-05,
+      "loss": 0.4749,
+      "step": 448
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999682774761963e-05,
+      "loss": 0.4692,
+      "step": 449
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999966384014363e-05,
+      "loss": 0.4748,
+      "step": 450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999644356708262e-05,
+      "loss": 0.4389,
+      "step": 451
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999962432445692e-05,
+      "loss": 0.4869,
+      "step": 452
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999603743390703e-05,
+      "loss": 0.5039,
+      "step": 453
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999582613510742e-05,
+      "loss": 0.4751,
+      "step": 454
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99995609348182e-05,
+      "loss": 0.4759,
+      "step": 455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999538707314264e-05,
+      "loss": 0.4652,
+      "step": 456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999951593100015e-05,
+      "loss": 0.4792,
+      "step": 457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999949260587712e-05,
+      "loss": 0.4731,
+      "step": 458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999946873194644e-05,
+      "loss": 0.4571,
+      "step": 459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999444309209432e-05,
+      "loss": 0.4535,
+      "step": 460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999941933766743e-05,
+      "loss": 0.4373,
+      "step": 461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999393817321808e-05,
+      "loss": 0.5346,
+      "step": 462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999936774817396e-05,
+      "loss": 0.4507,
+      "step": 463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999341130225324e-05,
+      "loss": 0.4626,
+      "step": 464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999313963477357e-05,
+      "loss": 0.4334,
+      "step": 465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999928624793155e-05,
+      "loss": 0.4869,
+      "step": 466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999925798358943e-05,
+      "loss": 0.4723,
+      "step": 467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999229170452535e-05,
+      "loss": 0.4454,
+      "step": 468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999919980852246e-05,
+      "loss": 0.4725,
+      "step": 469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999169897800814e-05,
+      "loss": 0.4732,
+      "step": 470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999139438289233e-05,
+      "loss": 0.47,
+      "step": 471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999108429989393e-05,
+      "loss": 0.4594,
+      "step": 472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999907687290299e-05,
+      "loss": 0.4605,
+      "step": 473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999044767031766e-05,
+      "loss": 0.4866,
+      "step": 474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999012112377473e-05,
+      "loss": 0.466,
+      "step": 475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999897890894191e-05,
+      "loss": 0.4162,
+      "step": 476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998945156726898e-05,
+      "loss": 0.4607,
+      "step": 477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999891085573429e-05,
+      "loss": 0.4582,
+      "step": 478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998876005965965e-05,
+      "loss": 0.4557,
+      "step": 479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998840607423835e-05,
+      "loss": 0.493,
+      "step": 480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999880466010985e-05,
+      "loss": 0.4748,
+      "step": 481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998768164025973e-05,
+      "loss": 0.4782,
+      "step": 482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998731119174216e-05,
+      "loss": 0.442,
+      "step": 483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999869352555661e-05,
+      "loss": 0.4482,
+      "step": 484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998655383175217e-05,
+      "loss": 0.4545,
+      "step": 485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998616692032127e-05,
+      "loss": 0.4525,
+      "step": 486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999857745212947e-05,
+      "loss": 0.4398,
+      "step": 487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998537663469398e-05,
+      "loss": 0.4721,
+      "step": 488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998497326054088e-05,
+      "loss": 0.4643,
+      "step": 489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998456439885763e-05,
+      "loss": 0.4375,
+      "step": 490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999841500496666e-05,
+      "loss": 0.4276,
+      "step": 491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999837302129906e-05,
+      "loss": 0.4521,
+      "step": 492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999833048888526e-05,
+      "loss": 0.4632,
+      "step": 493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998287407727602e-05,
+      "loss": 0.4664,
+      "step": 494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999824377782844e-05,
+      "loss": 0.4613,
+      "step": 495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998199599190177e-05,
+      "loss": 0.4263,
+      "step": 496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999815487181524e-05,
+      "loss": 0.4386,
+      "step": 497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998109595706073e-05,
+      "loss": 0.4776,
+      "step": 498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999806377086517e-05,
+      "loss": 0.4449,
+      "step": 499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999801739729504e-05,
+      "loss": 0.461,
+      "step": 500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997970474998237e-05,
+      "loss": 0.4764,
+      "step": 501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997923003977328e-05,
+      "loss": 0.4857,
+      "step": 502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999787498423492e-05,
+      "loss": 0.5005,
+      "step": 503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999782641577365e-05,
+      "loss": 0.4952,
+      "step": 504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999777729859618e-05,
+      "loss": 0.4743,
+      "step": 505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997727632705213e-05,
+      "loss": 0.4811,
+      "step": 506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999767741810347e-05,
+      "loss": 0.4574,
+      "step": 507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997626654793708e-05,
+      "loss": 0.4619,
+      "step": 508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999757534277871e-05,
+      "loss": 0.4818,
+      "step": 509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997523482061296e-05,
+      "loss": 0.4555,
+      "step": 510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997471072644314e-05,
+      "loss": 0.4444,
+      "step": 511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997418114530634e-05,
+      "loss": 0.4631,
+      "step": 512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997364607723166e-05,
+      "loss": 0.4544,
+      "step": 513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997310552224848e-05,
+      "loss": 0.4659,
+      "step": 514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997255948038644e-05,
+      "loss": 0.4633,
+      "step": 515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997200795167553e-05,
+      "loss": 0.4826,
+      "step": 516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99971450936146e-05,
+      "loss": 0.5052,
+      "step": 517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997088843382844e-05,
+      "loss": 0.4722,
+      "step": 518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997032044475375e-05,
+      "loss": 0.4763,
+      "step": 519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996974696895304e-05,
+      "loss": 0.4561,
+      "step": 520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999691680064578e-05,
+      "loss": 0.478,
+      "step": 521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996858355729985e-05,
+      "loss": 0.477,
+      "step": 522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996799362151124e-05,
+      "loss": 0.4331,
+      "step": 523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999673981991243e-05,
+      "loss": 0.4757,
+      "step": 524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999667972901718e-05,
+      "loss": 0.4606,
+      "step": 525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996619089468664e-05,
+      "loss": 0.5061,
+      "step": 526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996557901270215e-05,
+      "loss": 0.4661,
+      "step": 527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999649616442519e-05,
+      "loss": 0.4686,
+      "step": 528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996433878936973e-05,
+      "loss": 0.4651,
+      "step": 529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999637104480899e-05,
+      "loss": 0.4529,
+      "step": 530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996307662044685e-05,
+      "loss": 0.4764,
+      "step": 531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996243730647538e-05,
+      "loss": 0.4546,
+      "step": 532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996179250621056e-05,
+      "loss": 0.4495,
+      "step": 533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996114221968783e-05,
+      "loss": 0.4673,
+      "step": 534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999604864469428e-05,
+      "loss": 0.4761,
+      "step": 535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995982518801155e-05,
+      "loss": 0.4818,
+      "step": 536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995915844293027e-05,
+      "loss": 0.4516,
+      "step": 537
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995848621173564e-05,
+      "loss": 0.4619,
+      "step": 538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995780849446454e-05,
+      "loss": 0.4489,
+      "step": 539
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995712529115412e-05,
+      "loss": 0.4827,
+      "step": 540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999564366018419e-05,
+      "loss": 0.4898,
+      "step": 541
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995574242656573e-05,
+      "loss": 0.4721,
+      "step": 542
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995504276536365e-05,
+      "loss": 0.4689,
+      "step": 543
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995433761827405e-05,
+      "loss": 0.4784,
+      "step": 544
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995362698533565e-05,
+      "loss": 0.4573,
+      "step": 545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995291086658745e-05,
+      "loss": 0.4591,
+      "step": 546
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999521892620688e-05,
+      "loss": 0.4645,
+      "step": 547
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999514621718192e-05,
+      "loss": 0.4635,
+      "step": 548
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999507295958787e-05,
+      "loss": 0.4679,
+      "step": 549
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994999153428737e-05,
+      "loss": 0.4931,
+      "step": 550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999492479870858e-05,
+      "loss": 0.4546,
+      "step": 551
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994849895431472e-05,
+      "loss": 0.4675,
+      "step": 552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994774443601535e-05,
+      "loss": 0.4672,
+      "step": 553
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99946984432229e-05,
+      "loss": 0.4684,
+      "step": 554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994621894299745e-05,
+      "loss": 0.4726,
+      "step": 555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994544796836267e-05,
+      "loss": 0.4614,
+      "step": 556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99944671508367e-05,
+      "loss": 0.4532,
+      "step": 557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994388956305303e-05,
+      "loss": 0.4528,
+      "step": 558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999431021324637e-05,
+      "loss": 0.4598,
+      "step": 559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999423092166422e-05,
+      "loss": 0.4668,
+      "step": 560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994151081563206e-05,
+      "loss": 0.449,
+      "step": 561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994070692947714e-05,
+      "loss": 0.453,
+      "step": 562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999398975582215e-05,
+      "loss": 0.4506,
+      "step": 563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999390827019096e-05,
+      "loss": 0.4629,
+      "step": 564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993826236058612e-05,
+      "loss": 0.4457,
+      "step": 565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993743653429614e-05,
+      "loss": 0.4314,
+      "step": 566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993660522308493e-05,
+      "loss": 0.4657,
+      "step": 567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993576842699817e-05,
+      "loss": 0.449,
+      "step": 568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993492614608173e-05,
+      "loss": 0.4254,
+      "step": 569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999340783803819e-05,
+      "loss": 0.4137,
+      "step": 570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993322512994514e-05,
+      "loss": 0.465,
+      "step": 571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993236639481834e-05,
+      "loss": 0.4301,
+      "step": 572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993150217504854e-05,
+      "loss": 0.4684,
+      "step": 573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999306324706833e-05,
+      "loss": 0.4713,
+      "step": 574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992975728177025e-05,
+      "loss": 0.4566,
+      "step": 575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992887660835746e-05,
+      "loss": 0.4465,
+      "step": 576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999279904504933e-05,
+      "loss": 0.4649,
+      "step": 577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992709880822634e-05,
+      "loss": 0.462,
+      "step": 578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992620168160554e-05,
+      "loss": 0.4628,
+      "step": 579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992529907068015e-05,
+      "loss": 0.4436,
+      "step": 580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999243909754997e-05,
+      "loss": 0.4478,
+      "step": 581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992347739611402e-05,
+      "loss": 0.4661,
+      "step": 582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999225583325733e-05,
+      "loss": 0.449,
+      "step": 583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992163378492793e-05,
+      "loss": 0.439,
+      "step": 584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992070375322866e-05,
+      "loss": 0.4632,
+      "step": 585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991976823752655e-05,
+      "loss": 0.452,
+      "step": 586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991882723787293e-05,
+      "loss": 0.4734,
+      "step": 587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991788075431944e-05,
+      "loss": 0.4609,
+      "step": 588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99916928786918e-05,
+      "loss": 0.4801,
+      "step": 589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991597133572097e-05,
+      "loss": 0.4648,
+      "step": 590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991500840078078e-05,
+      "loss": 0.4491,
+      "step": 591
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999140399821503e-05,
+      "loss": 0.444,
+      "step": 592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991306607988275e-05,
+      "loss": 0.4864,
+      "step": 593
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999120866940315e-05,
+      "loss": 0.4921,
+      "step": 594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991110182465032e-05,
+      "loss": 0.4395,
+      "step": 595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991011147179333e-05,
+      "loss": 0.4557,
+      "step": 596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990911563551476e-05,
+      "loss": 0.4744,
+      "step": 597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990811431586938e-05,
+      "loss": 0.4564,
+      "step": 598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990710751291213e-05,
+      "loss": 0.4453,
+      "step": 599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990609522669818e-05,
+      "loss": 0.4549,
+      "step": 600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999050774572832e-05,
+      "loss": 0.4501,
+      "step": 601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990405420472293e-05,
+      "loss": 0.441,
+      "step": 602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990302546907364e-05,
+      "loss": 0.4769,
+      "step": 603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990199125039174e-05,
+      "loss": 0.4474,
+      "step": 604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99900951548734e-05,
+      "loss": 0.4475,
+      "step": 605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989990636415748e-05,
+      "loss": 0.4346,
+      "step": 606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989885569671957e-05,
+      "loss": 0.4907,
+      "step": 607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989779954647787e-05,
+      "loss": 0.4639,
+      "step": 608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998967379134904e-05,
+      "loss": 0.446,
+      "step": 609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998956707978154e-05,
+      "loss": 0.4561,
+      "step": 610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989459819951146e-05,
+      "loss": 0.456,
+      "step": 611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989352011863744e-05,
+      "loss": 0.4563,
+      "step": 612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989243655525248e-05,
+      "loss": 0.4628,
+      "step": 613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989134750941607e-05,
+      "loss": 0.4576,
+      "step": 614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989025298118802e-05,
+      "loss": 0.4578,
+      "step": 615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988915297062836e-05,
+      "loss": 0.475,
+      "step": 616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988804747779742e-05,
+      "loss": 0.4216,
+      "step": 617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988693650275598e-05,
+      "loss": 0.4666,
+      "step": 618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988582004556493e-05,
+      "loss": 0.4909,
+      "step": 619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988469810628554e-05,
+      "loss": 0.4455,
+      "step": 620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988357068497944e-05,
+      "loss": 0.4629,
+      "step": 621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988243778170853e-05,
+      "loss": 0.4802,
+      "step": 622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988129939653488e-05,
+      "loss": 0.464,
+      "step": 623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988015552952107e-05,
+      "loss": 0.4659,
+      "step": 624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998790061807298e-05,
+      "loss": 0.4497,
+      "step": 625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987785135022423e-05,
+      "loss": 0.4498,
+      "step": 626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998766910380677e-05,
+      "loss": 0.4639,
+      "step": 627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998755252443239e-05,
+      "loss": 0.4568,
+      "step": 628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987435396905677e-05,
+      "loss": 0.4729,
+      "step": 629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987317721233065e-05,
+      "loss": 0.4363,
+      "step": 630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998719949742101e-05,
+      "loss": 0.4469,
+      "step": 631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987080725476004e-05,
+      "loss": 0.4448,
+      "step": 632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986961405404558e-05,
+      "loss": 0.4729,
+      "step": 633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986841537213227e-05,
+      "loss": 0.4389,
+      "step": 634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986721120908587e-05,
+      "loss": 0.4331,
+      "step": 635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986600156497248e-05,
+      "loss": 0.4387,
+      "step": 636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986478643985854e-05,
+      "loss": 0.4538,
+      "step": 637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986356583381063e-05,
+      "loss": 0.4767,
+      "step": 638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986233974689583e-05,
+      "loss": 0.4556,
+      "step": 639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998611081791814e-05,
+      "loss": 0.4658,
+      "step": 640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985987113073494e-05,
+      "loss": 0.4686,
+      "step": 641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985862860162433e-05,
+      "loss": 0.4751,
+      "step": 642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998573805919178e-05,
+      "loss": 0.4448,
+      "step": 643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985612710168378e-05,
+      "loss": 0.4641,
+      "step": 644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985486813099112e-05,
+      "loss": 0.4442,
+      "step": 645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998536036799089e-05,
+      "loss": 0.4851,
+      "step": 646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985233374850653e-05,
+      "loss": 0.491,
+      "step": 647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985105833685368e-05,
+      "loss": 0.4684,
+      "step": 648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9984977744502038e-05,
+      "loss": 0.4682,
+      "step": 649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998484910730769e-05,
+      "loss": 0.4798,
+      "step": 650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984719922109388e-05,
+      "loss": 0.4446,
+      "step": 651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998459018891422e-05,
+      "loss": 0.4523,
+      "step": 652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984459907729305e-05,
+      "loss": 0.4547,
+      "step": 653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984329078561795e-05,
+      "loss": 0.4515,
+      "step": 654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998419770141887e-05,
+      "loss": 0.453,
+      "step": 655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998406577630774e-05,
+      "loss": 0.4656,
+      "step": 656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983933303235644e-05,
+      "loss": 0.4647,
+      "step": 657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983800282209857e-05,
+      "loss": 0.4572,
+      "step": 658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983666713237676e-05,
+      "loss": 0.4594,
+      "step": 659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983532596326435e-05,
+      "loss": 0.4504,
+      "step": 660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983397931483488e-05,
+      "loss": 0.4583,
+      "step": 661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983262718716233e-05,
+      "loss": 0.4632,
+      "step": 662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983126958032088e-05,
+      "loss": 0.4585,
+      "step": 663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.99829906494385e-05,
+      "loss": 0.4615,
+      "step": 664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998285379294296e-05,
+      "loss": 0.4914,
+      "step": 665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982716388552978e-05,
+      "loss": 0.4781,
+      "step": 666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982578436276084e-05,
+      "loss": 0.4452,
+      "step": 667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982439936119858e-05,
+      "loss": 0.449,
+      "step": 668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.99823008880919e-05,
+      "loss": 0.4721,
+      "step": 669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998216129219984e-05,
+      "loss": 0.4679,
+      "step": 670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998202114845134e-05,
+      "loss": 0.4102,
+      "step": 671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981880456854092e-05,
+      "loss": 0.4724,
+      "step": 672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998173921741582e-05,
+      "loss": 0.4789,
+      "step": 673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981597430144273e-05,
+      "loss": 0.4847,
+      "step": 674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981455095047233e-05,
+      "loss": 0.4744,
+      "step": 675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981312212132514e-05,
+      "loss": 0.4797,
+      "step": 676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981168781407955e-05,
+      "loss": 0.4478,
+      "step": 677
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9981024802881426e-05,
+      "loss": 0.4403,
+      "step": 678
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980880276560834e-05,
+      "loss": 0.48,
+      "step": 679
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998073520245411e-05,
+      "loss": 0.5155,
+      "step": 680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980589580569213e-05,
+      "loss": 0.5729,
+      "step": 681
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998044341091414e-05,
+      "loss": 0.4666,
+      "step": 682
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998029669349691e-05,
+      "loss": 0.48,
+      "step": 683
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998014942832558e-05,
+      "loss": 0.4559,
+      "step": 684
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980001615408228e-05,
+      "loss": 0.4329,
+      "step": 685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979853254752965e-05,
+      "loss": 0.4642,
+      "step": 686
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979704346367936e-05,
+      "loss": 0.5069,
+      "step": 687
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979554890261316e-05,
+      "loss": 0.5143,
+      "step": 688
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99794048864413e-05,
+      "loss": 0.5361,
+      "step": 689
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997925433491613e-05,
+      "loss": 0.4737,
+      "step": 690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979103235694065e-05,
+      "loss": 0.4609,
+      "step": 691
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99789515887834e-05,
+      "loss": 0.4637,
+      "step": 692
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978799394192447e-05,
+      "loss": 0.4286,
+      "step": 693
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997864665192957e-05,
+      "loss": 0.4779,
+      "step": 694
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978493362003155e-05,
+      "loss": 0.4665,
+      "step": 695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997833952442161e-05,
+      "loss": 0.4428,
+      "step": 696
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997818513919337e-05,
+      "loss": 0.4582,
+      "step": 697
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978030206326922e-05,
+      "loss": 0.4469,
+      "step": 698
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977874725830762e-05,
+      "loss": 0.4502,
+      "step": 699
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977718697713425e-05,
+      "loss": 0.4557,
+      "step": 700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997756212198347e-05,
+      "loss": 0.4695,
+      "step": 701
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977404998649497e-05,
+      "loss": 0.4475,
+      "step": 702
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977247327720128e-05,
+      "loss": 0.4425,
+      "step": 703
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977089109204016e-05,
+      "loss": 0.4646,
+      "step": 704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9976930343109842e-05,
+      "loss": 0.4591,
+      "step": 705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976771029446323e-05,
+      "loss": 0.4562,
+      "step": 706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976611168222198e-05,
+      "loss": 0.4424,
+      "step": 707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976450759446248e-05,
+      "loss": 0.4371,
+      "step": 708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976289803127273e-05,
+      "loss": 0.4305,
+      "step": 709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976128299274104e-05,
+      "loss": 0.4912,
+      "step": 710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975966247895607e-05,
+      "loss": 0.4611,
+      "step": 711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975803649000684e-05,
+      "loss": 0.4533,
+      "step": 712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975640502598243e-05,
+      "loss": 0.4966,
+      "step": 713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975476808697252e-05,
+      "loss": 0.4635,
+      "step": 714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975312567306687e-05,
+      "loss": 0.4759,
+      "step": 715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975147778435567e-05,
+      "loss": 0.4798,
+      "step": 716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997498244209293e-05,
+      "loss": 0.4628,
+      "step": 717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974816558287856e-05,
+      "loss": 0.4741,
+      "step": 718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997465012702945e-05,
+      "loss": 0.4358,
+      "step": 719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974483148326843e-05,
+      "loss": 0.4603,
+      "step": 720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.99743156221892e-05,
+      "loss": 0.4276,
+      "step": 721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974147548625718e-05,
+      "loss": 0.4651,
+      "step": 722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973978927645617e-05,
+      "loss": 0.4739,
+      "step": 723
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973809759258155e-05,
+      "loss": 0.4545,
+      "step": 724
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997364004347261e-05,
+      "loss": 0.4591,
+      "step": 725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973469780298308e-05,
+      "loss": 0.4564,
+      "step": 726
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973298969744588e-05,
+      "loss": 0.4484,
+      "step": 727
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973127611820823e-05,
+      "loss": 0.4568,
+      "step": 728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972955706536416e-05,
+      "loss": 0.4582,
+      "step": 729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997278325390081e-05,
+      "loss": 0.4517,
+      "step": 730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972610253923462e-05,
+      "loss": 0.4851,
+      "step": 731
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972436706613865e-05,
+      "loss": 0.4704,
+      "step": 732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972262611981556e-05,
+      "loss": 0.4628,
+      "step": 733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997208797003608e-05,
+      "loss": 0.4627,
+      "step": 734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971912780787023e-05,
+      "loss": 0.4524,
+      "step": 735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971737044244e-05,
+      "loss": 0.476,
+      "step": 736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997156076041666e-05,
+      "loss": 0.4211,
+      "step": 737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971383929314673e-05,
+      "loss": 0.4604,
+      "step": 738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971206550947748e-05,
+      "loss": 0.4891,
+      "step": 739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997102862532562e-05,
+      "loss": 0.5097,
+      "step": 740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997085015245805e-05,
+      "loss": 0.4456,
+      "step": 741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970671132354836e-05,
+      "loss": 0.5164,
+      "step": 742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970491565025806e-05,
+      "loss": 0.4368,
+      "step": 743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997031145048081e-05,
+      "loss": 0.4544,
+      "step": 744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970130788729737e-05,
+      "loss": 0.4733,
+      "step": 745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969949579782503e-05,
+      "loss": 0.4687,
+      "step": 746
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996976782364905e-05,
+      "loss": 0.4585,
+      "step": 747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969585520339355e-05,
+      "loss": 0.467,
+      "step": 748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969402669863424e-05,
+      "loss": 0.468,
+      "step": 749
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996921927223129e-05,
+      "loss": 0.4849,
+      "step": 750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969035327453024e-05,
+      "loss": 0.492,
+      "step": 751
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968850835538718e-05,
+      "loss": 0.4591,
+      "step": 752
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968665796498496e-05,
+      "loss": 0.4623,
+      "step": 753
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968480210342515e-05,
+      "loss": 0.4865,
+      "step": 754
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968294077080964e-05,
+      "loss": 0.4611,
+      "step": 755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996810739672405e-05,
+      "loss": 0.4711,
+      "step": 756
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996792016928203e-05,
+      "loss": 0.4923,
+      "step": 757
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967732394765173e-05,
+      "loss": 0.4507,
+      "step": 758
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996754407318378e-05,
+      "loss": 0.4799,
+      "step": 759
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99673552045482e-05,
+      "loss": 0.4675,
+      "step": 760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996716578886879e-05,
+      "loss": 0.4617,
+      "step": 761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966975826155942e-05,
+      "loss": 0.4631,
+      "step": 762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996678531642009e-05,
+      "loss": 0.4781,
+      "step": 763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966594259671685e-05,
+      "loss": 0.4625,
+      "step": 764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966402655921214e-05,
+      "loss": 0.5014,
+      "step": 765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966210505179197e-05,
+      "loss": 0.4418,
+      "step": 766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966017807456178e-05,
+      "loss": 0.4541,
+      "step": 767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965824562762727e-05,
+      "loss": 0.4769,
+      "step": 768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965630771109453e-05,
+      "loss": 0.4761,
+      "step": 769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965436432507e-05,
+      "loss": 0.4853,
+      "step": 770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965241546966022e-05,
+      "loss": 0.5016,
+      "step": 771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996504611449722e-05,
+      "loss": 0.4861,
+      "step": 772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964850135111322e-05,
+      "loss": 0.4754,
+      "step": 773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964653608819083e-05,
+      "loss": 0.4576,
+      "step": 774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964456535631287e-05,
+      "loss": 0.4524,
+      "step": 775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996425891555875e-05,
+      "loss": 0.4807,
+      "step": 776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996406074861232e-05,
+      "loss": 0.4845,
+      "step": 777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963862034802876e-05,
+      "loss": 0.4742,
+      "step": 778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996366277414132e-05,
+      "loss": 0.5038,
+      "step": 779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963462966638586e-05,
+      "loss": 0.4657,
+      "step": 780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996326261230565e-05,
+      "loss": 0.4655,
+      "step": 781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963061711153493e-05,
+      "loss": 0.4572,
+      "step": 782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962860263193153e-05,
+      "loss": 0.4745,
+      "step": 783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996265826843568e-05,
+      "loss": 0.4979,
+      "step": 784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962455726892163e-05,
+      "loss": 0.4872,
+      "step": 785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996225263857372e-05,
+      "loss": 0.5029,
+      "step": 786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962049003491497e-05,
+      "loss": 0.4678,
+      "step": 787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9961844821656666e-05,
+      "loss": 0.4665,
+      "step": 788
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961640093080435e-05,
+      "loss": 0.4468,
+      "step": 789
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961434817774042e-05,
+      "loss": 0.5028,
+      "step": 790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961228995748752e-05,
+      "loss": 0.4417,
+      "step": 791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996102262701586e-05,
+      "loss": 0.4753,
+      "step": 792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960815711586695e-05,
+      "loss": 0.4549,
+      "step": 793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960608249472614e-05,
+      "loss": 0.4892,
+      "step": 794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960400240684998e-05,
+      "loss": 0.4737,
+      "step": 795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960191685235263e-05,
+      "loss": 0.4629,
+      "step": 796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959982583134865e-05,
+      "loss": 0.4484,
+      "step": 797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995977293439527e-05,
+      "loss": 0.4627,
+      "step": 798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995956273902799e-05,
+      "loss": 0.4548,
+      "step": 799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995935199704456e-05,
+      "loss": 0.4674,
+      "step": 800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959140708456544e-05,
+      "loss": 0.4681,
+      "step": 801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958928873275542e-05,
+      "loss": 0.4336,
+      "step": 802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958716491513175e-05,
+      "loss": 0.4535,
+      "step": 803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958503563181104e-05,
+      "loss": 0.4447,
+      "step": 804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958290088291012e-05,
+      "loss": 0.4562,
+      "step": 805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958076066854622e-05,
+      "loss": 0.4334,
+      "step": 806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995786149888367e-05,
+      "loss": 0.4551,
+      "step": 807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995764638438994e-05,
+      "loss": 0.4471,
+      "step": 808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957430723385234e-05,
+      "loss": 0.4492,
+      "step": 809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957214515881392e-05,
+      "loss": 0.4548,
+      "step": 810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956997761890277e-05,
+      "loss": 0.4326,
+      "step": 811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956780461423787e-05,
+      "loss": 0.4558,
+      "step": 812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995656261449385e-05,
+      "loss": 0.4765,
+      "step": 813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956344221112416e-05,
+      "loss": 0.451,
+      "step": 814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956125281291475e-05,
+      "loss": 0.437,
+      "step": 815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955905795043048e-05,
+      "loss": 0.447,
+      "step": 816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955685762379172e-05,
+      "loss": 0.4415,
+      "step": 817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995546518331193e-05,
+      "loss": 0.4392,
+      "step": 818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955244057853423e-05,
+      "loss": 0.4482,
+      "step": 819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955022386015792e-05,
+      "loss": 0.4582,
+      "step": 820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.99548001678112e-05,
+      "loss": 0.4505,
+      "step": 821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954577403251845e-05,
+      "loss": 0.4676,
+      "step": 822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954354092349953e-05,
+      "loss": 0.4675,
+      "step": 823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954130235117777e-05,
+      "loss": 0.4577,
+      "step": 824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995390583156761e-05,
+      "loss": 0.4503,
+      "step": 825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953680881711757e-05,
+      "loss": 0.4634,
+      "step": 826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953455385562575e-05,
+      "loss": 0.4457,
+      "step": 827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953229343132437e-05,
+      "loss": 0.4583,
+      "step": 828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953002754433743e-05,
+      "loss": 0.4525,
+      "step": 829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995277561947894e-05,
+      "loss": 0.4483,
+      "step": 830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995254793828048e-05,
+      "loss": 0.4564,
+      "step": 831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995231971085087e-05,
+      "loss": 0.4433,
+      "step": 832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952090937202634e-05,
+      "loss": 0.4312,
+      "step": 833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951861617348327e-05,
+      "loss": 0.5094,
+      "step": 834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995163175130053e-05,
+      "loss": 0.4508,
+      "step": 835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995140133907187e-05,
+      "loss": 0.4783,
+      "step": 836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951170380674978e-05,
+      "loss": 0.4391,
+      "step": 837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950938876122542e-05,
+      "loss": 0.4449,
+      "step": 838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950706825427263e-05,
+      "loss": 0.4454,
+      "step": 839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995047422860188e-05,
+      "loss": 0.4609,
+      "step": 840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995024108565915e-05,
+      "loss": 0.4466,
+      "step": 841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995000739661188e-05,
+      "loss": 0.442,
+      "step": 842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.994977316147289e-05,
+      "loss": 0.4583,
+      "step": 843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949538380255035e-05,
+      "loss": 0.4552,
+      "step": 844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.99493030529712e-05,
+      "loss": 0.4915,
+      "step": 845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949067179634304e-05,
+      "loss": 0.4532,
+      "step": 846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994883076025729e-05,
+      "loss": 0.4574,
+      "step": 847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948593794853138e-05,
+      "loss": 0.4448,
+      "step": 848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948356283434846e-05,
+      "loss": 0.4622,
+      "step": 849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948118226015457e-05,
+      "loss": 0.4669,
+      "step": 850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994787962260803e-05,
+      "loss": 0.4607,
+      "step": 851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947640473225664e-05,
+      "loss": 0.4911,
+      "step": 852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947400777881484e-05,
+      "loss": 0.4774,
+      "step": 853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947160536588645e-05,
+      "loss": 0.4703,
+      "step": 854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946919749360334e-05,
+      "loss": 0.4639,
+      "step": 855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994667841620976e-05,
+      "loss": 0.4548,
+      "step": 856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946436537150176e-05,
+      "loss": 0.4688,
+      "step": 857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946194112194853e-05,
+      "loss": 0.4511,
+      "step": 858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945951141357098e-05,
+      "loss": 0.4573,
+      "step": 859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945707624650244e-05,
+      "loss": 0.4371,
+      "step": 860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945463562087658e-05,
+      "loss": 0.4559,
+      "step": 861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 0.4366,
+      "step": 862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.99449737994489e-05,
+      "loss": 0.4852,
+      "step": 863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944728099399603e-05,
+      "loss": 0.4577,
+      "step": 864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944481853548335e-05,
+      "loss": 0.4508,
+      "step": 865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944235061908612e-05,
+      "loss": 0.4443,
+      "step": 866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943987724493973e-05,
+      "loss": 0.4534,
+      "step": 867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943739841317996e-05,
+      "loss": 0.4557,
+      "step": 868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943491412394286e-05,
+      "loss": 0.4531,
+      "step": 869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943242437736474e-05,
+      "loss": 0.4509,
+      "step": 870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994299291735823e-05,
+      "loss": 0.4566,
+      "step": 871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942742851273245e-05,
+      "loss": 0.4525,
+      "step": 872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942492239495244e-05,
+      "loss": 0.4463,
+      "step": 873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942241082037983e-05,
+      "loss": 0.4635,
+      "step": 874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941989378915245e-05,
+      "loss": 0.4667,
+      "step": 875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941737130140843e-05,
+      "loss": 0.4464,
+      "step": 876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941484335728625e-05,
+      "loss": 0.4652,
+      "step": 877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941230995692464e-05,
+      "loss": 0.4609,
+      "step": 878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940977110046264e-05,
+      "loss": 0.4634,
+      "step": 879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940722678803958e-05,
+      "loss": 0.4813,
+      "step": 880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994046770197951e-05,
+      "loss": 0.4715,
+      "step": 881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940212179586915e-05,
+      "loss": 0.4736,
+      "step": 882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939956111640198e-05,
+      "loss": 0.4764,
+      "step": 883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993969949815341e-05,
+      "loss": 0.4645,
+      "step": 884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939442339140638e-05,
+      "loss": 0.4622,
+      "step": 885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939184634615993e-05,
+      "loss": 0.4521,
+      "step": 886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993892638459362e-05,
+      "loss": 0.4457,
+      "step": 887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938667589087694e-05,
+      "loss": 0.4568,
+      "step": 888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938408248112417e-05,
+      "loss": 0.4573,
+      "step": 889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993814836168202e-05,
+      "loss": 0.4544,
+      "step": 890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937887929810772e-05,
+      "loss": 0.4469,
+      "step": 891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937626952512966e-05,
+      "loss": 0.4565,
+      "step": 892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937365429802923e-05,
+      "loss": 0.4422,
+      "step": 893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937103361694992e-05,
+      "loss": 0.4431,
+      "step": 894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936840748203563e-05,
+      "loss": 0.4556,
+      "step": 895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993657758934305e-05,
+      "loss": 0.4555,
+      "step": 896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936313885127888e-05,
+      "loss": 0.4645,
+      "step": 897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993604963557256e-05,
+      "loss": 0.4772,
+      "step": 898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993578484069156e-05,
+      "loss": 0.4442,
+      "step": 899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935519500499427e-05,
+      "loss": 0.4513,
+      "step": 900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935253615010718e-05,
+      "loss": 0.443,
+      "step": 901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934987184240033e-05,
+      "loss": 0.4399,
+      "step": 902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993472020820199e-05,
+      "loss": 0.4663,
+      "step": 903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934452686911244e-05,
+      "loss": 0.4655,
+      "step": 904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934184620382475e-05,
+      "loss": 0.4574,
+      "step": 905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933916008630398e-05,
+      "loss": 0.4424,
+      "step": 906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993364685166975e-05,
+      "loss": 0.4608,
+      "step": 907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993337714951531e-05,
+      "loss": 0.448,
+      "step": 908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933106902181877e-05,
+      "loss": 0.4529,
+      "step": 909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932836109684287e-05,
+      "loss": 0.465,
+      "step": 910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932564772037397e-05,
+      "loss": 0.4524,
+      "step": 911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932292889256097e-05,
+      "loss": 0.4536,
+      "step": 912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932020461355315e-05,
+      "loss": 0.4641,
+      "step": 913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931747488349998e-05,
+      "loss": 0.4669,
+      "step": 914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931473970255136e-05,
+      "loss": 0.4763,
+      "step": 915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931199907085727e-05,
+      "loss": 0.4955,
+      "step": 916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930925298856823e-05,
+      "loss": 0.4506,
+      "step": 917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.99306501455835e-05,
+      "loss": 0.4363,
+      "step": 918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930374447280844e-05,
+      "loss": 0.4537,
+      "step": 919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930098203963998e-05,
+      "loss": 0.4606,
+      "step": 920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929821415648118e-05,
+      "loss": 0.4634,
+      "step": 921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929544082348394e-05,
+      "loss": 0.4454,
+      "step": 922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929266204080052e-05,
+      "loss": 0.4959,
+      "step": 923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928987780858345e-05,
+      "loss": 0.4561,
+      "step": 924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928708812698544e-05,
+      "loss": 0.4414,
+      "step": 925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928429299615967e-05,
+      "loss": 0.4487,
+      "step": 926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9928149241625957e-05,
+      "loss": 0.4599,
+      "step": 927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927868638743875e-05,
+      "loss": 0.4312,
+      "step": 928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992758749098513e-05,
+      "loss": 0.471,
+      "step": 929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992730579836515e-05,
+      "loss": 0.4582,
+      "step": 930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927023560899394e-05,
+      "loss": 0.456,
+      "step": 931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926740778603354e-05,
+      "loss": 0.4499,
+      "step": 932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926457451492544e-05,
+      "loss": 0.4578,
+      "step": 933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926173579582523e-05,
+      "loss": 0.4635,
+      "step": 934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925889162888866e-05,
+      "loss": 0.472,
+      "step": 935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925604201427188e-05,
+      "loss": 0.4568,
+      "step": 936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992531869521312e-05,
+      "loss": 0.443,
+      "step": 937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925032644262337e-05,
+      "loss": 0.4663,
+      "step": 938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924746048590538e-05,
+      "loss": 0.46,
+      "step": 939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992445890821345e-05,
+      "loss": 0.459,
+      "step": 940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924171223146834e-05,
+      "loss": 0.4615,
+      "step": 941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923882993406483e-05,
+      "loss": 0.4626,
+      "step": 942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923594219008212e-05,
+      "loss": 0.47,
+      "step": 943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992330489996787e-05,
+      "loss": 0.4538,
+      "step": 944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992301503630133e-05,
+      "loss": 0.4475,
+      "step": 945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922724628024516e-05,
+      "loss": 0.4694,
+      "step": 946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922433675153355e-05,
+      "loss": 0.4833,
+      "step": 947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922142177703814e-05,
+      "loss": 0.4514,
+      "step": 948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.99218501356919e-05,
+      "loss": 0.4535,
+      "step": 949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921557549133633e-05,
+      "loss": 0.4635,
+      "step": 950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992126441804508e-05,
+      "loss": 0.4452,
+      "step": 951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920970742442324e-05,
+      "loss": 0.4637,
+      "step": 952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992067652234148e-05,
+      "loss": 0.474,
+      "step": 953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.99203817577587e-05,
+      "loss": 0.4687,
+      "step": 954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9920086448710162e-05,
+      "loss": 0.4585,
+      "step": 955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991979059521207e-05,
+      "loss": 0.4587,
+      "step": 956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919494197280663e-05,
+      "loss": 0.4526,
+      "step": 957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919197254932213e-05,
+      "loss": 0.4516,
+      "step": 958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991889976818301e-05,
+      "loss": 0.4517,
+      "step": 959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918601737049387e-05,
+      "loss": 0.4858,
+      "step": 960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.99183031615477e-05,
+      "loss": 0.4652,
+      "step": 961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991800404169433e-05,
+      "loss": 0.4441,
+      "step": 962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.99177043775057e-05,
+      "loss": 0.4525,
+      "step": 963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917404168998256e-05,
+      "loss": 0.4294,
+      "step": 964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917103416188474e-05,
+      "loss": 0.4675,
+      "step": 965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916802119092854e-05,
+      "loss": 0.4801,
+      "step": 966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916500277727943e-05,
+      "loss": 0.4473,
+      "step": 967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916197892110302e-05,
+      "loss": 0.4391,
+      "step": 968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915894962256527e-05,
+      "loss": 0.4564,
+      "step": 969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915591488183243e-05,
+      "loss": 0.4471,
+      "step": 970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915287469907106e-05,
+      "loss": 0.4726,
+      "step": 971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914982907444806e-05,
+      "loss": 0.4266,
+      "step": 972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991467780081305e-05,
+      "loss": 0.4786,
+      "step": 973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991437215002859e-05,
+      "loss": 0.4583,
+      "step": 974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.99140659551082e-05,
+      "loss": 0.4511,
+      "step": 975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913759216068686e-05,
+      "loss": 0.4493,
+      "step": 976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913451932926877e-05,
+      "loss": 0.4313,
+      "step": 977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913144105699643e-05,
+      "loss": 0.4345,
+      "step": 978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991283573440388e-05,
+      "loss": 0.5173,
+      "step": 979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991252681905651e-05,
+      "loss": 0.4684,
+      "step": 980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912217359674488e-05,
+      "loss": 0.5112,
+      "step": 981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911907356274794e-05,
+      "loss": 0.4895,
+      "step": 982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911596808874452e-05,
+      "loss": 0.4135,
+      "step": 983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911285717490496e-05,
+      "loss": 0.4791,
+      "step": 984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910974082140005e-05,
+      "loss": 0.4403,
+      "step": 985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910661902840078e-05,
+      "loss": 0.4476,
+      "step": 986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910349179607857e-05,
+      "loss": 0.4362,
+      "step": 987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910035912460496e-05,
+      "loss": 0.44,
+      "step": 988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909722101415194e-05,
+      "loss": 0.4503,
+      "step": 989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909407746489173e-05,
+      "loss": 0.4515,
+      "step": 990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909092847699684e-05,
+      "loss": 0.4712,
+      "step": 991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908777405064012e-05,
+      "loss": 0.4658,
+      "step": 992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990846141859947e-05,
+      "loss": 0.4305,
+      "step": 993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908144888323398e-05,
+      "loss": 0.4832,
+      "step": 994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990782781425317e-05,
+      "loss": 0.4417,
+      "step": 995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907510196406184e-05,
+      "loss": 0.4509,
+      "step": 996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990719203479988e-05,
+      "loss": 0.443,
+      "step": 997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906873329451713e-05,
+      "loss": 0.4438,
+      "step": 998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906554080379182e-05,
+      "loss": 0.4521,
+      "step": 999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.99062342875998e-05,
+      "loss": 0.438,
+      "step": 1000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905913951131126e-05,
+      "loss": 0.4536,
+      "step": 1001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990559307099073e-05,
+      "loss": 0.4469,
+      "step": 1002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905271647196233e-05,
+      "loss": 0.441,
+      "step": 1003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904949679765276e-05,
+      "loss": 0.46,
+      "step": 1004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904627168715525e-05,
+      "loss": 0.4703,
+      "step": 1005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990430411406468e-05,
+      "loss": 0.4484,
+      "step": 1006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903980515830478e-05,
+      "loss": 0.4626,
+      "step": 1007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903656374030675e-05,
+      "loss": 0.4629,
+      "step": 1008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990333168868306e-05,
+      "loss": 0.4476,
+      "step": 1009
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990300645980545e-05,
+      "loss": 0.4424,
+      "step": 1010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902680687415704e-05,
+      "loss": 0.4425,
+      "step": 1011
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902354371531695e-05,
+      "loss": 0.4823,
+      "step": 1012
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902027512171334e-05,
+      "loss": 0.4351,
+      "step": 1013
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990170010935256e-05,
+      "loss": 0.4471,
+      "step": 1014
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990137216309334e-05,
+      "loss": 0.4402,
+      "step": 1015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901043673411677e-05,
+      "loss": 0.4361,
+      "step": 1016
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900714640325597e-05,
+      "loss": 0.4608,
+      "step": 1017
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900385063853154e-05,
+      "loss": 0.4655,
+      "step": 1018
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900054944012447e-05,
+      "loss": 0.4725,
+      "step": 1019
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899724280821587e-05,
+      "loss": 0.4747,
+      "step": 1020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899393074298724e-05,
+      "loss": 0.4597,
+      "step": 1021
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899061324462036e-05,
+      "loss": 0.4465,
+      "step": 1022
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898729031329725e-05,
+      "loss": 0.4522,
+      "step": 1023
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898396194920036e-05,
+      "loss": 0.4582,
+      "step": 1024
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898062815251232e-05,
+      "loss": 0.4765,
+      "step": 1025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897728892341615e-05,
+      "loss": 0.4171,
+      "step": 1026
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897394426209507e-05,
+      "loss": 0.4658,
+      "step": 1027
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897059416873263e-05,
+      "loss": 0.5116,
+      "step": 1028
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896723864351276e-05,
+      "loss": 0.4662,
+      "step": 1029
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896387768661958e-05,
+      "loss": 0.4586,
+      "step": 1030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896051129823757e-05,
+      "loss": 0.4833,
+      "step": 1031
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895713947855146e-05,
+      "loss": 0.4555,
+      "step": 1032
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895376222774632e-05,
+      "loss": 0.4603,
+      "step": 1033
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895037954600754e-05,
+      "loss": 0.4479,
+      "step": 1034
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894699143352075e-05,
+      "loss": 0.4356,
+      "step": 1035
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894359789047188e-05,
+      "loss": 0.4344,
+      "step": 1036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9894019891704718e-05,
+      "loss": 0.4661,
+      "step": 1037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893679451343324e-05,
+      "loss": 0.4344,
+      "step": 1038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893338467981684e-05,
+      "loss": 0.4708,
+      "step": 1039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989299694163852e-05,
+      "loss": 0.4697,
+      "step": 1040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892654872332572e-05,
+      "loss": 0.4605,
+      "step": 1041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892312260082617e-05,
+      "loss": 0.4177,
+      "step": 1042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891969104907454e-05,
+      "loss": 0.4474,
+      "step": 1043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989162540682592e-05,
+      "loss": 0.4956,
+      "step": 1044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891281165856876e-05,
+      "loss": 0.4435,
+      "step": 1045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890936382019216e-05,
+      "loss": 0.4698,
+      "step": 1046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890591055331866e-05,
+      "loss": 0.4708,
+      "step": 1047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989024518581377e-05,
+      "loss": 0.467,
+      "step": 1048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889898773483923e-05,
+      "loss": 0.4629,
+      "step": 1049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988955181836133e-05,
+      "loss": 0.4815,
+      "step": 1050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889204320465033e-05,
+      "loss": 0.4535,
+      "step": 1051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888856279814103e-05,
+      "loss": 0.4584,
+      "step": 1052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988850769642765e-05,
+      "loss": 0.4605,
+      "step": 1053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888158570324795e-05,
+      "loss": 0.4315,
+      "step": 1054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887808901524703e-05,
+      "loss": 0.4306,
+      "step": 1055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887458690046565e-05,
+      "loss": 0.4436,
+      "step": 1056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887107935909608e-05,
+      "loss": 0.4449,
+      "step": 1057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886756639133075e-05,
+      "loss": 0.4575,
+      "step": 1058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886404799736244e-05,
+      "loss": 0.438,
+      "step": 1059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886052417738436e-05,
+      "loss": 0.4657,
+      "step": 1060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988569949315898e-05,
+      "loss": 0.4594,
+      "step": 1061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988534602601725e-05,
+      "loss": 0.4511,
+      "step": 1062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988499201633265e-05,
+      "loss": 0.4711,
+      "step": 1063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9884637464124602e-05,
+      "loss": 0.4297,
+      "step": 1064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9884282369412572e-05,
+      "loss": 0.4516,
+      "step": 1065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988392673221604e-05,
+      "loss": 0.4551,
+      "step": 1066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883570552554535e-05,
+      "loss": 0.4569,
+      "step": 1067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883213830447597e-05,
+      "loss": 0.4434,
+      "step": 1068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988285656591481e-05,
+      "loss": 0.4572,
+      "step": 1069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882498758975777e-05,
+      "loss": 0.4675,
+      "step": 1070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882140409650138e-05,
+      "loss": 0.4559,
+      "step": 1071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881781517957562e-05,
+      "loss": 0.4395,
+      "step": 1072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881422083917745e-05,
+      "loss": 0.4487,
+      "step": 1073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881062107550414e-05,
+      "loss": 0.4446,
+      "step": 1074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880701588875327e-05,
+      "loss": 0.4721,
+      "step": 1075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880340527912267e-05,
+      "loss": 0.4473,
+      "step": 1076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879978924681053e-05,
+      "loss": 0.4464,
+      "step": 1077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879616779201528e-05,
+      "loss": 0.4621,
+      "step": 1078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879254091493577e-05,
+      "loss": 0.4463,
+      "step": 1079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878890861577092e-05,
+      "loss": 0.485,
+      "step": 1080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987852708947202e-05,
+      "loss": 0.4423,
+      "step": 1081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987816277519832e-05,
+      "loss": 0.4511,
+      "step": 1082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877797918775987e-05,
+      "loss": 0.4499,
+      "step": 1083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987743252022505e-05,
+      "loss": 0.449,
+      "step": 1084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877066579565557e-05,
+      "loss": 0.4633,
+      "step": 1085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876700096817596e-05,
+      "loss": 0.4409,
+      "step": 1086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876333072001282e-05,
+      "loss": 0.4871,
+      "step": 1087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875965505136757e-05,
+      "loss": 0.4711,
+      "step": 1088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875597396244193e-05,
+      "loss": 0.4437,
+      "step": 1089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987522874534379e-05,
+      "loss": 0.454,
+      "step": 1090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874859552455793e-05,
+      "loss": 0.4463,
+      "step": 1091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874489817600454e-05,
+      "loss": 0.4377,
+      "step": 1092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9874119540798067e-05,
+      "loss": 0.4767,
+      "step": 1093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873748722068957e-05,
+      "loss": 0.4575,
+      "step": 1094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873377361433474e-05,
+      "loss": 0.4499,
+      "step": 1095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873005458911996e-05,
+      "loss": 0.4472,
+      "step": 1096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872633014524945e-05,
+      "loss": 0.4477,
+      "step": 1097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987226002829275e-05,
+      "loss": 0.4668,
+      "step": 1098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987188650023589e-05,
+      "loss": 0.4603,
+      "step": 1099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871512430374863e-05,
+      "loss": 0.4669,
+      "step": 1100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871137818730197e-05,
+      "loss": 0.445,
+      "step": 1101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870762665322454e-05,
+      "loss": 0.4614,
+      "step": 1102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870386970172222e-05,
+      "loss": 0.4514,
+      "step": 1103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987001073330012e-05,
+      "loss": 0.4505,
+      "step": 1104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869633954726807e-05,
+      "loss": 0.4531,
+      "step": 1105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986925663447295e-05,
+      "loss": 0.4483,
+      "step": 1106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986887877255926e-05,
+      "loss": 0.4526,
+      "step": 1107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986850036900648e-05,
+      "loss": 0.4621,
+      "step": 1108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868121423835376e-05,
+      "loss": 0.4424,
+      "step": 1109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867741937066744e-05,
+      "loss": 0.4514,
+      "step": 1110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867361908721414e-05,
+      "loss": 0.4316,
+      "step": 1111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986698133882024e-05,
+      "loss": 0.4486,
+      "step": 1112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866600227384115e-05,
+      "loss": 0.4656,
+      "step": 1113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866218574433945e-05,
+      "loss": 0.4539,
+      "step": 1114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986583637999069e-05,
+      "loss": 0.4671,
+      "step": 1115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865453644075318e-05,
+      "loss": 0.4606,
+      "step": 1116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865070366708835e-05,
+      "loss": 0.4473,
+      "step": 1117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986468654791228e-05,
+      "loss": 0.4271,
+      "step": 1118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864302187706717e-05,
+      "loss": 0.4621,
+      "step": 1119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863917286113236e-05,
+      "loss": 0.4339,
+      "step": 1120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863531843152973e-05,
+      "loss": 0.4576,
+      "step": 1121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863145858847074e-05,
+      "loss": 0.4673,
+      "step": 1122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986275933321672e-05,
+      "loss": 0.4762,
+      "step": 1123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862372266283135e-05,
+      "loss": 0.4452,
+      "step": 1124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861984658067554e-05,
+      "loss": 0.4458,
+      "step": 1125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861596508591255e-05,
+      "loss": 0.451,
+      "step": 1126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986120781787554e-05,
+      "loss": 0.4585,
+      "step": 1127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986081858594174e-05,
+      "loss": 0.4533,
+      "step": 1128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860428812811224e-05,
+      "loss": 0.4517,
+      "step": 1129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860038498505375e-05,
+      "loss": 0.4629,
+      "step": 1130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859647643045622e-05,
+      "loss": 0.4557,
+      "step": 1131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985925624645341e-05,
+      "loss": 0.4688,
+      "step": 1132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858864308750223e-05,
+      "loss": 0.4664,
+      "step": 1133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858471829957577e-05,
+      "loss": 0.4596,
+      "step": 1134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858078810097004e-05,
+      "loss": 0.4344,
+      "step": 1135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857685249190083e-05,
+      "loss": 0.448,
+      "step": 1136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857291147258405e-05,
+      "loss": 0.4482,
+      "step": 1137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985689650432361e-05,
+      "loss": 0.4359,
+      "step": 1138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856501320407348e-05,
+      "loss": 0.4223,
+      "step": 1139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985610559553131e-05,
+      "loss": 0.4538,
+      "step": 1140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985570932971722e-05,
+      "loss": 0.4683,
+      "step": 1141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985531252298682e-05,
+      "loss": 0.4529,
+      "step": 1142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985491517536189e-05,
+      "loss": 0.4408,
+      "step": 1143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854517286864246e-05,
+      "loss": 0.4452,
+      "step": 1144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854118857515715e-05,
+      "loss": 0.4732,
+      "step": 1145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853719887338164e-05,
+      "loss": 0.4308,
+      "step": 1146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853320376353496e-05,
+      "loss": 0.4562,
+      "step": 1147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985292032458364e-05,
+      "loss": 0.4492,
+      "step": 1148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852519732050544e-05,
+      "loss": 0.4719,
+      "step": 1149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852118598776195e-05,
+      "loss": 0.4486,
+      "step": 1150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851716924782615e-05,
+      "loss": 0.4243,
+      "step": 1151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851314710091846e-05,
+      "loss": 0.4461,
+      "step": 1152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985091195472596e-05,
+      "loss": 0.426,
+      "step": 1153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850508658707065e-05,
+      "loss": 0.4743,
+      "step": 1154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850104822057292e-05,
+      "loss": 0.4759,
+      "step": 1155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984970044479881e-05,
+      "loss": 0.457,
+      "step": 1156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984929552695381e-05,
+      "loss": 0.4706,
+      "step": 1157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848890068544517e-05,
+      "loss": 0.4492,
+      "step": 1158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984848406959318e-05,
+      "loss": 0.4456,
+      "step": 1159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.4664,
+      "step": 1160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984767045015354e-05,
+      "loss": 0.4713,
+      "step": 1161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847262829709892e-05,
+      "loss": 0.4487,
+      "step": 1162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846854668813514e-05,
+      "loss": 0.4641,
+      "step": 1163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98464459674868e-05,
+      "loss": 0.4582,
+      "step": 1164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846036725752186e-05,
+      "loss": 0.4638,
+      "step": 1165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845626943632135e-05,
+      "loss": 0.448,
+      "step": 1166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845216621149133e-05,
+      "loss": 0.447,
+      "step": 1167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98448057583257e-05,
+      "loss": 0.4605,
+      "step": 1168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844394355184387e-05,
+      "loss": 0.4512,
+      "step": 1169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843982411747773e-05,
+      "loss": 0.4098,
+      "step": 1170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843569928038468e-05,
+      "loss": 0.4823,
+      "step": 1171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984315690407911e-05,
+      "loss": 0.4548,
+      "step": 1172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842743339892367e-05,
+      "loss": 0.4373,
+      "step": 1173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842329235500934e-05,
+      "loss": 0.4466,
+      "step": 1174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9841914590927544e-05,
+      "loss": 0.444,
+      "step": 1175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841499406194948e-05,
+      "loss": 0.4565,
+      "step": 1176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984108368132594e-05,
+      "loss": 0.4737,
+      "step": 1177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840667416343325e-05,
+      "loss": 0.4686,
+      "step": 1178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840250611269963e-05,
+      "loss": 0.4592,
+      "step": 1179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839833266128726e-05,
+      "loss": 0.4385,
+      "step": 1180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839415380942514e-05,
+      "loss": 0.4453,
+      "step": 1181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983899695573426e-05,
+      "loss": 0.465,
+      "step": 1182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983857799052694e-05,
+      "loss": 0.43,
+      "step": 1183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983815848534354e-05,
+      "loss": 0.43,
+      "step": 1184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9837738440207084e-05,
+      "loss": 0.4681,
+      "step": 1185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983731785514063e-05,
+      "loss": 0.4272,
+      "step": 1186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836896730167258e-05,
+      "loss": 0.4508,
+      "step": 1187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983647506531008e-05,
+      "loss": 0.432,
+      "step": 1188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836052860592237e-05,
+      "loss": 0.4355,
+      "step": 1189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983563011603691e-05,
+      "loss": 0.4907,
+      "step": 1190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983520683166729e-05,
+      "loss": 0.4456,
+      "step": 1191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834783007506615e-05,
+      "loss": 0.4582,
+      "step": 1192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834358643578143e-05,
+      "loss": 0.4815,
+      "step": 1193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833933739905164e-05,
+      "loss": 0.484,
+      "step": 1194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833508296511003e-05,
+      "loss": 0.472,
+      "step": 1195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833082313419005e-05,
+      "loss": 0.4387,
+      "step": 1196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983265579065255e-05,
+      "loss": 0.4735,
+      "step": 1197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983222872823505e-05,
+      "loss": 0.454,
+      "step": 1198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9831801126189943e-05,
+      "loss": 0.4554,
+      "step": 1199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9831372984540694e-05,
+      "loss": 0.468,
+      "step": 1200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9830944303310803e-05,
+      "loss": 0.4643,
+      "step": 1201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.98305150825238e-05,
+      "loss": 0.4402,
+      "step": 1202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9830085322203236e-05,
+      "loss": 0.4664,
+      "step": 1203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829655022372703e-05,
+      "loss": 0.4615,
+      "step": 1204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829224183055816e-05,
+      "loss": 0.4634,
+      "step": 1205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828792804276222e-05,
+      "loss": 0.4278,
+      "step": 1206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828360886057597e-05,
+      "loss": 0.434,
+      "step": 1207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982792842842364e-05,
+      "loss": 0.4442,
+      "step": 1208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827495431398094e-05,
+      "loss": 0.4672,
+      "step": 1209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827061895004718e-05,
+      "loss": 0.4556,
+      "step": 1210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982662781926731e-05,
+      "loss": 0.426,
+      "step": 1211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982619320420969e-05,
+      "loss": 0.4708,
+      "step": 1212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825758049855713e-05,
+      "loss": 0.4496,
+      "step": 1213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825322356229262e-05,
+      "loss": 0.4455,
+      "step": 1214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982488612335425e-05,
+      "loss": 0.4319,
+      "step": 1215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982444935125462e-05,
+      "loss": 0.47,
+      "step": 1216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9824012039954336e-05,
+      "loss": 0.4373,
+      "step": 1217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823574189477408e-05,
+      "loss": 0.4529,
+      "step": 1218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823135799847865e-05,
+      "loss": 0.4869,
+      "step": 1219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822696871089762e-05,
+      "loss": 0.453,
+      "step": 1220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822257403227193e-05,
+      "loss": 0.4502,
+      "step": 1221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9821817396284283e-05,
+      "loss": 0.4438,
+      "step": 1222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982137685028517e-05,
+      "loss": 0.4449,
+      "step": 1223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820935765254045e-05,
+      "loss": 0.4459,
+      "step": 1224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98204941412151e-05,
+      "loss": 0.4524,
+      "step": 1225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982005197819259e-05,
+      "loss": 0.4312,
+      "step": 1226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819609276210773e-05,
+      "loss": 0.4481,
+      "step": 1227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.981916603529395e-05,
+      "loss": 0.4593,
+      "step": 1228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818722255466445e-05,
+      "loss": 0.453,
+      "step": 1229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818277936752612e-05,
+      "loss": 0.4373,
+      "step": 1230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817833079176844e-05,
+      "loss": 0.4732,
+      "step": 1231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981738768276355e-05,
+      "loss": 0.4716,
+      "step": 1232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9816941747537176e-05,
+      "loss": 0.4816,
+      "step": 1233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98164952735222e-05,
+      "loss": 0.4557,
+      "step": 1234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981604826074312e-05,
+      "loss": 0.4448,
+      "step": 1235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981560070922448e-05,
+      "loss": 0.4331,
+      "step": 1236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981515261899083e-05,
+      "loss": 0.4555,
+      "step": 1237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814703990066773e-05,
+      "loss": 0.4538,
+      "step": 1238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814254822476925e-05,
+      "loss": 0.4392,
+      "step": 1239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981380511624594e-05,
+      "loss": 0.4403,
+      "step": 1240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813354871398504e-05,
+      "loss": 0.4271,
+      "step": 1241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812904087959322e-05,
+      "loss": 0.4149,
+      "step": 1242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812452765953137e-05,
+      "loss": 0.4446,
+      "step": 1243
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812000905404718e-05,
+      "loss": 0.4181,
+      "step": 1244
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811548506338867e-05,
+      "loss": 0.4869,
+      "step": 1245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981109556878041e-05,
+      "loss": 0.4258,
+      "step": 1246
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981064209275421e-05,
+      "loss": 0.4422,
+      "step": 1247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981018807828515e-05,
+      "loss": 0.4624,
+      "step": 1248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809733525398154e-05,
+      "loss": 0.4569,
+      "step": 1249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809278434118165e-05,
+      "loss": 0.4619,
+      "step": 1250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808822804470158e-05,
+      "loss": 0.4413,
+      "step": 1251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808366636479146e-05,
+      "loss": 0.4649,
+      "step": 1252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807909930170165e-05,
+      "loss": 0.4502,
+      "step": 1253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807452685568273e-05,
+      "loss": 0.434,
+      "step": 1254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806994902698572e-05,
+      "loss": 0.4442,
+      "step": 1255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806536581586187e-05,
+      "loss": 0.434,
+      "step": 1256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806077722256267e-05,
+      "loss": 0.4504,
+      "step": 1257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9805618324734e-05,
+      "loss": 0.4204,
+      "step": 1258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9805158389044597e-05,
+      "loss": 0.4552,
+      "step": 1259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.98046979152133e-05,
+      "loss": 0.4542,
+      "step": 1260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980423690326539e-05,
+      "loss": 0.4284,
+      "step": 1261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980377535322616e-05,
+      "loss": 0.4664,
+      "step": 1262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980331326512094e-05,
+      "loss": 0.4418,
+      "step": 1263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.98028506389751e-05,
+      "loss": 0.4345,
+      "step": 1264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802387474814022e-05,
+      "loss": 0.4743,
+      "step": 1265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9801923772663133e-05,
+      "loss": 0.4579,
+      "step": 1266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9801459532547877e-05,
+      "loss": 0.4527,
+      "step": 1267
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800994754493736e-05,
+      "loss": 0.4679,
+      "step": 1268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980052943852622e-05,
+      "loss": 0.4407,
+      "step": 1269
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800063584670864e-05,
+      "loss": 0.4274,
+      "step": 1270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9799597192953236e-05,
+      "loss": 0.4554,
+      "step": 1271
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9799130263398936e-05,
+      "loss": 0.4174,
+      "step": 1272
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798662796033588e-05,
+      "loss": 0.4365,
+      "step": 1273
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979819479088285e-05,
+      "loss": 0.4415,
+      "step": 1274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797726247972405e-05,
+      "loss": 0.4537,
+      "step": 1275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797257167327974e-05,
+      "loss": 0.4402,
+      "step": 1276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796787548975295e-05,
+      "loss": 0.4411,
+      "step": 1277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796317392940148e-05,
+      "loss": 0.4465,
+      "step": 1278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979584669924833e-05,
+      "loss": 0.446,
+      "step": 1279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795375467925684e-05,
+      "loss": 0.4471,
+      "step": 1280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794903698998067e-05,
+      "loss": 0.4443,
+      "step": 1281
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794431392491367e-05,
+      "loss": 0.4347,
+      "step": 1282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9793958548431518e-05,
+      "loss": 0.4763,
+      "step": 1283
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979348516684446e-05,
+      "loss": 0.4511,
+      "step": 1284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9793011247756177e-05,
+      "loss": 0.448,
+      "step": 1285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979253679119268e-05,
+      "loss": 0.4753,
+      "step": 1286
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979206179718001e-05,
+      "loss": 0.4319,
+      "step": 1287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791586265744238e-05,
+      "loss": 0.4625,
+      "step": 1288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979111019691146e-05,
+      "loss": 0.4264,
+      "step": 1289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790633590707804e-05,
+      "loss": 0.4223,
+      "step": 1290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790156447159425e-05,
+      "loss": 0.4467,
+      "step": 1291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978967876629252e-05,
+      "loss": 0.4304,
+      "step": 1292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9789200548133296e-05,
+      "loss": 0.4616,
+      "step": 1293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788721792708005e-05,
+      "loss": 0.4436,
+      "step": 1294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978824250004292e-05,
+      "loss": 0.4202,
+      "step": 1295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787762670164346e-05,
+      "loss": 0.4601,
+      "step": 1296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787282303098618e-05,
+      "loss": 0.453,
+      "step": 1297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786801398872105e-05,
+      "loss": 0.4359,
+      "step": 1298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786319957511196e-05,
+      "loss": 0.4657,
+      "step": 1299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978583797904231e-05,
+      "loss": 0.4465,
+      "step": 1300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785355463491908e-05,
+      "loss": 0.4519,
+      "step": 1301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784872410886475e-05,
+      "loss": 0.4529,
+      "step": 1302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978438882125251e-05,
+      "loss": 0.456,
+      "step": 1303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978390469461656e-05,
+      "loss": 0.4406,
+      "step": 1304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.97834200310052e-05,
+      "loss": 0.4156,
+      "step": 1305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9782934830445023e-05,
+      "loss": 0.4653,
+      "step": 1306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9782449092962662e-05,
+      "loss": 0.4499,
+      "step": 1307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978196281858478e-05,
+      "loss": 0.4754,
+      "step": 1308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 0.436,
+      "step": 1309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978098865924922e-05,
+      "loss": 0.4486,
+      "step": 1310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978050077434501e-05,
+      "loss": 0.4528,
+      "step": 1311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780012352652203e-05,
+      "loss": 0.4826,
+      "step": 1312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9779523394197606e-05,
+      "loss": 0.4307,
+      "step": 1313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977903389900806e-05,
+      "loss": 0.4533,
+      "step": 1314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9778543867110428e-05,
+      "loss": 0.4451,
+      "step": 1315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.97780532985316e-05,
+      "loss": 0.4379,
+      "step": 1316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9777562193298507e-05,
+      "loss": 0.4364,
+      "step": 1317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.97770705514381e-05,
+      "loss": 0.4555,
+      "step": 1318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776578372977358e-05,
+      "loss": 0.466,
+      "step": 1319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.97760856579433e-05,
+      "loss": 0.4283,
+      "step": 1320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977559240636296e-05,
+      "loss": 0.4227,
+      "step": 1321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977509861826342e-05,
+      "loss": 0.4474,
+      "step": 1322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977460429367177e-05,
+      "loss": 0.4764,
+      "step": 1323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774109432615148e-05,
+      "loss": 0.4788,
+      "step": 1324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977361403512071e-05,
+      "loss": 0.4843,
+      "step": 1325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773118101215648e-05,
+      "loss": 0.4643,
+      "step": 1326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772621630927177e-05,
+      "loss": 0.4623,
+      "step": 1327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772124624282546e-05,
+      "loss": 0.4806,
+      "step": 1328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771627081309037e-05,
+      "loss": 0.4432,
+      "step": 1329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977112900203395e-05,
+      "loss": 0.4507,
+      "step": 1330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770630386484626e-05,
+      "loss": 0.4501,
+      "step": 1331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977013123468843e-05,
+      "loss": 0.4554,
+      "step": 1332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9769631546672756e-05,
+      "loss": 0.4459,
+      "step": 1333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9769131322465028e-05,
+      "loss": 0.4612,
+      "step": 1334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768630562092705e-05,
+      "loss": 0.4656,
+      "step": 1335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768129265583265e-05,
+      "loss": 0.4525,
+      "step": 1336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976762743296422e-05,
+      "loss": 0.4693,
+      "step": 1337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767125064263117e-05,
+      "loss": 0.4291,
+      "step": 1338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9766622159507527e-05,
+      "loss": 0.4579,
+      "step": 1339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976611871872505e-05,
+      "loss": 0.4125,
+      "step": 1340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976561474194332e-05,
+      "loss": 0.4326,
+      "step": 1341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9765110229189987e-05,
+      "loss": 0.4529,
+      "step": 1342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976460518049275e-05,
+      "loss": 0.466,
+      "step": 1343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764099595879327e-05,
+      "loss": 0.4507,
+      "step": 1344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9763593475377464e-05,
+      "loss": 0.4592,
+      "step": 1345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9763086819014936e-05,
+      "loss": 0.4383,
+      "step": 1346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762579626819555e-05,
+      "loss": 0.4539,
+      "step": 1347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976207189881916e-05,
+      "loss": 0.453,
+      "step": 1348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761563635041606e-05,
+      "loss": 0.4476,
+      "step": 1349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97610548355148e-05,
+      "loss": 0.4393,
+      "step": 1350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9760545500266656e-05,
+      "loss": 0.4426,
+      "step": 1351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9760035629325137e-05,
+      "loss": 0.4406,
+      "step": 1352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9759525222718224e-05,
+      "loss": 0.4806,
+      "step": 1353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975901428047393e-05,
+      "loss": 0.4536,
+      "step": 1354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9758502802620295e-05,
+      "loss": 0.443,
+      "step": 1355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757990789185393e-05,
+      "loss": 0.4372,
+      "step": 1356
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757478240197322e-05,
+      "loss": 0.4333,
+      "step": 1357
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975696515568422e-05,
+      "loss": 0.4489,
+      "step": 1358
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9756451535674236e-05,
+      "loss": 0.4359,
+      "step": 1359
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975593738019557e-05,
+      "loss": 0.4287,
+      "step": 1360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755422689276433e-05,
+      "loss": 0.4579,
+      "step": 1361
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754907462945077e-05,
+      "loss": 0.4275,
+      "step": 1362
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975439170122978e-05,
+      "loss": 0.4458,
+      "step": 1363
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9753875404158842e-05,
+      "loss": 0.4661,
+      "step": 1364
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9753358571760607e-05,
+      "loss": 0.4588,
+      "step": 1365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752841204063438e-05,
+      "loss": 0.4369,
+      "step": 1366
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752323301095727e-05,
+      "loss": 0.4516,
+      "step": 1367
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9751804862885908e-05,
+      "loss": 0.4519,
+      "step": 1368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9751285889462424e-05,
+      "loss": 0.4603,
+      "step": 1369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750766380853763e-05,
+      "loss": 0.4389,
+      "step": 1370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750246337088435e-05,
+      "loss": 0.4315,
+      "step": 1371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9749725758194983e-05,
+      "loss": 0.4364,
+      "step": 1372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9749204644201982e-05,
+      "loss": 0.4587,
+      "step": 1373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748682995138027e-05,
+      "loss": 0.4664,
+      "step": 1374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974816081103175e-05,
+      "loss": 0.4309,
+      "step": 1375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974763809191181e-05,
+      "loss": 0.4867,
+      "step": 1376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.97471148378069e-05,
+      "loss": 0.4758,
+      "step": 1377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974659104874573e-05,
+      "loss": 0.426,
+      "step": 1378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974606672475705e-05,
+      "loss": 0.4386,
+      "step": 1379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9745541865869643e-05,
+      "loss": 0.4124,
+      "step": 1380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9745016472112308e-05,
+      "loss": 0.4411,
+      "step": 1381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744490543513883e-05,
+      "loss": 0.445,
+      "step": 1382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743964080103236e-05,
+      "loss": 0.4478,
+      "step": 1383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743437081909254e-05,
+      "loss": 0.431,
+      "step": 1384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742909548960866e-05,
+      "loss": 0.4028,
+      "step": 1385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742381481287025e-05,
+      "loss": 0.463,
+      "step": 1386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974185287891671e-05,
+      "loss": 0.4205,
+      "step": 1387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9741323741878933e-05,
+      "loss": 0.4414,
+      "step": 1388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740794070202737e-05,
+      "loss": 0.4733,
+      "step": 1389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974026386391719e-05,
+      "loss": 0.4563,
+      "step": 1390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739733123051397e-05,
+      "loss": 0.4478,
+      "step": 1391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739201847634478e-05,
+      "loss": 0.4272,
+      "step": 1392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.97386700376956e-05,
+      "loss": 0.4575,
+      "step": 1393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738137693263946e-05,
+      "loss": 0.4718,
+      "step": 1394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9737604814368734e-05,
+      "loss": 0.4634,
+      "step": 1395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973707140103921e-05,
+      "loss": 0.4473,
+      "step": 1396
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736537453304648e-05,
+      "loss": 0.4495,
+      "step": 1397
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736002971194358e-05,
+      "loss": 0.4383,
+      "step": 1398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973546795473767e-05,
+      "loss": 0.4481,
+      "step": 1399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9734932403963945e-05,
+      "loss": 0.4168,
+      "step": 1400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973439631890258e-05,
+      "loss": 0.4355,
+      "step": 1401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9733859699583e-05,
+      "loss": 0.4491,
+      "step": 1402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973332254603465e-05,
+      "loss": 0.4218,
+      "step": 1403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732784858287015e-05,
+      "loss": 0.4379,
+      "step": 1404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732246636369605e-05,
+      "loss": 0.434,
+      "step": 1405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973170788031196e-05,
+      "loss": 0.4916,
+      "step": 1406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9731168590143644e-05,
+      "loss": 0.4605,
+      "step": 1407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730628765894262e-05,
+      "loss": 0.4482,
+      "step": 1408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730088407593437e-05,
+      "loss": 0.434,
+      "step": 1409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972954751527083e-05,
+      "loss": 0.4448,
+      "step": 1410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9729006088956122e-05,
+      "loss": 0.4193,
+      "step": 1411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972846412867903e-05,
+      "loss": 0.4461,
+      "step": 1412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.97279216344693e-05,
+      "loss": 0.447,
+      "step": 1413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727378606356704e-05,
+      "loss": 0.4648,
+      "step": 1414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972683504437105e-05,
+      "loss": 0.458,
+      "step": 1415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726290948542166e-05,
+      "loss": 0.4624,
+      "step": 1416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725746318899914e-05,
+      "loss": 0.432,
+      "step": 1417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725201155474187e-05,
+      "loss": 0.4467,
+      "step": 1418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9724655458294902e-05,
+      "loss": 0.49,
+      "step": 1419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9724109227392016e-05,
+      "loss": 0.429,
+      "step": 1420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9723562462795504e-05,
+      "loss": 0.4428,
+      "step": 1421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972301516453537e-05,
+      "loss": 0.427,
+      "step": 1422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722467332641656e-05,
+      "loss": 0.4462,
+      "step": 1423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721918967144433e-05,
+      "loss": 0.4373,
+      "step": 1424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721370068073794e-05,
+      "loss": 0.4111,
+      "step": 1425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972082063545986e-05,
+      "loss": 0.4413,
+      "step": 1426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9720270669332787e-05,
+      "loss": 0.4389,
+      "step": 1427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719720169722765e-05,
+      "loss": 0.4641,
+      "step": 1428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719169136660002e-05,
+      "loss": 0.4406,
+      "step": 1429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718617570174744e-05,
+      "loss": 0.4641,
+      "step": 1430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971806547029726e-05,
+      "loss": 0.4593,
+      "step": 1431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9717512837057855e-05,
+      "loss": 0.4301,
+      "step": 1432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716959670486855e-05,
+      "loss": 0.4524,
+      "step": 1433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716405970614624e-05,
+      "loss": 0.4571,
+      "step": 1434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9715851737471548e-05,
+      "loss": 0.4448,
+      "step": 1435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9715296971088044e-05,
+      "loss": 0.4315,
+      "step": 1436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714741671494563e-05,
+      "loss": 0.455,
+      "step": 1437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971418583872158e-05,
+      "loss": 0.4458,
+      "step": 1438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97136294727996e-05,
+      "loss": 0.4425,
+      "step": 1439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9713072573759162e-05,
+      "loss": 0.4773,
+      "step": 1440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971251514163083e-05,
+      "loss": 0.4475,
+      "step": 1441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711957176445192e-05,
+      "loss": 0.4649,
+      "step": 1442
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971139867823288e-05,
+      "loss": 0.4502,
+      "step": 1443
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710839647024543e-05,
+      "loss": 0.4424,
+      "step": 1444
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710280082850856e-05,
+      "loss": 0.4503,
+      "step": 1445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9709719985742538e-05,
+      "loss": 0.4651,
+      "step": 1446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9709159355730326e-05,
+      "loss": 0.4518,
+      "step": 1447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970859819284499e-05,
+      "loss": 0.4647,
+      "step": 1448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708036497117328e-05,
+      "loss": 0.4343,
+      "step": 1449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9707474268578172e-05,
+      "loss": 0.4205,
+      "step": 1450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970691150725837e-05,
+      "loss": 0.445,
+      "step": 1451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9706348213188814e-05,
+      "loss": 0.4428,
+      "step": 1452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970578438640042e-05,
+      "loss": 0.4143,
+      "step": 1453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9705220026924135e-05,
+      "loss": 0.4627,
+      "step": 1454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9704655134790926e-05,
+      "loss": 0.4514,
+      "step": 1455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9704089710031805e-05,
+      "loss": 0.462,
+      "step": 1456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.97035237526778e-05,
+      "loss": 0.4379,
+      "step": 1457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9702957262759964e-05,
+      "loss": 0.4535,
+      "step": 1458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9702390240309402e-05,
+      "loss": 0.4551,
+      "step": 1459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9701822685357233e-05,
+      "loss": 0.4633,
+      "step": 1460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.97012545979346e-05,
+      "loss": 0.4498,
+      "step": 1461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9700685978072683e-05,
+      "loss": 0.4498,
+      "step": 1462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9700116825802688e-05,
+      "loss": 0.467,
+      "step": 1463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969954714115586e-05,
+      "loss": 0.4344,
+      "step": 1464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969897692416346e-05,
+      "loss": 0.442,
+      "step": 1465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969840617485678e-05,
+      "loss": 0.4215,
+      "step": 1466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697834893267146e-05,
+      "loss": 0.4374,
+      "step": 1467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969726307942592e-05,
+      "loss": 0.4661,
+      "step": 1468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969669073336448e-05,
+      "loss": 0.4775,
+      "step": 1469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9696117855114234e-05,
+      "loss": 0.5101,
+      "step": 1470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969554444470663e-05,
+      "loss": 0.4776,
+      "step": 1471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694970502173137e-05,
+      "loss": 0.4586,
+      "step": 1472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694396027545252e-05,
+      "loss": 0.4402,
+      "step": 1473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969382102085451e-05,
+      "loss": 0.4535,
+      "step": 1474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9693245482132464e-05,
+      "loss": 0.4442,
+      "step": 1475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692669411410703e-05,
+      "loss": 0.4395,
+      "step": 1476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692092808720846e-05,
+      "loss": 0.4421,
+      "step": 1477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9691515674094536e-05,
+      "loss": 0.4858,
+      "step": 1478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969093800756345e-05,
+      "loss": 0.4327,
+      "step": 1479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9690359809159294e-05,
+      "loss": 0.4451,
+      "step": 1480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689781078913798e-05,
+      "loss": 0.4319,
+      "step": 1481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689201816858724e-05,
+      "loss": 0.4513,
+      "step": 1482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968862202302587e-05,
+      "loss": 0.4356,
+      "step": 1483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688041697447055e-05,
+      "loss": 0.4041,
+      "step": 1484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687460840154123e-05,
+      "loss": 0.4796,
+      "step": 1485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686879451178962e-05,
+      "loss": 0.463,
+      "step": 1486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686297530553476e-05,
+      "loss": 0.4799,
+      "step": 1487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9685715078309603e-05,
+      "loss": 0.4624,
+      "step": 1488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968513209447931e-05,
+      "loss": 0.4481,
+      "step": 1489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9684548579094595e-05,
+      "loss": 0.4659,
+      "step": 1490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9683964532187482e-05,
+      "loss": 0.4469,
+      "step": 1491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968337995379003e-05,
+      "loss": 0.4519,
+      "step": 1492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682794843934315e-05,
+      "loss": 0.4702,
+      "step": 1493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682209202652454e-05,
+      "loss": 0.4772,
+      "step": 1494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968162302997659e-05,
+      "loss": 0.4694,
+      "step": 1495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968103632593889e-05,
+      "loss": 0.4713,
+      "step": 1496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9680449090571562e-05,
+      "loss": 0.4707,
+      "step": 1497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9679861323906827e-05,
+      "loss": 0.4566,
+      "step": 1498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967927302597695e-05,
+      "loss": 0.4536,
+      "step": 1499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967868419681422e-05,
+      "loss": 0.4466,
+      "step": 1500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9678094836450945e-05,
+      "loss": 0.4392,
+      "step": 1501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967750494491948e-05,
+      "loss": 0.4429,
+      "step": 1502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9676914522252193e-05,
+      "loss": 0.4338,
+      "step": 1503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.96763235684815e-05,
+      "loss": 0.4559,
+      "step": 1504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967573208363982e-05,
+      "loss": 0.453,
+      "step": 1505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9675140067759627e-05,
+      "loss": 0.4534,
+      "step": 1506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967454752087341e-05,
+      "loss": 0.4623,
+      "step": 1507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9673954443013684e-05,
+      "loss": 0.4315,
+      "step": 1508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967336083421301e-05,
+      "loss": 0.4445,
+      "step": 1509
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967276669450396e-05,
+      "loss": 0.4421,
+      "step": 1510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9672172023919144e-05,
+      "loss": 0.4318,
+      "step": 1511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96715768224912e-05,
+      "loss": 0.5111,
+      "step": 1512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670981090252794e-05,
+      "loss": 0.4641,
+      "step": 1513
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670384827236622e-05,
+      "loss": 0.4662,
+      "step": 1514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966978803347541e-05,
+      "loss": 0.4286,
+      "step": 1515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9669190709001912e-05,
+      "loss": 0.4313,
+      "step": 1516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9668592853848908e-05,
+      "loss": 0.4065,
+      "step": 1517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667994468049216e-05,
+      "loss": 0.4575,
+      "step": 1518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667395551635676e-05,
+      "loss": 0.4157,
+      "step": 1519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9666796104641158e-05,
+      "loss": 0.4164,
+      "step": 1520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9666196127098558e-05,
+      "loss": 0.4458,
+      "step": 1521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966559561904081e-05,
+      "loss": 0.4499,
+      "step": 1522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966499458050087e-05,
+      "loss": 0.487,
+      "step": 1523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9664393011511725e-05,
+      "loss": 0.4912,
+      "step": 1524
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9663790912106395e-05,
+      "loss": 0.473,
+      "step": 1525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966318828231792e-05,
+      "loss": 0.4201,
+      "step": 1526
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9662585122179372e-05,
+      "loss": 0.5038,
+      "step": 1527
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661981431723864e-05,
+      "loss": 0.4536,
+      "step": 1528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661377210984525e-05,
+      "loss": 0.4787,
+      "step": 1529
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966077245999451e-05,
+      "loss": 0.4486,
+      "step": 1530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966016717878702e-05,
+      "loss": 0.453,
+      "step": 1531
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.965956136739527e-05,
+      "loss": 0.4982,
+      "step": 1532
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658955025852506e-05,
+      "loss": 0.4382,
+      "step": 1533
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658348154192013e-05,
+      "loss": 0.4552,
+      "step": 1534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965774075244709e-05,
+      "loss": 0.4555,
+      "step": 1535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9657132820651082e-05,
+      "loss": 0.4747,
+      "step": 1536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9656524358837352e-05,
+      "loss": 0.4452,
+      "step": 1537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655915367039292e-05,
+      "loss": 0.4624,
+      "step": 1538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655305845290328e-05,
+      "loss": 0.45,
+      "step": 1539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9654695793623908e-05,
+      "loss": 0.4419,
+      "step": 1540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965408521207352e-05,
+      "loss": 0.4671,
+      "step": 1541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965347410067267e-05,
+      "loss": 0.437,
+      "step": 1542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.96528624594549e-05,
+      "loss": 0.4388,
+      "step": 1543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652250288453785e-05,
+      "loss": 0.4428,
+      "step": 1544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9651637587702912e-05,
+      "loss": 0.4285,
+      "step": 1545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9651024357235916e-05,
+      "loss": 0.4521,
+      "step": 1546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965041059708645e-05,
+      "loss": 0.4553,
+      "step": 1547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.96497963072882e-05,
+      "loss": 0.4269,
+      "step": 1548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964918148787488e-05,
+      "loss": 0.4566,
+      "step": 1549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9648566138880234e-05,
+      "loss": 0.4217,
+      "step": 1550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647950260338035e-05,
+      "loss": 0.446,
+      "step": 1551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647333852282086e-05,
+      "loss": 0.4532,
+      "step": 1552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9646716914746214e-05,
+      "loss": 0.4546,
+      "step": 1553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964609944776428e-05,
+      "loss": 0.4473,
+      "step": 1554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645481451370175e-05,
+      "loss": 0.4697,
+      "step": 1555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9644862925597813e-05,
+      "loss": 0.4363,
+      "step": 1556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9644243870481142e-05,
+      "loss": 0.4646,
+      "step": 1557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9643624286054145e-05,
+      "loss": 0.449,
+      "step": 1558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9643004172350814e-05,
+      "loss": 0.4453,
+      "step": 1559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9642383529405194e-05,
+      "loss": 0.465,
+      "step": 1560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641762357251343e-05,
+      "loss": 0.4312,
+      "step": 1561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641140655923358e-05,
+      "loss": 0.4686,
+      "step": 1562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.964051842545535e-05,
+      "loss": 0.4376,
+      "step": 1563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9639895665881478e-05,
+      "loss": 0.4499,
+      "step": 1564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9639272377235922e-05,
+      "loss": 0.4703,
+      "step": 1565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9638648559552885e-05,
+      "loss": 0.4225,
+      "step": 1566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9638024212866606e-05,
+      "loss": 0.4578,
+      "step": 1567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963739933721135e-05,
+      "loss": 0.4258,
+      "step": 1568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963677393262142e-05,
+      "loss": 0.4477,
+      "step": 1569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963614799913113e-05,
+      "loss": 0.4601,
+      "step": 1570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635521536774837e-05,
+      "loss": 0.4168,
+      "step": 1571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963489454558693e-05,
+      "loss": 0.4405,
+      "step": 1572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963426702560181e-05,
+      "loss": 0.4494,
+      "step": 1573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9633638976853925e-05,
+      "loss": 0.4292,
+      "step": 1574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963301039937774e-05,
+      "loss": 0.4457,
+      "step": 1575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963238129320776e-05,
+      "loss": 0.4437,
+      "step": 1576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631751658378506e-05,
+      "loss": 0.4272,
+      "step": 1577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963112149492454e-05,
+      "loss": 0.4391,
+      "step": 1578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9630490802880437e-05,
+      "loss": 0.4322,
+      "step": 1579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9629859582280827e-05,
+      "loss": 0.4666,
+      "step": 1580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962922783316034e-05,
+      "loss": 0.4601,
+      "step": 1581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9628595555553657e-05,
+      "loss": 0.452,
+      "step": 1582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9627962749495476e-05,
+      "loss": 0.447,
+      "step": 1583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9627329415020534e-05,
+      "loss": 0.4586,
+      "step": 1584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962669555216358e-05,
+      "loss": 0.4488,
+      "step": 1585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626061160959406e-05,
+      "loss": 0.4585,
+      "step": 1586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625426241442836e-05,
+      "loss": 0.4573,
+      "step": 1587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962479079364871e-05,
+      "loss": 0.4414,
+      "step": 1588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9624154817611907e-05,
+      "loss": 0.4274,
+      "step": 1589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9623518313367332e-05,
+      "loss": 0.44,
+      "step": 1590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9622881280949915e-05,
+      "loss": 0.445,
+      "step": 1591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962224372039462e-05,
+      "loss": 0.451,
+      "step": 1592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621605631736443e-05,
+      "loss": 0.4738,
+      "step": 1593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9620967015010397e-05,
+      "loss": 0.4797,
+      "step": 1594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9620327870251536e-05,
+      "loss": 0.4638,
+      "step": 1595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961968819749494e-05,
+      "loss": 0.4357,
+      "step": 1596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9619047996775713e-05,
+      "loss": 0.4309,
+      "step": 1597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9618407268128995e-05,
+      "loss": 0.4919,
+      "step": 1598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617766011589946e-05,
+      "loss": 0.4386,
+      "step": 1599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617124227193764e-05,
+      "loss": 0.4292,
+      "step": 1600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9616481914975675e-05,
+      "loss": 0.4769,
+      "step": 1601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961583907497093e-05,
+      "loss": 0.4807,
+      "step": 1602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9615195707214805e-05,
+      "loss": 0.4993,
+      "step": 1603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9614551811742613e-05,
+      "loss": 0.4503,
+      "step": 1604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9613907388589698e-05,
+      "loss": 0.4973,
+      "step": 1605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9613262437791425e-05,
+      "loss": 0.4255,
+      "step": 1606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961261695938319e-05,
+      "loss": 0.4872,
+      "step": 1607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961197095340042e-05,
+      "loss": 0.4467,
+      "step": 1608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961132441987857e-05,
+      "loss": 0.4543,
+      "step": 1609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610677358853127e-05,
+      "loss": 0.4481,
+      "step": 1610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.96100297703596e-05,
+      "loss": 0.4533,
+      "step": 1611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960938165443353e-05,
+      "loss": 0.4637,
+      "step": 1612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960873301111049e-05,
+      "loss": 0.4371,
+      "step": 1613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608083840426083e-05,
+      "loss": 0.4399,
+      "step": 1614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960743414241593e-05,
+      "loss": 0.4521,
+      "step": 1615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9606783917115698e-05,
+      "loss": 0.4563,
+      "step": 1616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960613316456107e-05,
+      "loss": 0.431,
+      "step": 1617
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9605481884787756e-05,
+      "loss": 0.4095,
+      "step": 1618
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960483007783151e-05,
+      "loss": 0.482,
+      "step": 1619
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.96041777437281e-05,
+      "loss": 0.4416,
+      "step": 1620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9603524882513326e-05,
+      "loss": 0.4539,
+      "step": 1621
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602871494223022e-05,
+      "loss": 0.4876,
+      "step": 1622
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602217578893053e-05,
+      "loss": 0.441,
+      "step": 1623
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9601563136559303e-05,
+      "loss": 0.4526,
+      "step": 1624
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960090816725769e-05,
+      "loss": 0.4653,
+      "step": 1625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9600252671024164e-05,
+      "loss": 0.4604,
+      "step": 1626
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.95995966478947e-05,
+      "loss": 0.4832,
+      "step": 1627
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.95989400979053e-05,
+      "loss": 0.4655,
+      "step": 1628
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9598283021092e-05,
+      "loss": 0.4433,
+      "step": 1629
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9597625417490863e-05,
+      "loss": 0.4421,
+      "step": 1630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9596967287137978e-05,
+      "loss": 0.4403,
+      "step": 1631
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959630863006947e-05,
+      "loss": 0.4471,
+      "step": 1632
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9595649446321484e-05,
+      "loss": 0.467,
+      "step": 1633
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9594989735930205e-05,
+      "loss": 0.4572,
+      "step": 1634
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959432949893183e-05,
+      "loss": 0.4655,
+      "step": 1635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593668735362605e-05,
+      "loss": 0.4345,
+      "step": 1636
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593007445258786e-05,
+      "loss": 0.4026,
+      "step": 1637
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9592345628656678e-05,
+      "loss": 0.4749,
+      "step": 1638
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9591683285592595e-05,
+      "loss": 0.4497,
+      "step": 1639
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959102041610289e-05,
+      "loss": 0.4455,
+      "step": 1640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590357020223945e-05,
+      "loss": 0.447,
+      "step": 1641
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589693097992172e-05,
+      "loss": 0.4423,
+      "step": 1642
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589028649444008e-05,
+      "loss": 0.4286,
+      "step": 1643
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9588363674615917e-05,
+      "loss": 0.4443,
+      "step": 1644
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9587698173544396e-05,
+      "loss": 0.4527,
+      "step": 1645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9587032146265973e-05,
+      "loss": 0.4554,
+      "step": 1646
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.95863655928172e-05,
+      "loss": 0.4476,
+      "step": 1647
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9585698513234663e-05,
+      "loss": 0.4718,
+      "step": 1648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958503090755497e-05,
+      "loss": 0.4459,
+      "step": 1649
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958436277581476e-05,
+      "loss": 0.4423,
+      "step": 1650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9583694118050708e-05,
+      "loss": 0.4424,
+      "step": 1651
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9583024934299507e-05,
+      "loss": 0.4576,
+      "step": 1652
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958235522459789e-05,
+      "loss": 0.443,
+      "step": 1653
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9581684988982607e-05,
+      "loss": 0.4613,
+      "step": 1654
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9581014227490443e-05,
+      "loss": 0.4282,
+      "step": 1655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9580342940158217e-05,
+      "loss": 0.4457,
+      "step": 1656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957967112702277e-05,
+      "loss": 0.4661,
+      "step": 1657
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957899878812097e-05,
+      "loss": 0.4392,
+      "step": 1658
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957832592348972e-05,
+      "loss": 0.4417,
+      "step": 1659
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9577652533165952e-05,
+      "loss": 0.4683,
+      "step": 1660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576978617186616e-05,
+      "loss": 0.4623,
+      "step": 1661
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957630417558871e-05,
+      "loss": 0.4276,
+      "step": 1662
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9575629208409233e-05,
+      "loss": 0.4766,
+      "step": 1663
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957495371568525e-05,
+      "loss": 0.4633,
+      "step": 1664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9574277697453824e-05,
+      "loss": 0.4451,
+      "step": 1665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9573601153752053e-05,
+      "loss": 0.4271,
+      "step": 1666
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9572924084617076e-05,
+      "loss": 0.4483,
+      "step": 1667
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9572246490086048e-05,
+      "loss": 0.4365,
+      "step": 1668
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9571568370196163e-05,
+      "loss": 0.3985,
+      "step": 1669
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570889724984633e-05,
+      "loss": 0.4297,
+      "step": 1670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570210554488708e-05,
+      "loss": 0.4343,
+      "step": 1671
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9569530858745663e-05,
+      "loss": 0.4461,
+      "step": 1672
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.95688506377928e-05,
+      "loss": 0.4776,
+      "step": 1673
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956816989166745e-05,
+      "loss": 0.4261,
+      "step": 1674
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9567488620406984e-05,
+      "loss": 0.4622,
+      "step": 1675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956680682404878e-05,
+      "loss": 0.5034,
+      "step": 1676
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9566124502630267e-05,
+      "loss": 0.4367,
+      "step": 1677
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9565441656188892e-05,
+      "loss": 0.443,
+      "step": 1678
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9564758284762127e-05,
+      "loss": 0.4396,
+      "step": 1679
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9564074388387483e-05,
+      "loss": 0.452,
+      "step": 1680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9563389967102493e-05,
+      "loss": 0.4437,
+      "step": 1681
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956270502094472e-05,
+      "loss": 0.4029,
+      "step": 1682
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9562019549951756e-05,
+      "loss": 0.4593,
+      "step": 1683
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9561333554161225e-05,
+      "loss": 0.5146,
+      "step": 1684
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956064703361077e-05,
+      "loss": 0.4541,
+      "step": 1685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9559959988338074e-05,
+      "loss": 0.464,
+      "step": 1686
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955927241838085e-05,
+      "loss": 0.4543,
+      "step": 1687
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9558584323776825e-05,
+      "loss": 0.4378,
+      "step": 1688
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9557895704563768e-05,
+      "loss": 0.4396,
+      "step": 1689
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9557206560779477e-05,
+      "loss": 0.4533,
+      "step": 1690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9556516892461762e-05,
+      "loss": 0.4408,
+      "step": 1691
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955582669964849e-05,
+      "loss": 0.427,
+      "step": 1692
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955513598237753e-05,
+      "loss": 0.4657,
+      "step": 1693
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.95544447406868e-05,
+      "loss": 0.4618,
+      "step": 1694
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955375297461423e-05,
+      "loss": 0.4524,
+      "step": 1695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955306068419779e-05,
+      "loss": 0.4395,
+      "step": 1696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9552367869475474e-05,
+      "loss": 0.4849,
+      "step": 1697
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551674530485307e-05,
+      "loss": 0.4586,
+      "step": 1698
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9550980667265344e-05,
+      "loss": 0.4873,
+      "step": 1699
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955028627985366e-05,
+      "loss": 0.4366,
+      "step": 1700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9549591368288376e-05,
+      "loss": 0.4813,
+      "step": 1701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954889593260762e-05,
+      "loss": 0.4265,
+      "step": 1702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548199972849565e-05,
+      "loss": 0.41,
+      "step": 1703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954750348905241e-05,
+      "loss": 0.432,
+      "step": 1704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954680648125438e-05,
+      "loss": 0.4528,
+      "step": 1705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9546108949493726e-05,
+      "loss": 0.4891,
+      "step": 1706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545410893808734e-05,
+      "loss": 0.4286,
+      "step": 1707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9544712314237716e-05,
+      "loss": 0.4274,
+      "step": 1708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954401321081901e-05,
+      "loss": 0.4282,
+      "step": 1709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9543313583590985e-05,
+      "loss": 0.4791,
+      "step": 1710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954261343259204e-05,
+      "loss": 0.4558,
+      "step": 1711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95419127578606e-05,
+      "loss": 0.4405,
+      "step": 1712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541211559435126e-05,
+      "loss": 0.452,
+      "step": 1713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9540509837354097e-05,
+      "loss": 0.4541,
+      "step": 1714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539807591656032e-05,
+      "loss": 0.4505,
+      "step": 1715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539104822379465e-05,
+      "loss": 0.4639,
+      "step": 1716
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953840152956297e-05,
+      "loss": 0.4383,
+      "step": 1717
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953769771324515e-05,
+      "loss": 0.4321,
+      "step": 1718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9536993373464626e-05,
+      "loss": 0.4498,
+      "step": 1719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953628851026006e-05,
+      "loss": 0.4391,
+      "step": 1720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953558312367013e-05,
+      "loss": 0.4174,
+      "step": 1721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953487721373356e-05,
+      "loss": 0.4399,
+      "step": 1722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9534170780489088e-05,
+      "loss": 0.4644,
+      "step": 1723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533463823975488e-05,
+      "loss": 0.4183,
+      "step": 1724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953275634423155e-05,
+      "loss": 0.4778,
+      "step": 1725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953204834129612e-05,
+      "loss": 0.4805,
+      "step": 1726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9531339815208043e-05,
+      "loss": 0.4356,
+      "step": 1727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953063076600621e-05,
+      "loss": 0.4188,
+      "step": 1728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9529921193729535e-05,
+      "loss": 0.4496,
+      "step": 1729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952921109841696e-05,
+      "loss": 0.4398,
+      "step": 1730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528500480107464e-05,
+      "loss": 0.446,
+      "step": 1731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9527789338840043e-05,
+      "loss": 0.4355,
+      "step": 1732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9527077674653725e-05,
+      "loss": 0.4508,
+      "step": 1733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9526365487587575e-05,
+      "loss": 0.4365,
+      "step": 1734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952565277768068e-05,
+      "loss": 0.4693,
+      "step": 1735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9524939544972145e-05,
+      "loss": 0.4546,
+      "step": 1736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9524225789501128e-05,
+      "loss": 0.4361,
+      "step": 1737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9523511511306794e-05,
+      "loss": 0.4345,
+      "step": 1738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952279671042835e-05,
+      "loss": 0.4655,
+      "step": 1739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9522081386905025e-05,
+      "loss": 0.4166,
+      "step": 1740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952136554077608e-05,
+      "loss": 0.4565,
+      "step": 1741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9520649172080803e-05,
+      "loss": 0.4675,
+      "step": 1742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9519932280858507e-05,
+      "loss": 0.4497,
+      "step": 1743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951921486714854e-05,
+      "loss": 0.4497,
+      "step": 1744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951849693099028e-05,
+      "loss": 0.4402,
+      "step": 1745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951777847242312e-05,
+      "loss": 0.4257,
+      "step": 1746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.95170594914865e-05,
+      "loss": 0.4538,
+      "step": 1747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951633998821988e-05,
+      "loss": 0.4397,
+      "step": 1748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951561996266275e-05,
+      "loss": 0.482,
+      "step": 1749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951489941485462e-05,
+      "loss": 0.4331,
+      "step": 1750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951417834483504e-05,
+      "loss": 0.4542,
+      "step": 1751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951345675264359e-05,
+      "loss": 0.4188,
+      "step": 1752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951273463831987e-05,
+      "loss": 0.4424,
+      "step": 1753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9512012001903505e-05,
+      "loss": 0.4447,
+      "step": 1754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9511288843434166e-05,
+      "loss": 0.4511,
+      "step": 1755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.4582,
+      "step": 1756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950984096049534e-05,
+      "loss": 0.4482,
+      "step": 1757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509116236105318e-05,
+      "loss": 0.4222,
+      "step": 1758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9508390989821252e-05,
+      "loss": 0.472,
+      "step": 1759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9507665221682936e-05,
+      "loss": 0.4383,
+      "step": 1760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9506938931730213e-05,
+      "loss": 0.4409,
+      "step": 1761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9506212120002938e-05,
+      "loss": 0.4038,
+      "step": 1762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9505484786541006e-05,
+      "loss": 0.4495,
+      "step": 1763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9504756931384333e-05,
+      "loss": 0.4421,
+      "step": 1764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9504028554572865e-05,
+      "loss": 0.4512,
+      "step": 1765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950329965614658e-05,
+      "loss": 0.4396,
+      "step": 1766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950257023614548e-05,
+      "loss": 0.4507,
+      "step": 1767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9501840294609602e-05,
+      "loss": 0.4389,
+      "step": 1768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9501109831579004e-05,
+      "loss": 0.4483,
+      "step": 1769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9500378847093778e-05,
+      "loss": 0.4326,
+      "step": 1770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9499647341194047e-05,
+      "loss": 0.4418,
+      "step": 1771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949891531391995e-05,
+      "loss": 0.4615,
+      "step": 1772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949818276531167e-05,
+      "loss": 0.4545,
+      "step": 1773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949744969540941e-05,
+      "loss": 0.4444,
+      "step": 1774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.94967161042534e-05,
+      "loss": 0.4285,
+      "step": 1775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9495981991883908e-05,
+      "loss": 0.4244,
+      "step": 1776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9495247358341223e-05,
+      "loss": 0.4427,
+      "step": 1777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9494512203665664e-05,
+      "loss": 0.4567,
+      "step": 1778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9493776527897574e-05,
+      "loss": 0.4367,
+      "step": 1779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9493040331077336e-05,
+      "loss": 0.4454,
+      "step": 1780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9492303613245353e-05,
+      "loss": 0.4428,
+      "step": 1781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949156637444206e-05,
+      "loss": 0.4282,
+      "step": 1782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9490828614707917e-05,
+      "loss": 0.4571,
+      "step": 1783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9490090334083416e-05,
+      "loss": 0.4459,
+      "step": 1784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9489351532609075e-05,
+      "loss": 0.4678,
+      "step": 1785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9488612210325445e-05,
+      "loss": 0.4294,
+      "step": 1786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.94878723672731e-05,
+      "loss": 0.451,
+      "step": 1787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9487132003492645e-05,
+      "loss": 0.455,
+      "step": 1788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9486391119024716e-05,
+      "loss": 0.4027,
+      "step": 1789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9485649713909972e-05,
+      "loss": 0.425,
+      "step": 1790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948490778818911e-05,
+      "loss": 0.4083,
+      "step": 1791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9484165341902847e-05,
+      "loss": 0.4606,
+      "step": 1792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9483422375091926e-05,
+      "loss": 0.4805,
+      "step": 1793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9482678887797125e-05,
+      "loss": 0.4921,
+      "step": 1794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9481934880059257e-05,
+      "loss": 0.4576,
+      "step": 1795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9481190351919145e-05,
+      "loss": 0.4238,
+      "step": 1796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948044530341766e-05,
+      "loss": 0.4435,
+      "step": 1797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9479699734595693e-05,
+      "loss": 0.4139,
+      "step": 1798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9478953645494155e-05,
+      "loss": 0.4533,
+      "step": 1799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9478207036154e-05,
+      "loss": 0.46,
+      "step": 1800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947745990661621e-05,
+      "loss": 0.4235,
+      "step": 1801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947671225692178e-05,
+      "loss": 0.4227,
+      "step": 1802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9475964087111744e-05,
+      "loss": 0.4523,
+      "step": 1803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9475215397227167e-05,
+      "loss": 0.4409,
+      "step": 1804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9474466187309144e-05,
+      "loss": 0.4517,
+      "step": 1805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947371645739879e-05,
+      "loss": 0.466,
+      "step": 1806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9472966207537257e-05,
+      "loss": 0.4443,
+      "step": 1807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9472215437765713e-05,
+      "loss": 0.4245,
+      "step": 1808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947146414812537e-05,
+      "loss": 0.4427,
+      "step": 1809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947071233865746e-05,
+      "loss": 0.4419,
+      "step": 1810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9469960009403244e-05,
+      "loss": 0.4518,
+      "step": 1811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946920716040401e-05,
+      "loss": 0.4309,
+      "step": 1812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9468453791701082e-05,
+      "loss": 0.4376,
+      "step": 1813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9467699903335807e-05,
+      "loss": 0.4334,
+      "step": 1814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9466945495349558e-05,
+      "loss": 0.4612,
+      "step": 1815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946619056778374e-05,
+      "loss": 0.4417,
+      "step": 1816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946543512067979e-05,
+      "loss": 0.443,
+      "step": 1817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9464679154079165e-05,
+      "loss": 0.4603,
+      "step": 1818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9463922668023358e-05,
+      "loss": 0.455,
+      "step": 1819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9463165662553885e-05,
+      "loss": 0.4498,
+      "step": 1820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9462408137712298e-05,
+      "loss": 0.4182,
+      "step": 1821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9461650093540166e-05,
+      "loss": 0.4577,
+      "step": 1822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9460891530079094e-05,
+      "loss": 0.4682,
+      "step": 1823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946013244737072e-05,
+      "loss": 0.4355,
+      "step": 1824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9459372845456706e-05,
+      "loss": 0.4719,
+      "step": 1825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9458612724378738e-05,
+      "loss": 0.4665,
+      "step": 1826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9457852084178528e-05,
+      "loss": 0.4352,
+      "step": 1827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945709092489783e-05,
+      "loss": 0.4489,
+      "step": 1828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945632924657842e-05,
+      "loss": 0.44,
+      "step": 1829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9455567049262098e-05,
+      "loss": 0.4435,
+      "step": 1830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.94548043329907e-05,
+      "loss": 0.4196,
+      "step": 1831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9454041097806084e-05,
+      "loss": 0.4702,
+      "step": 1832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9453277343750137e-05,
+      "loss": 0.456,
+      "step": 1833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945251307086478e-05,
+      "loss": 0.4463,
+      "step": 1834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9451748279191956e-05,
+      "loss": 0.4269,
+      "step": 1835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9450982968773643e-05,
+      "loss": 0.4281,
+      "step": 1836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9450217139651845e-05,
+      "loss": 0.4359,
+      "step": 1837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9449450791868585e-05,
+      "loss": 0.4365,
+      "step": 1838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944868392546593e-05,
+      "loss": 0.4307,
+      "step": 1839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944791654048597e-05,
+      "loss": 0.4562,
+      "step": 1840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944714863697082e-05,
+      "loss": 0.4447,
+      "step": 1841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9446380214962616e-05,
+      "loss": 0.4376,
+      "step": 1842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9445611274503547e-05,
+      "loss": 0.4185,
+      "step": 1843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9444841815635804e-05,
+      "loss": 0.4503,
+      "step": 1844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9444071838401624e-05,
+      "loss": 0.4475,
+      "step": 1845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944330134284326e-05,
+      "loss": 0.4835,
+      "step": 1846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9442530329003007e-05,
+      "loss": 0.4186,
+      "step": 1847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9441758796923175e-05,
+      "loss": 0.4351,
+      "step": 1848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944098674664611e-05,
+      "loss": 0.4332,
+      "step": 1849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9440214178214186e-05,
+      "loss": 0.417,
+      "step": 1850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9439441091669803e-05,
+      "loss": 0.4394,
+      "step": 1851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9438667487055394e-05,
+      "loss": 0.4468,
+      "step": 1852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9437893364413408e-05,
+      "loss": 0.4492,
+      "step": 1853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9437118723786346e-05,
+      "loss": 0.4288,
+      "step": 1854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943634356521671e-05,
+      "loss": 0.4552,
+      "step": 1855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9435567888747054e-05,
+      "loss": 0.4225,
+      "step": 1856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943479169441994e-05,
+      "loss": 0.4485,
+      "step": 1857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9434014982277972e-05,
+      "loss": 0.4617,
+      "step": 1858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943323775236378e-05,
+      "loss": 0.4083,
+      "step": 1859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9432460004720025e-05,
+      "loss": 0.4097,
+      "step": 1860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9431681739389385e-05,
+      "loss": 0.4513,
+      "step": 1861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9430902956414573e-05,
+      "loss": 0.4298,
+      "step": 1862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9430123655838345e-05,
+      "loss": 0.4539,
+      "step": 1863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9429343837703457e-05,
+      "loss": 0.4532,
+      "step": 1864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9428563502052712e-05,
+      "loss": 0.4473,
+      "step": 1865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.942778264892894e-05,
+      "loss": 0.4752,
+      "step": 1866
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9427001278374998e-05,
+      "loss": 0.4594,
+      "step": 1867
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9426219390433765e-05,
+      "loss": 0.449,
+      "step": 1868
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942543698514816e-05,
+      "loss": 0.4641,
+      "step": 1869
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942465406256112e-05,
+      "loss": 0.4297,
+      "step": 1870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942387062271562e-05,
+      "loss": 0.4472,
+      "step": 1871
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942308666565465e-05,
+      "loss": 0.4103,
+      "step": 1872
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942230219142124e-05,
+      "loss": 0.457,
+      "step": 1873
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9421517200058445e-05,
+      "loss": 0.4452,
+      "step": 1874
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942073169160935e-05,
+      "loss": 0.4601,
+      "step": 1875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9419945666117064e-05,
+      "loss": 0.4364,
+      "step": 1876
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941915912362473e-05,
+      "loss": 0.4328,
+      "step": 1877
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941837206417551e-05,
+      "loss": 0.4316,
+      "step": 1878
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9417584487812605e-05,
+      "loss": 0.4193,
+      "step": 1879
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941679639457924e-05,
+      "loss": 0.4512,
+      "step": 1880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9416007784518667e-05,
+      "loss": 0.4391,
+      "step": 1881
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941521865767417e-05,
+      "loss": 0.4386,
+      "step": 1882
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941442901408906e-05,
+      "loss": 0.4557,
+      "step": 1883
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9413638853806667e-05,
+      "loss": 0.4477,
+      "step": 1884
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9412848176870365e-05,
+      "loss": 0.4478,
+      "step": 1885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941205698332355e-05,
+      "loss": 0.4661,
+      "step": 1886
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9411265273209643e-05,
+      "loss": 0.4413,
+      "step": 1887
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9410473046572097e-05,
+      "loss": 0.4519,
+      "step": 1888
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9409680303454392e-05,
+      "loss": 0.4638,
+      "step": 1889
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940888704390003e-05,
+      "loss": 0.4529,
+      "step": 1890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940809326795256e-05,
+      "loss": 0.4077,
+      "step": 1891
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9407298975655538e-05,
+      "loss": 0.4297,
+      "step": 1892
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9406504167052565e-05,
+      "loss": 0.428,
+      "step": 1893
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9405708842187256e-05,
+      "loss": 0.4344,
+      "step": 1894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9404913001103264e-05,
+      "loss": 0.418,
+      "step": 1895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9404116643844268e-05,
+      "loss": 0.4393,
+      "step": 1896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9403319770453972e-05,
+      "loss": 0.4514,
+      "step": 1897
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9402522380976117e-05,
+      "loss": 0.4359,
+      "step": 1898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9401724475454464e-05,
+      "loss": 0.4412,
+      "step": 1899
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.94009260539328e-05,
+      "loss": 0.4317,
+      "step": 1900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940012711645495e-05,
+      "loss": 0.4376,
+      "step": 1901
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9399327663064763e-05,
+      "loss": 0.421,
+      "step": 1902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9398527693806115e-05,
+      "loss": 0.445,
+      "step": 1903
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9397727208722914e-05,
+      "loss": 0.4125,
+      "step": 1904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.4584,
+      "step": 1905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.93961246912586e-05,
+      "loss": 0.3969,
+      "step": 1906
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939532265896544e-05,
+      "loss": 0.4466,
+      "step": 1907
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939452011102363e-05,
+      "loss": 0.4294,
+      "step": 1908
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9393717047477213e-05,
+      "loss": 0.4313,
+      "step": 1909
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9392913468370264e-05,
+      "loss": 0.443,
+      "step": 1910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939210937374689e-05,
+      "loss": 0.4327,
+      "step": 1911
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9391304763651216e-05,
+      "loss": 0.4392,
+      "step": 1912
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939049963812741e-05,
+      "loss": 0.4332,
+      "step": 1913
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9389693997219654e-05,
+      "loss": 0.4453,
+      "step": 1914
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9388887840972167e-05,
+      "loss": 0.4417,
+      "step": 1915
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9388081169429195e-05,
+      "loss": 0.4572,
+      "step": 1916
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9387273982635007e-05,
+      "loss": 0.4143,
+      "step": 1917
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9386466280633906e-05,
+      "loss": 0.4476,
+      "step": 1918
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9385658063470223e-05,
+      "loss": 0.4398,
+      "step": 1919
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9384849331188316e-05,
+      "loss": 0.4723,
+      "step": 1920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9384040083832567e-05,
+      "loss": 0.4547,
+      "step": 1921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9383230321447394e-05,
+      "loss": 0.4556,
+      "step": 1922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938242004407724e-05,
+      "loss": 0.4435,
+      "step": 1923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9381609251766577e-05,
+      "loss": 0.4254,
+      "step": 1924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9380797944559898e-05,
+      "loss": 0.4317,
+      "step": 1925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9379986122501736e-05,
+      "loss": 0.4313,
+      "step": 1926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9379173785636647e-05,
+      "loss": 0.4563,
+      "step": 1927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9378360934009212e-05,
+      "loss": 0.4355,
+      "step": 1928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9377547567664043e-05,
+      "loss": 0.4452,
+      "step": 1929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9376733686645782e-05,
+      "loss": 0.4594,
+      "step": 1930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9375919290999098e-05,
+      "loss": 0.4554,
+      "step": 1931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9375104380768687e-05,
+      "loss": 0.3941,
+      "step": 1932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9374288955999273e-05,
+      "loss": 0.4406,
+      "step": 1933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9373473016735612e-05,
+      "loss": 0.4605,
+      "step": 1934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9372656563022483e-05,
+      "loss": 0.4926,
+      "step": 1935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.93718395949047e-05,
+      "loss": 0.4508,
+      "step": 1936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9371022112427095e-05,
+      "loss": 0.4671,
+      "step": 1937
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937020411563454e-05,
+      "loss": 0.4469,
+      "step": 1938
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9369385604571928e-05,
+      "loss": 0.4704,
+      "step": 1939
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9368566579284182e-05,
+      "loss": 0.4145,
+      "step": 1940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936774703981625e-05,
+      "loss": 0.4297,
+      "step": 1941
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9366926986213115e-05,
+      "loss": 0.4461,
+      "step": 1942
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9366106418519783e-05,
+      "loss": 0.4451,
+      "step": 1943
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9365285336781284e-05,
+      "loss": 0.4275,
+      "step": 1944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9364463741042694e-05,
+      "loss": 0.4614,
+      "step": 1945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9363641631349096e-05,
+      "loss": 0.4283,
+      "step": 1946
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9362819007745614e-05,
+      "loss": 0.429,
+      "step": 1947
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9361995870277393e-05,
+      "loss": 0.4463,
+      "step": 1948
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936117221898961e-05,
+      "loss": 0.4474,
+      "step": 1949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9360348053927477e-05,
+      "loss": 0.4466,
+      "step": 1950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935952337513622e-05,
+      "loss": 0.4387,
+      "step": 1951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9358698182661103e-05,
+      "loss": 0.4662,
+      "step": 1952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9357872476547414e-05,
+      "loss": 0.4426,
+      "step": 1953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935704625684047e-05,
+      "loss": 0.4381,
+      "step": 1954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935621952358562e-05,
+      "loss": 0.4363,
+      "step": 1955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935539227682824e-05,
+      "loss": 0.4247,
+      "step": 1956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9354564516613723e-05,
+      "loss": 0.4479,
+      "step": 1957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9353736242987504e-05,
+      "loss": 0.4367,
+      "step": 1958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9352907455995045e-05,
+      "loss": 0.4197,
+      "step": 1959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9352078155681834e-05,
+      "loss": 0.4328,
+      "step": 1960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935124834209338e-05,
+      "loss": 0.4408,
+      "step": 1961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935041801527523e-05,
+      "loss": 0.43,
+      "step": 1962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934958717527295e-05,
+      "loss": 0.4558,
+      "step": 1963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934875582213214e-05,
+      "loss": 0.4342,
+      "step": 1964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9347923955898437e-05,
+      "loss": 0.4337,
+      "step": 1965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934709157661749e-05,
+      "loss": 0.4322,
+      "step": 1966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934625868433498e-05,
+      "loss": 0.4556,
+      "step": 1967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934542527909663e-05,
+      "loss": 0.4281,
+      "step": 1968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9344591360948166e-05,
+      "loss": 0.4372,
+      "step": 1969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9343756929935364e-05,
+      "loss": 0.4641,
+      "step": 1970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9342921986104022e-05,
+      "loss": 0.482,
+      "step": 1971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934208652949996e-05,
+      "loss": 0.4258,
+      "step": 1972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9341250560169035e-05,
+      "loss": 0.4525,
+      "step": 1973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9340414078157126e-05,
+      "loss": 0.4488,
+      "step": 1974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9339577083510145e-05,
+      "loss": 0.4584,
+      "step": 1975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9338739576274022e-05,
+      "loss": 0.4467,
+      "step": 1976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9337901556494727e-05,
+      "loss": 0.4192,
+      "step": 1977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9337063024218258e-05,
+      "loss": 0.4237,
+      "step": 1978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933622397949063e-05,
+      "loss": 0.4484,
+      "step": 1979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9335384422357893e-05,
+      "loss": 0.429,
+      "step": 1980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9334544352866126e-05,
+      "loss": 0.4325,
+      "step": 1981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933370377106144e-05,
+      "loss": 0.4349,
+      "step": 1982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933286267698996e-05,
+      "loss": 0.4473,
+      "step": 1983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9332021070697854e-05,
+      "loss": 0.4435,
+      "step": 1984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9331178952231314e-05,
+      "loss": 0.4482,
+      "step": 1985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9330336321636552e-05,
+      "loss": 0.4369,
+      "step": 1986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9329493178959823e-05,
+      "loss": 0.4303,
+      "step": 1987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9328649524247398e-05,
+      "loss": 0.443,
+      "step": 1988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9327805357545572e-05,
+      "loss": 0.4502,
+      "step": 1989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932696067890069e-05,
+      "loss": 0.4552,
+      "step": 1990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9326115488359098e-05,
+      "loss": 0.4265,
+      "step": 1991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932526978596719e-05,
+      "loss": 0.43,
+      "step": 1992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9324423571771377e-05,
+      "loss": 0.4312,
+      "step": 1993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932357684581811e-05,
+      "loss": 0.4732,
+      "step": 1994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9322729608153854e-05,
+      "loss": 0.4601,
+      "step": 1995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932188185882511e-05,
+      "loss": 0.4657,
+      "step": 1996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9321033597878406e-05,
+      "loss": 0.4724,
+      "step": 1997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9320184825360296e-05,
+      "loss": 0.4463,
+      "step": 1998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9319335541317363e-05,
+      "loss": 0.4279,
+      "step": 1999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9318485745796224e-05,
+      "loss": 0.4505,
+      "step": 2000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9317635438843513e-05,
+      "loss": 0.4293,
+      "step": 2001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.93167846205059e-05,
+      "loss": 0.4461,
+      "step": 2002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9315933290830076e-05,
+      "loss": 0.4389,
+      "step": 2003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9315081449862775e-05,
+      "loss": 0.4448,
+      "step": 2004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9314229097650744e-05,
+      "loss": 0.4294,
+      "step": 2005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9313376234240763e-05,
+      "loss": 0.4932,
+      "step": 2006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931252285967964e-05,
+      "loss": 0.439,
+      "step": 2007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9311668974014207e-05,
+      "loss": 0.4322,
+      "step": 2008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9310814577291337e-05,
+      "loss": 0.4386,
+      "step": 2009
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930995966955792e-05,
+      "loss": 0.4629,
+      "step": 2010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9309104250860868e-05,
+      "loss": 0.4338,
+      "step": 2011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930824832124714e-05,
+      "loss": 0.4304,
+      "step": 2012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9307391880763706e-05,
+      "loss": 0.4569,
+      "step": 2013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9306534929457576e-05,
+      "loss": 0.4265,
+      "step": 2014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9305677467375776e-05,
+      "loss": 0.4283,
+      "step": 2015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930481949456537e-05,
+      "loss": 0.4154,
+      "step": 2016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9303961011073448e-05,
+      "loss": 0.4526,
+      "step": 2017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9303102016947127e-05,
+      "loss": 0.4578,
+      "step": 2018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9302242512233547e-05,
+      "loss": 0.4462,
+      "step": 2019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9301382496979884e-05,
+      "loss": 0.4315,
+      "step": 2020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9300521971233338e-05,
+      "loss": 0.4527,
+      "step": 2021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929966093504114e-05,
+      "loss": 0.4377,
+      "step": 2022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929879938845054e-05,
+      "loss": 0.4543,
+      "step": 2023
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929793733150883e-05,
+      "loss": 0.4825,
+      "step": 2024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9297074764263318e-05,
+      "loss": 0.4651,
+      "step": 2025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9296211686761347e-05,
+      "loss": 0.4645,
+      "step": 2026
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9295348099050287e-05,
+      "loss": 0.4475,
+      "step": 2027
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929448400117753e-05,
+      "loss": 0.4323,
+      "step": 2028
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929361939319051e-05,
+      "loss": 0.4501,
+      "step": 2029
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929275427513667e-05,
+      "loss": 0.4196,
+      "step": 2030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9291888647063493e-05,
+      "loss": 0.4087,
+      "step": 2031
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9291022509018492e-05,
+      "loss": 0.4166,
+      "step": 2032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.92901558610492e-05,
+      "loss": 0.4245,
+      "step": 2033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928928870320318e-05,
+      "loss": 0.4335,
+      "step": 2034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928842103552803e-05,
+      "loss": 0.43,
+      "step": 2035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928755285807137e-05,
+      "loss": 0.464,
+      "step": 2036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9286684170880843e-05,
+      "loss": 0.4427,
+      "step": 2037
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928581497400413e-05,
+      "loss": 0.4431,
+      "step": 2038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9284945267488938e-05,
+      "loss": 0.4384,
+      "step": 2039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9284075051382993e-05,
+      "loss": 0.4419,
+      "step": 2040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9283204325734058e-05,
+      "loss": 0.411,
+      "step": 2041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9282333090589923e-05,
+      "loss": 0.4348,
+      "step": 2042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928146134599841e-05,
+      "loss": 0.4338,
+      "step": 2043
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9280589092007354e-05,
+      "loss": 0.425,
+      "step": 2044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9279716328664628e-05,
+      "loss": 0.4289,
+      "step": 2045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9278843056018138e-05,
+      "loss": 0.4188,
+      "step": 2046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927796927411581e-05,
+      "loss": 0.4459,
+      "step": 2047
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9277094983005597e-05,
+      "loss": 0.4287,
+      "step": 2048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927622018273549e-05,
+      "loss": 0.4009,
+      "step": 2049
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9275344873353492e-05,
+      "loss": 0.4502,
+      "step": 2050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9274469054907656e-05,
+      "loss": 0.4429,
+      "step": 2051
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9273592727446037e-05,
+      "loss": 0.4596,
+      "step": 2052
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9272715891016736e-05,
+      "loss": 0.4224,
+      "step": 2053
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9271838545667876e-05,
+      "loss": 0.4491,
+      "step": 2054
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927096069144761e-05,
+      "loss": 0.439,
+      "step": 2055
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927008232840412e-05,
+      "loss": 0.4851,
+      "step": 2056
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.926920345658561e-05,
+      "loss": 0.4622,
+      "step": 2057
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9268324076040318e-05,
+      "loss": 0.4341,
+      "step": 2058
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9267444186816506e-05,
+      "loss": 0.4549,
+      "step": 2059
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926656378896247e-05,
+      "loss": 0.4566,
+      "step": 2060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9265682882526516e-05,
+      "loss": 0.4048,
+      "step": 2061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9264801467557008e-05,
+      "loss": 0.4333,
+      "step": 2062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9263919544102312e-05,
+      "loss": 0.4176,
+      "step": 2063
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9263037112210835e-05,
+      "loss": 0.4032,
+      "step": 2064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9262154171931003e-05,
+      "loss": 0.4514,
+      "step": 2065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9261270723311276e-05,
+      "loss": 0.4404,
+      "step": 2066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9260386766400145e-05,
+      "loss": 0.44,
+      "step": 2067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9259502301246122e-05,
+      "loss": 0.4546,
+      "step": 2068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9258617327897752e-05,
+      "loss": 0.458,
+      "step": 2069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.92577318464036e-05,
+      "loss": 0.472,
+      "step": 2070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9256845856812266e-05,
+      "loss": 0.4283,
+      "step": 2071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9255959359172383e-05,
+      "loss": 0.4299,
+      "step": 2072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9255072353532597e-05,
+      "loss": 0.4301,
+      "step": 2073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9254184839941593e-05,
+      "loss": 0.4185,
+      "step": 2074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9253296818448082e-05,
+      "loss": 0.3968,
+      "step": 2075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.92524082891008e-05,
+      "loss": 0.4735,
+      "step": 2076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9251519251948512e-05,
+      "loss": 0.4809,
+      "step": 2077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925062970704002e-05,
+      "loss": 0.4895,
+      "step": 2078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924973965442413e-05,
+      "loss": 0.4381,
+      "step": 2079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9248849094149703e-05,
+      "loss": 0.4475,
+      "step": 2080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9247958026265612e-05,
+      "loss": 0.4521,
+      "step": 2081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924706645082076e-05,
+      "loss": 0.4376,
+      "step": 2082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9246174367864086e-05,
+      "loss": 0.4392,
+      "step": 2083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9245281777444547e-05,
+      "loss": 0.446,
+      "step": 2084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924438867961113e-05,
+      "loss": 0.4793,
+      "step": 2085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9243495074412853e-05,
+      "loss": 0.4383,
+      "step": 2086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924260096189876e-05,
+      "loss": 0.4641,
+      "step": 2087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9241706342117928e-05,
+      "loss": 0.4543,
+      "step": 2088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9240811215119447e-05,
+      "loss": 0.446,
+      "step": 2089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923991558095245e-05,
+      "loss": 0.4354,
+      "step": 2090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9239019439666097e-05,
+      "loss": 0.4734,
+      "step": 2091
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923812279130956e-05,
+      "loss": 0.4217,
+      "step": 2092
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9237225635932063e-05,
+      "loss": 0.4141,
+      "step": 2093
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923632797358284e-05,
+      "loss": 0.4389,
+      "step": 2094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923542980431115e-05,
+      "loss": 0.4322,
+      "step": 2095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9234531128166302e-05,
+      "loss": 0.455,
+      "step": 2096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9233631945197608e-05,
+      "loss": 0.4628,
+      "step": 2097
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9232732255454424e-05,
+      "loss": 0.4437,
+      "step": 2098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9231832058986123e-05,
+      "loss": 0.4832,
+      "step": 2099
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9230931355842116e-05,
+      "loss": 0.4089,
+      "step": 2100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923003014607183e-05,
+      "loss": 0.4553,
+      "step": 2101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922912842972474e-05,
+      "loss": 0.4437,
+      "step": 2102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922822620685032e-05,
+      "loss": 0.4735,
+      "step": 2103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9227323477498098e-05,
+      "loss": 0.4392,
+      "step": 2104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922642024171761e-05,
+      "loss": 0.4279,
+      "step": 2105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9225516499558438e-05,
+      "loss": 0.446,
+      "step": 2106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9224612251070176e-05,
+      "loss": 0.4452,
+      "step": 2107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9223707496302455e-05,
+      "loss": 0.4198,
+      "step": 2108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922280223530493e-05,
+      "loss": 0.4299,
+      "step": 2109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922189646812729e-05,
+      "loss": 0.4378,
+      "step": 2110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9220990194819235e-05,
+      "loss": 0.4359,
+      "step": 2111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9220083415430517e-05,
+      "loss": 0.4477,
+      "step": 2112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9219176130010896e-05,
+      "loss": 0.4347,
+      "step": 2113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.921826833861017e-05,
+      "loss": 0.4335,
+      "step": 2114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9217360041278162e-05,
+      "loss": 0.4441,
+      "step": 2115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9216451238064723e-05,
+      "loss": 0.4302,
+      "step": 2116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9215541929019727e-05,
+      "loss": 0.4341,
+      "step": 2117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9214632114193084e-05,
+      "loss": 0.4629,
+      "step": 2118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9213721793634726e-05,
+      "loss": 0.4299,
+      "step": 2119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921281096739462e-05,
+      "loss": 0.4192,
+      "step": 2120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9211899635522744e-05,
+      "loss": 0.4191,
+      "step": 2121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9210987798069124e-05,
+      "loss": 0.4696,
+      "step": 2122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9210075455083805e-05,
+      "loss": 0.4408,
+      "step": 2123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9209162606616857e-05,
+      "loss": 0.4375,
+      "step": 2124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920824925271838e-05,
+      "loss": 0.4071,
+      "step": 2125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9207335393438504e-05,
+      "loss": 0.4467,
+      "step": 2126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9206421028827384e-05,
+      "loss": 0.4403,
+      "step": 2127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9205506158935203e-05,
+      "loss": 0.4532,
+      "step": 2128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9204590783812175e-05,
+      "loss": 0.4346,
+      "step": 2129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9203674903508533e-05,
+      "loss": 0.4359,
+      "step": 2130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9202758518074546e-05,
+      "loss": 0.4286,
+      "step": 2131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9201841627560516e-05,
+      "loss": 0.4433,
+      "step": 2132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9200924232016754e-05,
+      "loss": 0.4206,
+      "step": 2133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920000633149362e-05,
+      "loss": 0.467,
+      "step": 2134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9199087926041482e-05,
+      "loss": 0.4657,
+      "step": 2135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9198169015710754e-05,
+      "loss": 0.4556,
+      "step": 2136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9197249600551862e-05,
+      "loss": 0.4361,
+      "step": 2137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919632968061527e-05,
+      "loss": 0.4176,
+      "step": 2138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919540925595147e-05,
+      "loss": 0.428,
+      "step": 2139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9194488326610972e-05,
+      "loss": 0.4439,
+      "step": 2140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9193566892644325e-05,
+      "loss": 0.4305,
+      "step": 2141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9192644954102096e-05,
+      "loss": 0.4187,
+      "step": 2142
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9191722511034884e-05,
+      "loss": 0.4609,
+      "step": 2143
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.919079956349332e-05,
+      "loss": 0.4846,
+      "step": 2144
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9189876111528057e-05,
+      "loss": 0.4399,
+      "step": 2145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918895215518978e-05,
+      "loss": 0.4925,
+      "step": 2146
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9188027694529192e-05,
+      "loss": 0.4423,
+      "step": 2147
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9187102729597037e-05,
+      "loss": 0.4567,
+      "step": 2148
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9186177260444078e-05,
+      "loss": 0.4927,
+      "step": 2149
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918525128712111e-05,
+      "loss": 0.4416,
+      "step": 2150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918432480967895e-05,
+      "loss": 0.4201,
+      "step": 2151
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9183397828168448e-05,
+      "loss": 0.428,
+      "step": 2152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9182470342640483e-05,
+      "loss": 0.4649,
+      "step": 2153
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9181542353145953e-05,
+      "loss": 0.4694,
+      "step": 2154
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918061385973579e-05,
+      "loss": 0.4483,
+      "step": 2155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917968486246096e-05,
+      "loss": 0.4626,
+      "step": 2156
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9178755361372445e-05,
+      "loss": 0.4455,
+      "step": 2157
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917782535652126e-05,
+      "loss": 0.454,
+      "step": 2158
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9176894847958446e-05,
+      "loss": 0.4495,
+      "step": 2159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917596383573507e-05,
+      "loss": 0.4417,
+      "step": 2160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9175032319902236e-05,
+      "loss": 0.4357,
+      "step": 2161
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9174100300511064e-05,
+      "loss": 0.4346,
+      "step": 2162
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9173167777612706e-05,
+      "loss": 0.4412,
+      "step": 2163
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9172234751258345e-05,
+      "loss": 0.4721,
+      "step": 2164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9171301221499187e-05,
+      "loss": 0.4634,
+      "step": 2165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917036718838647e-05,
+      "loss": 0.4868,
+      "step": 2166
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9169432651971455e-05,
+      "loss": 0.4306,
+      "step": 2167
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.916849761230543e-05,
+      "loss": 0.4473,
+      "step": 2168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.916756206943972e-05,
+      "loss": 0.4377,
+      "step": 2169
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9166626023425663e-05,
+      "loss": 0.4075,
+      "step": 2170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916568947431464e-05,
+      "loss": 0.4357,
+      "step": 2171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9164752422158045e-05,
+      "loss": 0.4657,
+      "step": 2172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916381486700731e-05,
+      "loss": 0.4306,
+      "step": 2173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9162876808913897e-05,
+      "loss": 0.4271,
+      "step": 2174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9161938247929282e-05,
+      "loss": 0.4385,
+      "step": 2175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9160999184104982e-05,
+      "loss": 0.4135,
+      "step": 2176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916005961749253e-05,
+      "loss": 0.4388,
+      "step": 2177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9159119548143497e-05,
+      "loss": 0.4519,
+      "step": 2178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9158178976109475e-05,
+      "loss": 0.4292,
+      "step": 2179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9157237901442093e-05,
+      "loss": 0.4205,
+      "step": 2180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9156296324192988e-05,
+      "loss": 0.4268,
+      "step": 2181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9155354244413848e-05,
+      "loss": 0.4511,
+      "step": 2182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9154411662156373e-05,
+      "loss": 0.4275,
+      "step": 2183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9153468577472296e-05,
+      "loss": 0.4341,
+      "step": 2184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9152524990413378e-05,
+      "loss": 0.452,
+      "step": 2185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9151580901031406e-05,
+      "loss": 0.4335,
+      "step": 2186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915063630937819e-05,
+      "loss": 0.4358,
+      "step": 2187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9149691215505578e-05,
+      "loss": 0.4314,
+      "step": 2188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9148745619465442e-05,
+      "loss": 0.4461,
+      "step": 2189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9147799521309677e-05,
+      "loss": 0.4309,
+      "step": 2190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9146852921090207e-05,
+      "loss": 0.4402,
+      "step": 2191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9145905818858984e-05,
+      "loss": 0.4447,
+      "step": 2192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9144958214667987e-05,
+      "loss": 0.4506,
+      "step": 2193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9144010108569233e-05,
+      "loss": 0.4651,
+      "step": 2194
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9143061500614747e-05,
+      "loss": 0.4274,
+      "step": 2195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9142112390856598e-05,
+      "loss": 0.4726,
+      "step": 2196
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9141162779346875e-05,
+      "loss": 0.4622,
+      "step": 2197
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9140212666137694e-05,
+      "loss": 0.4336,
+      "step": 2198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9139262051281205e-05,
+      "loss": 0.4437,
+      "step": 2199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9138310934829576e-05,
+      "loss": 0.4266,
+      "step": 2200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913735931683501e-05,
+      "loss": 0.4469,
+      "step": 2201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9136407197349737e-05,
+      "loss": 0.4004,
+      "step": 2202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 0.4585,
+      "step": 2203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9134501454116115e-05,
+      "loss": 0.4297,
+      "step": 2204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913354783047236e-05,
+      "loss": 0.4592,
+      "step": 2205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9132593705547082e-05,
+      "loss": 0.4526,
+      "step": 2206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913163907939265e-05,
+      "loss": 0.4795,
+      "step": 2207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9130683952061458e-05,
+      "loss": 0.4486,
+      "step": 2208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9129728323605926e-05,
+      "loss": 0.4485,
+      "step": 2209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9128772194078497e-05,
+      "loss": 0.4241,
+      "step": 2210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9127815563531654e-05,
+      "loss": 0.427,
+      "step": 2211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9126858432017896e-05,
+      "loss": 0.4265,
+      "step": 2212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9125900799589754e-05,
+      "loss": 0.4273,
+      "step": 2213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9124942666299788e-05,
+      "loss": 0.4542,
+      "step": 2214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9123984032200586e-05,
+      "loss": 0.4255,
+      "step": 2215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9123024897344758e-05,
+      "loss": 0.4433,
+      "step": 2216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912206526178494e-05,
+      "loss": 0.4294,
+      "step": 2217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9121105125573813e-05,
+      "loss": 0.4229,
+      "step": 2218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912014448876406e-05,
+      "loss": 0.43,
+      "step": 2219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911918335140841e-05,
+      "loss": 0.4381,
+      "step": 2220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911822171355961e-05,
+      "loss": 0.4191,
+      "step": 2221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9117259575270444e-05,
+      "loss": 0.4329,
+      "step": 2222
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9116296936593716e-05,
+      "loss": 0.4564,
+      "step": 2223
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9115333797582252e-05,
+      "loss": 0.4344,
+      "step": 2224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9114370158288925e-05,
+      "loss": 0.4454,
+      "step": 2225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911340601876661e-05,
+      "loss": 0.4198,
+      "step": 2226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9112441379068232e-05,
+      "loss": 0.431,
+      "step": 2227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9111476239246732e-05,
+      "loss": 0.4446,
+      "step": 2228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9110510599355076e-05,
+      "loss": 0.4207,
+      "step": 2229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910954445944626e-05,
+      "loss": 0.4226,
+      "step": 2230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9108577819573317e-05,
+      "loss": 0.469,
+      "step": 2231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9107610679789298e-05,
+      "loss": 0.453,
+      "step": 2232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910664304014728e-05,
+      "loss": 0.4781,
+      "step": 2233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910567490070037e-05,
+      "loss": 0.4547,
+      "step": 2234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9104706261501704e-05,
+      "loss": 0.4304,
+      "step": 2235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9103737122604444e-05,
+      "loss": 0.4217,
+      "step": 2236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9102767484061785e-05,
+      "loss": 0.4072,
+      "step": 2237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9101797345926937e-05,
+      "loss": 0.4388,
+      "step": 2238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9100826708253148e-05,
+      "loss": 0.435,
+      "step": 2239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909985557109369e-05,
+      "loss": 0.4603,
+      "step": 2240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909888393450186e-05,
+      "loss": 0.4361,
+      "step": 2241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909791179853099e-05,
+      "loss": 0.449,
+      "step": 2242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9096939163234426e-05,
+      "loss": 0.4503,
+      "step": 2243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909596602866556e-05,
+      "loss": 0.4417,
+      "step": 2244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9094992394877792e-05,
+      "loss": 0.4812,
+      "step": 2245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9094018261924566e-05,
+      "loss": 0.4227,
+      "step": 2246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909304362985934e-05,
+      "loss": 0.4646,
+      "step": 2247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9092068498735607e-05,
+      "loss": 0.4555,
+      "step": 2248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909109286860689e-05,
+      "loss": 0.4329,
+      "step": 2249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9090116739526725e-05,
+      "loss": 0.4053,
+      "step": 2250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9089140111548695e-05,
+      "loss": 0.437,
+      "step": 2251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.90881629847264e-05,
+      "loss": 0.4469,
+      "step": 2252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908718535911346e-05,
+      "loss": 0.4243,
+      "step": 2253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908620723476354e-05,
+      "loss": 0.4491,
+      "step": 2254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9085228611730316e-05,
+      "loss": 0.4321,
+      "step": 2255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9084249490067507e-05,
+      "loss": 0.4391,
+      "step": 2256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908326986982884e-05,
+      "loss": 0.4362,
+      "step": 2257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9082289751068087e-05,
+      "loss": 0.4375,
+      "step": 2258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9081309133839037e-05,
+      "loss": 0.4405,
+      "step": 2259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9080328018195512e-05,
+      "loss": 0.4405,
+      "step": 2260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907934640419136e-05,
+      "loss": 0.4649,
+      "step": 2261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9078364291880454e-05,
+      "loss": 0.4201,
+      "step": 2262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9077381681316693e-05,
+      "loss": 0.4532,
+      "step": 2263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907639857255401e-05,
+      "loss": 0.481,
+      "step": 2264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907541496564636e-05,
+      "loss": 0.4378,
+      "step": 2265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9074430860647728e-05,
+      "loss": 0.425,
+      "step": 2266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9073446257612123e-05,
+      "loss": 0.4313,
+      "step": 2267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9072461156593584e-05,
+      "loss": 0.4361,
+      "step": 2268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907147555764618e-05,
+      "loss": 0.4376,
+      "step": 2269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9070489460824002e-05,
+      "loss": 0.4623,
+      "step": 2270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906950286618117e-05,
+      "loss": 0.4282,
+      "step": 2271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906851577377183e-05,
+      "loss": 0.439,
+      "step": 2272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906752818365016e-05,
+      "loss": 0.4556,
+      "step": 2273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9066540095870364e-05,
+      "loss": 0.45,
+      "step": 2274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9065551510486668e-05,
+      "loss": 0.4394,
+      "step": 2275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906456242755333e-05,
+      "loss": 0.4436,
+      "step": 2276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9063572847124635e-05,
+      "loss": 0.4564,
+      "step": 2277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9062582769254896e-05,
+      "loss": 0.4088,
+      "step": 2278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9061592193998448e-05,
+      "loss": 0.457,
+      "step": 2279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9060601121409662e-05,
+      "loss": 0.4624,
+      "step": 2280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.905960955154293e-05,
+      "loss": 0.4398,
+      "step": 2281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905861748445267e-05,
+      "loss": 0.4319,
+      "step": 2282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9057624920193335e-05,
+      "loss": 0.4396,
+      "step": 2283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.90566318588194e-05,
+      "loss": 0.4488,
+      "step": 2284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9055638300385362e-05,
+      "loss": 0.4287,
+      "step": 2285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9054644244945756e-05,
+      "loss": 0.4378,
+      "step": 2286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9053649692555135e-05,
+      "loss": 0.4174,
+      "step": 2287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9052654643268088e-05,
+      "loss": 0.4349,
+      "step": 2288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905165909713923e-05,
+      "loss": 0.4549,
+      "step": 2289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905066305422319e-05,
+      "loss": 0.425,
+      "step": 2290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9049666514574642e-05,
+      "loss": 0.4332,
+      "step": 2291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9048669478248275e-05,
+      "loss": 0.4246,
+      "step": 2292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9047671945298813e-05,
+      "loss": 0.4427,
+      "step": 2293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9046673915781005e-05,
+      "loss": 0.4361,
+      "step": 2294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904567538974963e-05,
+      "loss": 0.4419,
+      "step": 2295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9044676367259474e-05,
+      "loss": 0.4461,
+      "step": 2296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9043676848365387e-05,
+      "loss": 0.4422,
+      "step": 2297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9042676833122214e-05,
+      "loss": 0.4686,
+      "step": 2298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904167632158485e-05,
+      "loss": 0.4617,
+      "step": 2299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904067531380819e-05,
+      "loss": 0.4384,
+      "step": 2300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9039673809847187e-05,
+      "loss": 0.431,
+      "step": 2301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9038671809756805e-05,
+      "loss": 0.4272,
+      "step": 2302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9037669313592032e-05,
+      "loss": 0.4149,
+      "step": 2303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903666632140789e-05,
+      "loss": 0.4494,
+      "step": 2304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9035662833259433e-05,
+      "loss": 0.4531,
+      "step": 2305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9034658849201725e-05,
+      "loss": 0.4368,
+      "step": 2306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9033654369289883e-05,
+      "loss": 0.4276,
+      "step": 2307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9032649393579018e-05,
+      "loss": 0.4434,
+      "step": 2308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9031643922124302e-05,
+      "loss": 0.4534,
+      "step": 2309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9030637954980915e-05,
+      "loss": 0.448,
+      "step": 2310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9029631492204062e-05,
+      "loss": 0.4301,
+      "step": 2311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9028624533848988e-05,
+      "loss": 0.4534,
+      "step": 2312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9027617079970956e-05,
+      "loss": 0.4388,
+      "step": 2313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9026609130625257e-05,
+      "loss": 0.4304,
+      "step": 2314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9025600685867215e-05,
+      "loss": 0.4655,
+      "step": 2315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9024591745752174e-05,
+      "loss": 0.4398,
+      "step": 2316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902358231033551e-05,
+      "loss": 0.4556,
+      "step": 2317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9022572379672622e-05,
+      "loss": 0.4395,
+      "step": 2318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9021561953818937e-05,
+      "loss": 0.4319,
+      "step": 2319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9020551032829918e-05,
+      "loss": 0.4403,
+      "step": 2320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901953961676104e-05,
+      "loss": 0.4318,
+      "step": 2321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901852770566782e-05,
+      "loss": 0.4244,
+      "step": 2322
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901751529960579e-05,
+      "loss": 0.4153,
+      "step": 2323
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9016502398630516e-05,
+      "loss": 0.423,
+      "step": 2324
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9015489002797588e-05,
+      "loss": 0.432,
+      "step": 2325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901447511216263e-05,
+      "loss": 0.4463,
+      "step": 2326
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9013460726781283e-05,
+      "loss": 0.4601,
+      "step": 2327
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9012445846709222e-05,
+      "loss": 0.4604,
+      "step": 2328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9011430472002148e-05,
+      "loss": 0.3947,
+      "step": 2329
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9010414602715786e-05,
+      "loss": 0.4591,
+      "step": 2330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900939823890589e-05,
+      "loss": 0.409,
+      "step": 2331
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900838138062825e-05,
+      "loss": 0.4422,
+      "step": 2332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900736402793866e-05,
+      "loss": 0.4225,
+      "step": 2333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9006346180892967e-05,
+      "loss": 0.4827,
+      "step": 2334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9005327839547032e-05,
+      "loss": 0.4557,
+      "step": 2335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9004309003956743e-05,
+      "loss": 0.4415,
+      "step": 2336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900328967417802e-05,
+      "loss": 0.4541,
+      "step": 2337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9002269850266805e-05,
+      "loss": 0.4362,
+      "step": 2338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9001249532279073e-05,
+      "loss": 0.4481,
+      "step": 2339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9000228720270816e-05,
+      "loss": 0.4647,
+      "step": 2340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899920741429807e-05,
+      "loss": 0.4406,
+      "step": 2341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8998185614416876e-05,
+      "loss": 0.4543,
+      "step": 2342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8997163320683323e-05,
+      "loss": 0.4297,
+      "step": 2343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8996140533153517e-05,
+      "loss": 0.4267,
+      "step": 2344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8995117251883587e-05,
+      "loss": 0.4331,
+      "step": 2345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.89940934769297e-05,
+      "loss": 0.4562,
+      "step": 2346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899306920834804e-05,
+      "loss": 0.4503,
+      "step": 2347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8992044446194826e-05,
+      "loss": 0.4449,
+      "step": 2348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.89910191905263e-05,
+      "loss": 0.4201,
+      "step": 2349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8989993441398725e-05,
+      "loss": 0.4254,
+      "step": 2350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898896719886841e-05,
+      "loss": 0.4363,
+      "step": 2351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8987940462991673e-05,
+      "loss": 0.416,
+      "step": 2352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8986913233824863e-05,
+      "loss": 0.4324,
+      "step": 2353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8985885511424354e-05,
+      "loss": 0.4292,
+      "step": 2354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8984857295846563e-05,
+      "loss": 0.4504,
+      "step": 2355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8983828587147914e-05,
+      "loss": 0.4445,
+      "step": 2356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898279938538486e-05,
+      "loss": 0.4429,
+      "step": 2357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8981769690613904e-05,
+      "loss": 0.4269,
+      "step": 2358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8980739502891548e-05,
+      "loss": 0.442,
+      "step": 2359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8979708822274333e-05,
+      "loss": 0.4535,
+      "step": 2360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897867764881883e-05,
+      "loss": 0.4589,
+      "step": 2361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8977645982581633e-05,
+      "loss": 0.4211,
+      "step": 2362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8976613823619357e-05,
+      "loss": 0.4211,
+      "step": 2363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8975581171988657e-05,
+      "loss": 0.4305,
+      "step": 2364
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8974548027746208e-05,
+      "loss": 0.4473,
+      "step": 2365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8973514390948713e-05,
+      "loss": 0.4602,
+      "step": 2366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8972480261652896e-05,
+      "loss": 0.4552,
+      "step": 2367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8971445639915522e-05,
+      "loss": 0.4078,
+      "step": 2368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897041052579337e-05,
+      "loss": 0.4272,
+      "step": 2369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896937491934325e-05,
+      "loss": 0.4471,
+      "step": 2370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8968338820622002e-05,
+      "loss": 0.4416,
+      "step": 2371
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896730222968649e-05,
+      "loss": 0.4355,
+      "step": 2372
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8966265146593606e-05,
+      "loss": 0.4362,
+      "step": 2373
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8965227571400267e-05,
+      "loss": 0.4308,
+      "step": 2374
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8964189504163423e-05,
+      "loss": 0.4619,
+      "step": 2375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8963150944940044e-05,
+      "loss": 0.4388,
+      "step": 2376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896211189378713e-05,
+      "loss": 0.4301,
+      "step": 2377
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8961072350761707e-05,
+      "loss": 0.4595,
+      "step": 2378
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896003231592083e-05,
+      "loss": 0.4269,
+      "step": 2379
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895899178932158e-05,
+      "loss": 0.4434,
+      "step": 2380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895795077102106e-05,
+      "loss": 0.4524,
+      "step": 2381
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8956909261076414e-05,
+      "loss": 0.433,
+      "step": 2382
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8955867259544795e-05,
+      "loss": 0.4178,
+      "step": 2383
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.89548247664834e-05,
+      "loss": 0.4472,
+      "step": 2384
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895378178194944e-05,
+      "loss": 0.4272,
+      "step": 2385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8952738306000152e-05,
+      "loss": 0.4312,
+      "step": 2386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8951694338692816e-05,
+      "loss": 0.4483,
+      "step": 2387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8950649880084724e-05,
+      "loss": 0.4539,
+      "step": 2388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.89496049302332e-05,
+      "loss": 0.4277,
+      "step": 2389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8948559489195593e-05,
+      "loss": 0.4404,
+      "step": 2390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8947513557029283e-05,
+      "loss": 0.403,
+      "step": 2391
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894646713379167e-05,
+      "loss": 0.4411,
+      "step": 2392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8945420219540197e-05,
+      "loss": 0.4607,
+      "step": 2393
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8944372814332306e-05,
+      "loss": 0.4589,
+      "step": 2394
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8943324918225495e-05,
+      "loss": 0.4489,
+      "step": 2395
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894227653127727e-05,
+      "loss": 0.4278,
+      "step": 2396
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8941227653545172e-05,
+      "loss": 0.4207,
+      "step": 2397
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894017828508677e-05,
+      "loss": 0.4181,
+      "step": 2398
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8939128425959648e-05,
+      "loss": 0.4281,
+      "step": 2399
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8938078076221436e-05,
+      "loss": 0.4241,
+      "step": 2400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8937027235929778e-05,
+      "loss": 0.4617,
+      "step": 2401
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8935975905142343e-05,
+      "loss": 0.395,
+      "step": 2402
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8934924083916838e-05,
+      "loss": 0.4144,
+      "step": 2403
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893387177231099e-05,
+      "loss": 0.4329,
+      "step": 2404
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893281897038255e-05,
+      "loss": 0.4591,
+      "step": 2405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.89317656781893e-05,
+      "loss": 0.4543,
+      "step": 2406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8930711895789053e-05,
+      "loss": 0.4064,
+      "step": 2407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8929657623239638e-05,
+      "loss": 0.4516,
+      "step": 2408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8928602860598924e-05,
+      "loss": 0.4222,
+      "step": 2409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8927547607924796e-05,
+      "loss": 0.413,
+      "step": 2410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892649186527517e-05,
+      "loss": 0.4194,
+      "step": 2411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8925435632707986e-05,
+      "loss": 0.4398,
+      "step": 2412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892437891028122e-05,
+      "loss": 0.4468,
+      "step": 2413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8923321698052865e-05,
+      "loss": 0.4242,
+      "step": 2414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8922263996080946e-05,
+      "loss": 0.4382,
+      "step": 2415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8921205804423515e-05,
+      "loss": 0.4162,
+      "step": 2416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8920147123138645e-05,
+      "loss": 0.4246,
+      "step": 2417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8919087952284442e-05,
+      "loss": 0.4475,
+      "step": 2418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.891802829191904e-05,
+      "loss": 0.4439,
+      "step": 2419
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8916968142100593e-05,
+      "loss": 0.4387,
+      "step": 2420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8915907502887285e-05,
+      "loss": 0.4313,
+      "step": 2421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8914846374337334e-05,
+      "loss": 0.432,
+      "step": 2422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891378475650897e-05,
+      "loss": 0.4307,
+      "step": 2423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8912722649460465e-05,
+      "loss": 0.4227,
+      "step": 2424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8911660053250106e-05,
+      "loss": 0.4215,
+      "step": 2425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8910596967936218e-05,
+      "loss": 0.4831,
+      "step": 2426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890953339357714e-05,
+      "loss": 0.43,
+      "step": 2427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8908469330231253e-05,
+      "loss": 0.4401,
+      "step": 2428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8907404777956952e-05,
+      "loss": 0.4369,
+      "step": 2429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8906339736812662e-05,
+      "loss": 0.4478,
+      "step": 2430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890527420685684e-05,
+      "loss": 0.4193,
+      "step": 2431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890420818814796e-05,
+      "loss": 0.4012,
+      "step": 2432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890314168074453e-05,
+      "loss": 0.4396,
+      "step": 2433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8902074684705096e-05,
+      "loss": 0.4621,
+      "step": 2434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8901007200088203e-05,
+      "loss": 0.4626,
+      "step": 2435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8899939226952444e-05,
+      "loss": 0.4123,
+      "step": 2436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8898870765356434e-05,
+      "loss": 0.4473,
+      "step": 2437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8897801815358814e-05,
+      "loss": 0.4417,
+      "step": 2438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889673237701825e-05,
+      "loss": 0.4315,
+      "step": 2439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889566245039344e-05,
+      "loss": 0.4596,
+      "step": 2440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.88945920355431e-05,
+      "loss": 0.4478,
+      "step": 2441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8893521132525984e-05,
+      "loss": 0.4347,
+      "step": 2442
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8892449741400864e-05,
+      "loss": 0.4326,
+      "step": 2443
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889137786222654e-05,
+      "loss": 0.4262,
+      "step": 2444
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8890305495061845e-05,
+      "loss": 0.4317,
+      "step": 2445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8889232639965633e-05,
+      "loss": 0.4541,
+      "step": 2446
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.888815929699678e-05,
+      "loss": 0.4091,
+      "step": 2447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8887085466214205e-05,
+      "loss": 0.4598,
+      "step": 2448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8886011147676835e-05,
+      "loss": 0.4465,
+      "step": 2449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8884936341443638e-05,
+      "loss": 0.4703,
+      "step": 2450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88838610475736e-05,
+      "loss": 0.4565,
+      "step": 2451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8882785266125735e-05,
+      "loss": 0.4299,
+      "step": 2452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8881708997159093e-05,
+      "loss": 0.4159,
+      "step": 2453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8880632240732738e-05,
+      "loss": 0.4189,
+      "step": 2454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8879554996905766e-05,
+      "loss": 0.4308,
+      "step": 2455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88784772657373e-05,
+      "loss": 0.4232,
+      "step": 2456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88773990472865e-05,
+      "loss": 0.4171,
+      "step": 2457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8876320341612523e-05,
+      "loss": 0.4175,
+      "step": 2458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887524114877459e-05,
+      "loss": 0.4553,
+      "step": 2459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8874161468831916e-05,
+      "loss": 0.4312,
+      "step": 2460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887308130184377e-05,
+      "loss": 0.4202,
+      "step": 2461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887200064786943e-05,
+      "loss": 0.428,
+      "step": 2462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8870919506968207e-05,
+      "loss": 0.446,
+      "step": 2463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8869837879199438e-05,
+      "loss": 0.4392,
+      "step": 2464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8868755764622488e-05,
+      "loss": 0.4396,
+      "step": 2465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8867673163296746e-05,
+      "loss": 0.4196,
+      "step": 2466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8866590075281624e-05,
+      "loss": 0.4265,
+      "step": 2467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886550650063657e-05,
+      "loss": 0.4131,
+      "step": 2468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886442243942106e-05,
+      "loss": 0.4459,
+      "step": 2469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8863337891694583e-05,
+      "loss": 0.4621,
+      "step": 2470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8862252857516667e-05,
+      "loss": 0.4113,
+      "step": 2471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886116733694686e-05,
+      "loss": 0.4413,
+      "step": 2472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886008133004474e-05,
+      "loss": 0.3948,
+      "step": 2473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8858994836869908e-05,
+      "loss": 0.4365,
+      "step": 2474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8857907857482004e-05,
+      "loss": 0.4469,
+      "step": 2475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8856820391940675e-05,
+      "loss": 0.4436,
+      "step": 2476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8855732440305613e-05,
+      "loss": 0.4417,
+      "step": 2477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885464400263652e-05,
+      "loss": 0.4247,
+      "step": 2478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885355507899314e-05,
+      "loss": 0.4502,
+      "step": 2479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8852465669435235e-05,
+      "loss": 0.4236,
+      "step": 2480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.88513757740226e-05,
+      "loss": 0.4152,
+      "step": 2481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8850285392815043e-05,
+      "loss": 0.4116,
+      "step": 2482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8849194525872415e-05,
+      "loss": 0.4487,
+      "step": 2483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8848103173254584e-05,
+      "loss": 0.4564,
+      "step": 2484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8847011335021447e-05,
+      "loss": 0.452,
+      "step": 2485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8845919011232928e-05,
+      "loss": 0.3977,
+      "step": 2486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8844826201948986e-05,
+      "loss": 0.4204,
+      "step": 2487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8843732907229585e-05,
+      "loss": 0.4249,
+      "step": 2488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8842639127134734e-05,
+      "loss": 0.4579,
+      "step": 2489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8841544861724466e-05,
+      "loss": 0.4323,
+      "step": 2490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8840450111058837e-05,
+      "loss": 0.4385,
+      "step": 2491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883935487519793e-05,
+      "loss": 0.4608,
+      "step": 2492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8838259154201853e-05,
+      "loss": 0.4535,
+      "step": 2493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883716294813075e-05,
+      "loss": 0.4414,
+      "step": 2494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883606625704478e-05,
+      "loss": 0.4202,
+      "step": 2495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883496908100413e-05,
+      "loss": 0.4281,
+      "step": 2496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8833871420069022e-05,
+      "loss": 0.4362,
+      "step": 2497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.88327732742997e-05,
+      "loss": 0.4623,
+      "step": 2498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883167464375643e-05,
+      "loss": 0.4451,
+      "step": 2499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8830575528499513e-05,
+      "loss": 0.4753,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8829475928589272e-05,
+      "loss": 0.4304,
+      "step": 2501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.882837584408605e-05,
+      "loss": 0.4781,
+      "step": 2502
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8827275275050233e-05,
+      "loss": 0.4826,
+      "step": 2503
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882617422154222e-05,
+      "loss": 0.4375,
+      "step": 2504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882507268362244e-05,
+      "loss": 0.4091,
+      "step": 2505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882397066135135e-05,
+      "loss": 0.4606,
+      "step": 2506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8822868154789435e-05,
+      "loss": 0.4274,
+      "step": 2507
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8821765163997197e-05,
+      "loss": 0.4126,
+      "step": 2508
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882066168903518e-05,
+      "loss": 0.4465,
+      "step": 2509
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8819557729963945e-05,
+      "loss": 0.4364,
+      "step": 2510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881845328684408e-05,
+      "loss": 0.4334,
+      "step": 2511
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8817348359736202e-05,
+      "loss": 0.4116,
+      "step": 2512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8816242948700953e-05,
+      "loss": 0.4404,
+      "step": 2513
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8815137053799e-05,
+      "loss": 0.4308,
+      "step": 2514
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881403067509104e-05,
+      "loss": 0.448,
+      "step": 2515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8812923812637792e-05,
+      "loss": 0.4462,
+      "step": 2516
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881181646650001e-05,
+      "loss": 0.4169,
+      "step": 2517
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8810708636738466e-05,
+      "loss": 0.4446,
+      "step": 2518
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880960032341396e-05,
+      "loss": 0.4313,
+      "step": 2519
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8808491526587325e-05,
+      "loss": 0.4247,
+      "step": 2520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8807382246319413e-05,
+      "loss": 0.4279,
+      "step": 2521
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8806272482671103e-05,
+      "loss": 0.4163,
+      "step": 2522
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8805162235703306e-05,
+      "loss": 0.4362,
+      "step": 2523
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880405150547695e-05,
+      "loss": 0.4556,
+      "step": 2524
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8802940292053008e-05,
+      "loss": 0.4204,
+      "step": 2525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8801828595492456e-05,
+      "loss": 0.4393,
+      "step": 2526
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8800716415856316e-05,
+      "loss": 0.4791,
+      "step": 2527
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.879960375320562e-05,
+      "loss": 0.4139,
+      "step": 2528
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.879849060760144e-05,
+      "loss": 0.4231,
+      "step": 2529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8797376979104874e-05,
+      "loss": 0.4296,
+      "step": 2530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8796262867777032e-05,
+      "loss": 0.4175,
+      "step": 2531
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8795148273679066e-05,
+      "loss": 0.427,
+      "step": 2532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8794033196872145e-05,
+      "loss": 0.4326,
+      "step": 2533
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8792917637417478e-05,
+      "loss": 0.4135,
+      "step": 2534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879180159537628e-05,
+      "loss": 0.4161,
+      "step": 2535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8790685070809808e-05,
+      "loss": 0.4468,
+      "step": 2536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8789568063779338e-05,
+      "loss": 0.4248,
+      "step": 2537
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878845057434618e-05,
+      "loss": 0.421,
+      "step": 2538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8787332602571663e-05,
+      "loss": 0.4835,
+      "step": 2539
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8786214148517142e-05,
+      "loss": 0.4543,
+      "step": 2540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878509521224401e-05,
+      "loss": 0.455,
+      "step": 2541
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8783975793813674e-05,
+      "loss": 0.4471,
+      "step": 2542
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878285589328757e-05,
+      "loss": 0.4388,
+      "step": 2543
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878173551072716e-05,
+      "loss": 0.4418,
+      "step": 2544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8780614646193944e-05,
+      "loss": 0.4497,
+      "step": 2545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877949329974943e-05,
+      "loss": 0.3986,
+      "step": 2546
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8778371471455164e-05,
+      "loss": 0.4449,
+      "step": 2547
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8777249161372713e-05,
+      "loss": 0.4317,
+      "step": 2548
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8776126369563683e-05,
+      "loss": 0.4291,
+      "step": 2549
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8775003096089688e-05,
+      "loss": 0.4088,
+      "step": 2550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877387934101238e-05,
+      "loss": 0.4118,
+      "step": 2551
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877275510439343e-05,
+      "loss": 0.4205,
+      "step": 2552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877163038629455e-05,
+      "loss": 0.4402,
+      "step": 2553
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8770505186777463e-05,
+      "loss": 0.4743,
+      "step": 2554
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8769379505903922e-05,
+      "loss": 0.4261,
+      "step": 2555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.876825334373571e-05,
+      "loss": 0.4633,
+      "step": 2556
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8767126700334633e-05,
+      "loss": 0.406,
+      "step": 2557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876599957576253e-05,
+      "loss": 0.4441,
+      "step": 2558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8764871970081262e-05,
+      "loss": 0.4333,
+      "step": 2559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876374388335271e-05,
+      "loss": 0.4382,
+      "step": 2560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8762615315638788e-05,
+      "loss": 0.4383,
+      "step": 2561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876148626700144e-05,
+      "loss": 0.4369,
+      "step": 2562
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8760356737502633e-05,
+      "loss": 0.4297,
+      "step": 2563
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875922672720435e-05,
+      "loss": 0.433,
+      "step": 2564
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8758096236168623e-05,
+      "loss": 0.435,
+      "step": 2565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875696526445749e-05,
+      "loss": 0.4627,
+      "step": 2566
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8755833812133023e-05,
+      "loss": 0.457,
+      "step": 2567
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875470187925732e-05,
+      "loss": 0.4368,
+      "step": 2568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875356946589251e-05,
+      "loss": 0.4523,
+      "step": 2569
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875243657210074e-05,
+      "loss": 0.4465,
+      "step": 2570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8751303197944183e-05,
+      "loss": 0.4292,
+      "step": 2571
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875016934348505e-05,
+      "loss": 0.4111,
+      "step": 2572
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8749035008785567e-05,
+      "loss": 0.444,
+      "step": 2573
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8747900193907994e-05,
+      "loss": 0.4345,
+      "step": 2574
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874676489891461e-05,
+      "loss": 0.4491,
+      "step": 2575
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8745629123867725e-05,
+      "loss": 0.438,
+      "step": 2576
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8744492868829675e-05,
+      "loss": 0.3987,
+      "step": 2577
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874335613386282e-05,
+      "loss": 0.4399,
+      "step": 2578
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8742218919029553e-05,
+      "loss": 0.4583,
+      "step": 2579
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874108122439228e-05,
+      "loss": 0.4502,
+      "step": 2580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873994305001345e-05,
+      "loss": 0.4293,
+      "step": 2581
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8738804395955524e-05,
+      "loss": 0.4303,
+      "step": 2582
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8737665262280997e-05,
+      "loss": 0.4752,
+      "step": 2583
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8736525649052394e-05,
+      "loss": 0.4173,
+      "step": 2584
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8735385556332252e-05,
+      "loss": 0.4669,
+      "step": 2585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.873424498418315e-05,
+      "loss": 0.4276,
+      "step": 2586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8733103932667687e-05,
+      "loss": 0.4192,
+      "step": 2587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.873196240184848e-05,
+      "loss": 0.4643,
+      "step": 2588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.873082039178819e-05,
+      "loss": 0.4313,
+      "step": 2589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8729677902549487e-05,
+      "loss": 0.4477,
+      "step": 2590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8728534934195082e-05,
+      "loss": 0.4258,
+      "step": 2591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8727391486787702e-05,
+      "loss": 0.4632,
+      "step": 2592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.87262475603901e-05,
+      "loss": 0.4351,
+      "step": 2593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872510315506506e-05,
+      "loss": 0.4286,
+      "step": 2594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8723958270875394e-05,
+      "loss": 0.4384,
+      "step": 2595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8722812907883936e-05,
+      "loss": 0.4264,
+      "step": 2596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8721667066153548e-05,
+      "loss": 0.4452,
+      "step": 2597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872052074574712e-05,
+      "loss": 0.4486,
+      "step": 2598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8719373946727558e-05,
+      "loss": 0.4713,
+      "step": 2599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871822666915781e-05,
+      "loss": 0.4393,
+      "step": 2600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8717078913100843e-05,
+      "loss": 0.4231,
+      "step": 2601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8715930678619646e-05,
+      "loss": 0.4457,
+      "step": 2602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8714781965777236e-05,
+      "loss": 0.4498,
+      "step": 2603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871363277463667e-05,
+      "loss": 0.4496,
+      "step": 2604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8712483105261006e-05,
+      "loss": 0.4355,
+      "step": 2605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871133295771335e-05,
+      "loss": 0.4336,
+      "step": 2606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871018233205682e-05,
+      "loss": 0.4367,
+      "step": 2607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870903122835457e-05,
+      "loss": 0.4787,
+      "step": 2608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8707879646669784e-05,
+      "loss": 0.4505,
+      "step": 2609
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8706727587065653e-05,
+      "loss": 0.4427,
+      "step": 2610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8705575049605415e-05,
+      "loss": 0.4343,
+      "step": 2611
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8704422034352316e-05,
+      "loss": 0.4864,
+      "step": 2612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8703268541369643e-05,
+      "loss": 0.4308,
+      "step": 2613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.870211457072071e-05,
+      "loss": 0.4537,
+      "step": 2614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8700960122468838e-05,
+      "loss": 0.4554,
+      "step": 2615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8699805196677395e-05,
+      "loss": 0.4298,
+      "step": 2616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869864979340977e-05,
+      "loss": 0.4259,
+      "step": 2617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869749391272937e-05,
+      "loss": 0.4458,
+      "step": 2618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8696337554699634e-05,
+      "loss": 0.4224,
+      "step": 2619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869518071938403e-05,
+      "loss": 0.4163,
+      "step": 2620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8694023406846046e-05,
+      "loss": 0.4503,
+      "step": 2621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8692865617149204e-05,
+      "loss": 0.4589,
+      "step": 2622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8691707350357047e-05,
+      "loss": 0.4383,
+      "step": 2623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869054860653314e-05,
+      "loss": 0.4384,
+      "step": 2624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8689389385741082e-05,
+      "loss": 0.4407,
+      "step": 2625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8688229688044495e-05,
+      "loss": 0.448,
+      "step": 2626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8687069513507027e-05,
+      "loss": 0.4056,
+      "step": 2627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8685908862192356e-05,
+      "loss": 0.4124,
+      "step": 2628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8684747734164177e-05,
+      "loss": 0.4418,
+      "step": 2629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8683586129486218e-05,
+      "loss": 0.4569,
+      "step": 2630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8682424048222238e-05,
+      "loss": 0.432,
+      "step": 2631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8681261490436007e-05,
+      "loss": 0.437,
+      "step": 2632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8680098456191334e-05,
+      "loss": 0.4129,
+      "step": 2633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8678934945552055e-05,
+      "loss": 0.4366,
+      "step": 2634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8677770958582023e-05,
+      "loss": 0.4192,
+      "step": 2635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867660649534512e-05,
+      "loss": 0.4403,
+      "step": 2636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867544155590526e-05,
+      "loss": 0.4306,
+      "step": 2637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8674276140326376e-05,
+      "loss": 0.4255,
+      "step": 2638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8673110248672434e-05,
+      "loss": 0.4114,
+      "step": 2639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867194388100742e-05,
+      "loss": 0.4332,
+      "step": 2640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8670777037395344e-05,
+      "loss": 0.4191,
+      "step": 2641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8669609717900253e-05,
+      "loss": 0.4388,
+      "step": 2642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866844192258621e-05,
+      "loss": 0.4326,
+      "step": 2643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8667273651517315e-05,
+      "loss": 0.4439,
+      "step": 2644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8666104904757675e-05,
+      "loss": 0.4195,
+      "step": 2645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8664935682371444e-05,
+      "loss": 0.4363,
+      "step": 2646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866376598442279e-05,
+      "loss": 0.4402,
+      "step": 2647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866259581097591e-05,
+      "loss": 0.4236,
+      "step": 2648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8661425162095024e-05,
+      "loss": 0.4518,
+      "step": 2649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.3907,
+      "step": 2650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8659082438288274e-05,
+      "loss": 0.4195,
+      "step": 2651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8657910363490983e-05,
+      "loss": 0.4528,
+      "step": 2652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8656737813516845e-05,
+      "loss": 0.4358,
+      "step": 2653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865556478843021e-05,
+      "loss": 0.4509,
+      "step": 2654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865439128829546e-05,
+      "loss": 0.4289,
+      "step": 2655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8653217313177004e-05,
+      "loss": 0.4658,
+      "step": 2656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8652042863139267e-05,
+      "loss": 0.4208,
+      "step": 2657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8650867938246714e-05,
+      "loss": 0.4099,
+      "step": 2658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8649692538563823e-05,
+      "loss": 0.4554,
+      "step": 2659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864851666415511e-05,
+      "loss": 0.4437,
+      "step": 2660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8647340315085106e-05,
+      "loss": 0.4203,
+      "step": 2661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8646163491418375e-05,
+      "loss": 0.4275,
+      "step": 2662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8644986193219505e-05,
+      "loss": 0.4046,
+      "step": 2663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864380842055311e-05,
+      "loss": 0.4649,
+      "step": 2664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8642630173483832e-05,
+      "loss": 0.4661,
+      "step": 2665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8641451452076338e-05,
+      "loss": 0.4453,
+      "step": 2666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8640272256395316e-05,
+      "loss": 0.4351,
+      "step": 2667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.863909258650549e-05,
+      "loss": 0.4735,
+      "step": 2668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8637912442471603e-05,
+      "loss": 0.4257,
+      "step": 2669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8636731824358426e-05,
+      "loss": 0.4486,
+      "step": 2670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8635550732230748e-05,
+      "loss": 0.4418,
+      "step": 2671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8634369166153398e-05,
+      "loss": 0.4255,
+      "step": 2672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8633187126191227e-05,
+      "loss": 0.4234,
+      "step": 2673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8632004612409105e-05,
+      "loss": 0.4306,
+      "step": 2674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8630821624871935e-05,
+      "loss": 0.4667,
+      "step": 2675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8629638163644644e-05,
+      "loss": 0.4615,
+      "step": 2676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8628454228792178e-05,
+      "loss": 0.4128,
+      "step": 2677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8627269820379526e-05,
+      "loss": 0.4353,
+      "step": 2678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8626084938471686e-05,
+      "loss": 0.4004,
+      "step": 2679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8624899583133686e-05,
+      "loss": 0.4328,
+      "step": 2680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862371375443059e-05,
+      "loss": 0.4523,
+      "step": 2681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8622527452427474e-05,
+      "loss": 0.4418,
+      "step": 2682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862134067718945e-05,
+      "loss": 0.4405,
+      "step": 2683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8620153428781654e-05,
+      "loss": 0.4271,
+      "step": 2684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861896570726924e-05,
+      "loss": 0.4437,
+      "step": 2685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8617777512717402e-05,
+      "loss": 0.4547,
+      "step": 2686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8616588845191346e-05,
+      "loss": 0.419,
+      "step": 2687
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8615399704756317e-05,
+      "loss": 0.4021,
+      "step": 2688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8614210091477572e-05,
+      "loss": 0.446,
+      "step": 2689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8613020005420405e-05,
+      "loss": 0.4351,
+      "step": 2690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8611829446650132e-05,
+      "loss": 0.4337,
+      "step": 2691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8610638415232097e-05,
+      "loss": 0.4431,
+      "step": 2692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8609446911231662e-05,
+      "loss": 0.4729,
+      "step": 2693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8608254934714227e-05,
+      "loss": 0.4918,
+      "step": 2694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8607062485745212e-05,
+      "loss": 0.4688,
+      "step": 2695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8605869564390056e-05,
+      "loss": 0.4176,
+      "step": 2696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8604676170714238e-05,
+      "loss": 0.4249,
+      "step": 2697
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860348230478325e-05,
+      "loss": 0.4051,
+      "step": 2698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860228796666262e-05,
+      "loss": 0.4392,
+      "step": 2699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8601093156417896e-05,
+      "loss": 0.4493,
+      "step": 2700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859989787411465e-05,
+      "loss": 0.4742,
+      "step": 2701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8598702119818493e-05,
+      "loss": 0.4421,
+      "step": 2702
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8597505893595046e-05,
+      "loss": 0.4477,
+      "step": 2703
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8596309195509958e-05,
+      "loss": 0.4689,
+      "step": 2704
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8595112025628913e-05,
+      "loss": 0.4388,
+      "step": 2705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8593914384017616e-05,
+      "loss": 0.4371,
+      "step": 2706
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8592716270741796e-05,
+      "loss": 0.4345,
+      "step": 2707
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8591517685867212e-05,
+      "loss": 0.4469,
+      "step": 2708
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8590318629459643e-05,
+      "loss": 0.4731,
+      "step": 2709
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8589119101584902e-05,
+      "loss": 0.4476,
+      "step": 2710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858791910230882e-05,
+      "loss": 0.4877,
+      "step": 2711
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858671863169726e-05,
+      "loss": 0.4626,
+      "step": 2712
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8585517689816104e-05,
+      "loss": 0.4339,
+      "step": 2713
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8584316276731266e-05,
+      "loss": 0.5082,
+      "step": 2714
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8583114392508684e-05,
+      "loss": 0.456,
+      "step": 2715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8581912037214323e-05,
+      "loss": 0.4116,
+      "step": 2716
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8580709210914173e-05,
+      "loss": 0.4431,
+      "step": 2717
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8579505913674248e-05,
+      "loss": 0.4568,
+      "step": 2718
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8578302145560586e-05,
+      "loss": 0.4434,
+      "step": 2719
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8577097906639256e-05,
+      "loss": 0.4492,
+      "step": 2720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8575893196976357e-05,
+      "loss": 0.4741,
+      "step": 2721
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8574688016637998e-05,
+      "loss": 0.426,
+      "step": 2722
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857348236569033e-05,
+      "loss": 0.46,
+      "step": 2723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.857227624419952e-05,
+      "loss": 0.4018,
+      "step": 2724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.857106965223177e-05,
+      "loss": 0.4288,
+      "step": 2725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8569862589853293e-05,
+      "loss": 0.4566,
+      "step": 2726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8568655057130344e-05,
+      "loss": 0.4459,
+      "step": 2727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8567447054129196e-05,
+      "loss": 0.4285,
+      "step": 2728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8566238580916147e-05,
+      "loss": 0.4593,
+      "step": 2729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8565029637557517e-05,
+      "loss": 0.4098,
+      "step": 2730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8563820224119668e-05,
+      "loss": 0.4323,
+      "step": 2731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856261034066897e-05,
+      "loss": 0.4409,
+      "step": 2732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8561399987271827e-05,
+      "loss": 0.454,
+      "step": 2733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8560189163994664e-05,
+      "loss": 0.4246,
+      "step": 2734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8558977870903942e-05,
+      "loss": 0.4759,
+      "step": 2735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8557766108066135e-05,
+      "loss": 0.4184,
+      "step": 2736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8556553875547755e-05,
+      "loss": 0.452,
+      "step": 2737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8555341173415333e-05,
+      "loss": 0.4332,
+      "step": 2738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8554128001735417e-05,
+      "loss": 0.4241,
+      "step": 2739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85529143605746e-05,
+      "loss": 0.4325,
+      "step": 2740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8551700249999488e-05,
+      "loss": 0.4535,
+      "step": 2741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8550485670076713e-05,
+      "loss": 0.4102,
+      "step": 2742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8549270620872938e-05,
+      "loss": 0.4151,
+      "step": 2743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8548055102454852e-05,
+      "loss": 0.4513,
+      "step": 2744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854683911488916e-05,
+      "loss": 0.4362,
+      "step": 2745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854562265824261e-05,
+      "loss": 0.4138,
+      "step": 2746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8544405732581952e-05,
+      "loss": 0.4444,
+      "step": 2747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854318833797399e-05,
+      "loss": 0.4365,
+      "step": 2748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8541970474485523e-05,
+      "loss": 0.4302,
+      "step": 2749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8540752142183407e-05,
+      "loss": 0.4072,
+      "step": 2750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8539533341134498e-05,
+      "loss": 0.4431,
+      "step": 2751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8538314071405694e-05,
+      "loss": 0.4484,
+      "step": 2752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8537094333063908e-05,
+      "loss": 0.4419,
+      "step": 2753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8535874126176087e-05,
+      "loss": 0.4234,
+      "step": 2754
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.85346534508092e-05,
+      "loss": 0.4614,
+      "step": 2755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853343230703024e-05,
+      "loss": 0.4443,
+      "step": 2756
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8532210694906232e-05,
+      "loss": 0.4218,
+      "step": 2757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8530988614504216e-05,
+      "loss": 0.447,
+      "step": 2758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852976606589127e-05,
+      "loss": 0.4625,
+      "step": 2759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8528543049134485e-05,
+      "loss": 0.4579,
+      "step": 2760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8527319564300993e-05,
+      "loss": 0.4012,
+      "step": 2761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8526095611457938e-05,
+      "loss": 0.4214,
+      "step": 2762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8524871190672497e-05,
+      "loss": 0.4317,
+      "step": 2763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8523646302011867e-05,
+      "loss": 0.417,
+      "step": 2764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8522420945543274e-05,
+      "loss": 0.4387,
+      "step": 2765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8521195121333978e-05,
+      "loss": 0.4383,
+      "step": 2766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851996882945125e-05,
+      "loss": 0.4359,
+      "step": 2767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8518742069962395e-05,
+      "loss": 0.4271,
+      "step": 2768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851751484293474e-05,
+      "loss": 0.4944,
+      "step": 2769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8516287148435638e-05,
+      "loss": 0.4243,
+      "step": 2770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8515058986532474e-05,
+      "loss": 0.4001,
+      "step": 2771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8513830357292654e-05,
+      "loss": 0.4377,
+      "step": 2772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851260126078361e-05,
+      "loss": 0.405,
+      "step": 2773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8511371697072792e-05,
+      "loss": 0.4251,
+      "step": 2774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851014166622769e-05,
+      "loss": 0.4292,
+      "step": 2775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850891116831581e-05,
+      "loss": 0.4201,
+      "step": 2776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850768020340468e-05,
+      "loss": 0.4183,
+      "step": 2777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8506448771561872e-05,
+      "loss": 0.4453,
+      "step": 2778
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8505216872854964e-05,
+      "loss": 0.4725,
+      "step": 2779
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.850398450735157e-05,
+      "loss": 0.4245,
+      "step": 2780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8502751675119324e-05,
+      "loss": 0.4409,
+      "step": 2781
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8501518376225887e-05,
+      "loss": 0.405,
+      "step": 2782
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8500284610738947e-05,
+      "loss": 0.4077,
+      "step": 2783
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8499050378726226e-05,
+      "loss": 0.4113,
+      "step": 2784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8497815680255452e-05,
+      "loss": 0.4218,
+      "step": 2785
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8496580515394392e-05,
+      "loss": 0.4514,
+      "step": 2786
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849534488421084e-05,
+      "loss": 0.4542,
+      "step": 2787
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8494108786772613e-05,
+      "loss": 0.4468,
+      "step": 2788
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849287222314754e-05,
+      "loss": 0.4585,
+      "step": 2789
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8491635193403507e-05,
+      "loss": 0.4453,
+      "step": 2790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8490397697608394e-05,
+      "loss": 0.4247,
+      "step": 2791
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8489159735830122e-05,
+      "loss": 0.4341,
+      "step": 2792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8487921308136634e-05,
+      "loss": 0.4388,
+      "step": 2793
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8486682414595903e-05,
+      "loss": 0.4539,
+      "step": 2794
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848544305527592e-05,
+      "loss": 0.4171,
+      "step": 2795
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.84842032302447e-05,
+      "loss": 0.4327,
+      "step": 2796
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8482962939570302e-05,
+      "loss": 0.4401,
+      "step": 2797
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848172218332079e-05,
+      "loss": 0.4211,
+      "step": 2798
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848048096156426e-05,
+      "loss": 0.4192,
+      "step": 2799
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847923927436884e-05,
+      "loss": 0.444,
+      "step": 2800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8477997121802674e-05,
+      "loss": 0.4699,
+      "step": 2801
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8476754503933937e-05,
+      "loss": 0.424,
+      "step": 2802
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8475511420830827e-05,
+      "loss": 0.4533,
+      "step": 2803
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8474267872561568e-05,
+      "loss": 0.4202,
+      "step": 2804
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8473023859194413e-05,
+      "loss": 0.3983,
+      "step": 2805
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8471779380797638e-05,
+      "loss": 0.4301,
+      "step": 2806
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8470534437439546e-05,
+      "loss": 0.4292,
+      "step": 2807
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8469289029188457e-05,
+      "loss": 0.4285,
+      "step": 2808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846804315611273e-05,
+      "loss": 0.4306,
+      "step": 2809
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846679681828074e-05,
+      "loss": 0.4314,
+      "step": 2810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846555001576089e-05,
+      "loss": 0.4451,
+      "step": 2811
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846430274862161e-05,
+      "loss": 0.4207,
+      "step": 2812
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8463055016931358e-05,
+      "loss": 0.4469,
+      "step": 2813
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846180682075861e-05,
+      "loss": 0.4569,
+      "step": 2814
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8460558160171866e-05,
+      "loss": 0.4595,
+      "step": 2815
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8459309035239667e-05,
+      "loss": 0.4111,
+      "step": 2816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8458059446030565e-05,
+      "loss": 0.4426,
+      "step": 2817
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845680939261314e-05,
+      "loss": 0.4292,
+      "step": 2818
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8455558875056004e-05,
+      "loss": 0.4319,
+      "step": 2819
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8454307893427787e-05,
+      "loss": 0.4365,
+      "step": 2820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8453056447797148e-05,
+      "loss": 0.42,
+      "step": 2821
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845180453823277e-05,
+      "loss": 0.4128,
+      "step": 2822
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8450552164803363e-05,
+      "loss": 0.4327,
+      "step": 2823
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844929932757766e-05,
+      "loss": 0.4392,
+      "step": 2824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8448046026624423e-05,
+      "loss": 0.4376,
+      "step": 2825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844679226201244e-05,
+      "loss": 0.4165,
+      "step": 2826
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8445538033810515e-05,
+      "loss": 0.4522,
+      "step": 2827
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844428334208749e-05,
+      "loss": 0.4261,
+      "step": 2828
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8443028186912226e-05,
+      "loss": 0.4327,
+      "step": 2829
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844177256835361e-05,
+      "loss": 0.4298,
+      "step": 2830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8440516486480554e-05,
+      "loss": 0.4562,
+      "step": 2831
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8439259941361998e-05,
+      "loss": 0.432,
+      "step": 2832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8438002933066904e-05,
+      "loss": 0.4221,
+      "step": 2833
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.843674546166426e-05,
+      "loss": 0.4141,
+      "step": 2834
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843548752722309e-05,
+      "loss": 0.4226,
+      "step": 2835
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8434229129812417e-05,
+      "loss": 0.4533,
+      "step": 2836
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843297026950132e-05,
+      "loss": 0.4198,
+      "step": 2837
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843171094635888e-05,
+      "loss": 0.4411,
+      "step": 2838
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8430451160454222e-05,
+      "loss": 0.4502,
+      "step": 2839
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8429190911856486e-05,
+      "loss": 0.4301,
+      "step": 2840
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8427930200634833e-05,
+      "loss": 0.4384,
+      "step": 2841
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8426669026858462e-05,
+      "loss": 0.4154,
+      "step": 2842
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8425407390596586e-05,
+      "loss": 0.4058,
+      "step": 2843
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842414529191845e-05,
+      "loss": 0.3966,
+      "step": 2844
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8422882730893323e-05,
+      "loss": 0.4481,
+      "step": 2845
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8421619707590496e-05,
+      "loss": 0.442,
+      "step": 2846
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8420356222079293e-05,
+      "loss": 0.417,
+      "step": 2847
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8419092274429052e-05,
+      "loss": 0.4253,
+      "step": 2848
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841782786470915e-05,
+      "loss": 0.4713,
+      "step": 2849
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8416562992988977e-05,
+      "loss": 0.4606,
+      "step": 2850
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841529765933796e-05,
+      "loss": 0.4462,
+      "step": 2851
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8414031863825536e-05,
+      "loss": 0.4561,
+      "step": 2852
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8412765606521183e-05,
+      "loss": 0.4458,
+      "step": 2853
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8411498887494396e-05,
+      "loss": 0.4636,
+      "step": 2854
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.84102317068147e-05,
+      "loss": 0.4486,
+      "step": 2855
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8408964064551633e-05,
+      "loss": 0.4336,
+      "step": 2856
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8407695960774777e-05,
+      "loss": 0.4386,
+      "step": 2857
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8406427395553725e-05,
+      "loss": 0.438,
+      "step": 2858
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8405158368958105e-05,
+      "loss": 0.4271,
+      "step": 2859
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8403888881057562e-05,
+      "loss": 0.4365,
+      "step": 2860
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840261893192177e-05,
+      "loss": 0.4439,
+      "step": 2861
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8401348521620427e-05,
+      "loss": 0.4244,
+      "step": 2862
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8400077650223264e-05,
+      "loss": 0.44,
+      "step": 2863
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8398806317800027e-05,
+      "loss": 0.4275,
+      "step": 2864
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8397534524420487e-05,
+      "loss": 0.4248,
+      "step": 2865
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839626227015445e-05,
+      "loss": 0.4632,
+      "step": 2866
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839498955507175e-05,
+      "loss": 0.4342,
+      "step": 2867
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8393716379242214e-05,
+      "loss": 0.4584,
+      "step": 2868
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8392442742735743e-05,
+      "loss": 0.4199,
+      "step": 2869
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839116864562222e-05,
+      "loss": 0.4355,
+      "step": 2870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8389894087971587e-05,
+      "loss": 0.4346,
+      "step": 2871
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838861906985379e-05,
+      "loss": 0.4396,
+      "step": 2872
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8387343591338803e-05,
+      "loss": 0.4295,
+      "step": 2873
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8386067652496635e-05,
+      "loss": 0.4629,
+      "step": 2874
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8384791253397312e-05,
+      "loss": 0.4065,
+      "step": 2875
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8383514394110884e-05,
+      "loss": 0.4177,
+      "step": 2876
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838223707470743e-05,
+      "loss": 0.4354,
+      "step": 2877
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838095929525706e-05,
+      "loss": 0.4399,
+      "step": 2878
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8379681055829897e-05,
+      "loss": 0.4124,
+      "step": 2879
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8378402356496102e-05,
+      "loss": 0.4219,
+      "step": 2880
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8377123197325843e-05,
+      "loss": 0.4169,
+      "step": 2881
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8375843578389335e-05,
+      "loss": 0.4148,
+      "step": 2882
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8374563499756805e-05,
+      "loss": 0.4492,
+      "step": 2883
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837328296149851e-05,
+      "loss": 0.4094,
+      "step": 2884
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8372001963684725e-05,
+      "loss": 0.4458,
+      "step": 2885
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8370720506385757e-05,
+      "loss": 0.4801,
+      "step": 2886
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8369438589671945e-05,
+      "loss": 0.4453,
+      "step": 2887
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8368156213613636e-05,
+      "loss": 0.4355,
+      "step": 2888
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8366873378281216e-05,
+      "loss": 0.4743,
+      "step": 2889
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8365590083745085e-05,
+      "loss": 0.4491,
+      "step": 2890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8364306330075688e-05,
+      "loss": 0.4464,
+      "step": 2891
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836302211734347e-05,
+      "loss": 0.4353,
+      "step": 2892
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8361737445618916e-05,
+      "loss": 0.4546,
+      "step": 2893
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8360452314972536e-05,
+      "loss": 0.4146,
+      "step": 2894
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835916672547486e-05,
+      "loss": 0.4456,
+      "step": 2895
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8357880677196446e-05,
+      "loss": 0.448,
+      "step": 2896
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8356594170207876e-05,
+      "loss": 0.466,
+      "step": 2897
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835530720457976e-05,
+      "loss": 0.4414,
+      "step": 2898
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8354019780382736e-05,
+      "loss": 0.4474,
+      "step": 2899
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835273189768745e-05,
+      "loss": 0.429,
+      "step": 2900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.83514435565646e-05,
+      "loss": 0.4716,
+      "step": 2901
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8350154757084882e-05,
+      "loss": 0.4348,
+      "step": 2902
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834886549931904e-05,
+      "loss": 0.4047,
+      "step": 2903
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8347575783337823e-05,
+      "loss": 0.4511,
+      "step": 2904
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8346285609212025e-05,
+      "loss": 0.4048,
+      "step": 2905
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834499497701245e-05,
+      "loss": 0.429,
+      "step": 2906
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8343703886809933e-05,
+      "loss": 0.4319,
+      "step": 2907
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834241233867533e-05,
+      "loss": 0.4317,
+      "step": 2908
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8341120332679537e-05,
+      "loss": 0.4093,
+      "step": 2909
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8339827868893452e-05,
+      "loss": 0.4282,
+      "step": 2910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8338534947388015e-05,
+      "loss": 0.4343,
+      "step": 2911
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8337241568234187e-05,
+      "loss": 0.4811,
+      "step": 2912
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8335947731502953e-05,
+      "loss": 0.4981,
+      "step": 2913
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8334653437265323e-05,
+      "loss": 0.4479,
+      "step": 2914
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8333358685592325e-05,
+      "loss": 0.4571,
+      "step": 2915
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8332063476555033e-05,
+      "loss": 0.4565,
+      "step": 2916
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8330767810224525e-05,
+      "loss": 0.478,
+      "step": 2917
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8329471686671913e-05,
+      "loss": 0.4306,
+      "step": 2918
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832817510596833e-05,
+      "loss": 0.4429,
+      "step": 2919
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8326878068184945e-05,
+      "loss": 0.4374,
+      "step": 2920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832558057339293e-05,
+      "loss": 0.4482,
+      "step": 2921
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8324282621663515e-05,
+      "loss": 0.4353,
+      "step": 2922
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832298421306792e-05,
+      "loss": 0.4625,
+      "step": 2923
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8321685347677413e-05,
+      "loss": 0.4237,
+      "step": 2924
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8320386025563278e-05,
+      "loss": 0.4273,
+      "step": 2925
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8319086246796833e-05,
+      "loss": 0.4241,
+      "step": 2926
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8317786011449406e-05,
+      "loss": 0.4237,
+      "step": 2927
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8316485319592364e-05,
+      "loss": 0.4069,
+      "step": 2928
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831518417129709e-05,
+      "loss": 0.4195,
+      "step": 2929
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8313882566634994e-05,
+      "loss": 0.463,
+      "step": 2930
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831258050567752e-05,
+      "loss": 0.4639,
+      "step": 2931
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831127798849613e-05,
+      "loss": 0.4217,
+      "step": 2932
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8309975015162297e-05,
+      "loss": 0.4188,
+      "step": 2933
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830867158574755e-05,
+      "loss": 0.434,
+      "step": 2934
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8307367700323412e-05,
+      "loss": 0.4326,
+      "step": 2935
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8306063358961453e-05,
+      "loss": 0.426,
+      "step": 2936
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830475856173326e-05,
+      "loss": 0.4339,
+      "step": 2937
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830345330871044e-05,
+      "loss": 0.4307,
+      "step": 2938
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8302147599964635e-05,
+      "loss": 0.3935,
+      "step": 2939
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8300841435567504e-05,
+      "loss": 0.4651,
+      "step": 2940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8299534815590735e-05,
+      "loss": 0.5073,
+      "step": 2941
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.829822774010604e-05,
+      "loss": 0.4516,
+      "step": 2942
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8296920209185153e-05,
+      "loss": 0.4424,
+      "step": 2943
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8295612222899838e-05,
+      "loss": 0.4842,
+      "step": 2944
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8294303781321883e-05,
+      "loss": 0.4221,
+      "step": 2945
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.82929948845231e-05,
+      "loss": 0.451,
+      "step": 2946
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8291685532575325e-05,
+      "loss": 0.4391,
+      "step": 2947
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8290375725550417e-05,
+      "loss": 0.4429,
+      "step": 2948
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828906546352027e-05,
+      "loss": 0.4247,
+      "step": 2949
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8287754746556793e-05,
+      "loss": 0.4634,
+      "step": 2950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8286443574731915e-05,
+      "loss": 0.4375,
+      "step": 2951
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828513194811761e-05,
+      "loss": 0.4597,
+      "step": 2952
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8283819866785855e-05,
+      "loss": 0.4523,
+      "step": 2953
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8282507330808666e-05,
+      "loss": 0.4278,
+      "step": 2954
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8281194340258082e-05,
+      "loss": 0.4428,
+      "step": 2955
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827988089520616e-05,
+      "loss": 0.4197,
+      "step": 2956
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827856699572499e-05,
+      "loss": 0.4144,
+      "step": 2957
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8277252641886677e-05,
+      "loss": 0.4315,
+      "step": 2958
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8275937833763367e-05,
+      "loss": 0.4059,
+      "step": 2959
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8274622571427216e-05,
+      "loss": 0.4562,
+      "step": 2960
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8273306854950408e-05,
+      "loss": 0.4247,
+      "step": 2961
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827199068440516e-05,
+      "loss": 0.4956,
+      "step": 2962
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8270674059863705e-05,
+      "loss": 0.4804,
+      "step": 2963
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8269356981398303e-05,
+      "loss": 0.419,
+      "step": 2964
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826803944908124e-05,
+      "loss": 0.4441,
+      "step": 2965
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826672146298483e-05,
+      "loss": 0.4662,
+      "step": 2966
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8265403023181406e-05,
+      "loss": 0.4241,
+      "step": 2967
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826408412974333e-05,
+      "loss": 0.4519,
+      "step": 2968
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8262764782742986e-05,
+      "loss": 0.449,
+      "step": 2969
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8261444982252784e-05,
+      "loss": 0.4387,
+      "step": 2970
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8260124728345163e-05,
+      "loss": 0.4556,
+      "step": 2971
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.825880402109258e-05,
+      "loss": 0.4385,
+      "step": 2972
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825748286056752e-05,
+      "loss": 0.4544,
+      "step": 2973
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8256161246842497e-05,
+      "loss": 0.4483,
+      "step": 2974
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825483917999004e-05,
+      "loss": 0.4543,
+      "step": 2975
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8253516660082714e-05,
+      "loss": 0.4143,
+      "step": 2976
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82521936871931e-05,
+      "loss": 0.4214,
+      "step": 2977
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8250870261393808e-05,
+      "loss": 0.4305,
+      "step": 2978
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824954638275747e-05,
+      "loss": 0.4284,
+      "step": 2979
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8248222051356756e-05,
+      "loss": 0.4326,
+      "step": 2980
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8246897267264335e-05,
+      "loss": 0.4539,
+      "step": 2981
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8245572030552923e-05,
+      "loss": 0.4238,
+      "step": 2982
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8244246341295256e-05,
+      "loss": 0.467,
+      "step": 2983
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824292019956409e-05,
+      "loss": 0.4552,
+      "step": 2984
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8241593605432207e-05,
+      "loss": 0.4293,
+      "step": 2985
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824026655897242e-05,
+      "loss": 0.4324,
+      "step": 2986
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8238939060257557e-05,
+      "loss": 0.439,
+      "step": 2987
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8237611109360478e-05,
+      "loss": 0.4461,
+      "step": 2988
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8236282706354064e-05,
+      "loss": 0.432,
+      "step": 2989
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8234953851311225e-05,
+      "loss": 0.4359,
+      "step": 2990
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823362454430489e-05,
+      "loss": 0.4303,
+      "step": 2991
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823229478540802e-05,
+      "loss": 0.4294,
+      "step": 2992
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8230964574693593e-05,
+      "loss": 0.4491,
+      "step": 2993
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822963391223462e-05,
+      "loss": 0.4142,
+      "step": 2994
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8228302798104128e-05,
+      "loss": 0.4299,
+      "step": 2995
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8226971232375175e-05,
+      "loss": 0.4431,
+      "step": 2996
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8225639215120846e-05,
+      "loss": 0.461,
+      "step": 2997
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822430674641424e-05,
+      "loss": 0.4306,
+      "step": 2998
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822297382632849e-05,
+      "loss": 0.4571,
+      "step": 2999
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8221640454936755e-05,
+      "loss": 0.4334,
+      "step": 3000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.822030663231221e-05,
+      "loss": 0.4314,
+      "step": 3001
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8218972358528063e-05,
+      "loss": 0.432,
+      "step": 3002
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8217637633657546e-05,
+      "loss": 0.428,
+      "step": 3003
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821630245777391e-05,
+      "loss": 0.4399,
+      "step": 3004
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8214966830950432e-05,
+      "loss": 0.4194,
+      "step": 3005
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821363075326042e-05,
+      "loss": 0.4262,
+      "step": 3006
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.82122942247772e-05,
+      "loss": 0.4329,
+      "step": 3007
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8210957245574124e-05,
+      "loss": 0.4114,
+      "step": 3008
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8209619815724578e-05,
+      "loss": 0.4515,
+      "step": 3009
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8208281935301957e-05,
+      "loss": 0.4591,
+      "step": 3010
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8206943604379693e-05,
+      "loss": 0.4712,
+      "step": 3011
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8205604823031237e-05,
+      "loss": 0.4844,
+      "step": 3012
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8204265591330067e-05,
+      "loss": 0.4205,
+      "step": 3013
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820292590934968e-05,
+      "loss": 0.4406,
+      "step": 3014
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8201585777163606e-05,
+      "loss": 0.4167,
+      "step": 3015
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.82002451948454e-05,
+      "loss": 0.4658,
+      "step": 3016
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8198904162468628e-05,
+      "loss": 0.4232,
+      "step": 3017
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.81975626801069e-05,
+      "loss": 0.4358,
+      "step": 3018
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8196220747833844e-05,
+      "loss": 0.4315,
+      "step": 3019
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8194878365723098e-05,
+      "loss": 0.4497,
+      "step": 3020
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8193535533848343e-05,
+      "loss": 0.4388,
+      "step": 3021
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819219225228328e-05,
+      "loss": 0.4362,
+      "step": 3022
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8190848521101632e-05,
+      "loss": 0.4398,
+      "step": 3023
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8189504340377144e-05,
+      "loss": 0.4308,
+      "step": 3024
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8188159710183595e-05,
+      "loss": 0.4385,
+      "step": 3025
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8186814630594777e-05,
+      "loss": 0.4353,
+      "step": 3026
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8185469101684522e-05,
+      "loss": 0.4127,
+      "step": 3027
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8184123123526664e-05,
+      "loss": 0.4061,
+      "step": 3028
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.818277669619509e-05,
+      "loss": 0.4221,
+      "step": 3029
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8181429819763685e-05,
+      "loss": 0.4238,
+      "step": 3030
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8180082494306378e-05,
+      "loss": 0.4209,
+      "step": 3031
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817873471989711e-05,
+      "loss": 0.4472,
+      "step": 3032
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8177386496609855e-05,
+      "loss": 0.4105,
+      "step": 3033
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8176037824518603e-05,
+      "loss": 0.4403,
+      "step": 3034
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817468870369738e-05,
+      "loss": 0.4006,
+      "step": 3035
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817333913422023e-05,
+      "loss": 0.4556,
+      "step": 3036
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8171989116161216e-05,
+      "loss": 0.4532,
+      "step": 3037
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817063864959444e-05,
+      "loss": 0.4359,
+      "step": 3038
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8169287734594013e-05,
+      "loss": 0.4325,
+      "step": 3039
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8167936371234084e-05,
+      "loss": 0.4179,
+      "step": 3040
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816658455958882e-05,
+      "loss": 0.4536,
+      "step": 3041
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8165232299732408e-05,
+      "loss": 0.422,
+      "step": 3042
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8163879591739067e-05,
+      "loss": 0.4197,
+      "step": 3043
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816252643568304e-05,
+      "loss": 0.4597,
+      "step": 3044
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8161172831638593e-05,
+      "loss": 0.4178,
+      "step": 3045
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815981877968002e-05,
+      "loss": 0.4364,
+      "step": 3046
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8158464279881628e-05,
+      "loss": 0.4571,
+      "step": 3047
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8157109332317765e-05,
+      "loss": 0.4589,
+      "step": 3048
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8155753937062786e-05,
+      "loss": 0.4115,
+      "step": 3049
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8154398094191092e-05,
+      "loss": 0.4587,
+      "step": 3050
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8153041803777085e-05,
+      "loss": 0.4495,
+      "step": 3051
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815168506589521e-05,
+      "loss": 0.4007,
+      "step": 3052
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8150327880619926e-05,
+      "loss": 0.478,
+      "step": 3053
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8148970248025724e-05,
+      "loss": 0.458,
+      "step": 3054
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.814761216818711e-05,
+      "loss": 0.4525,
+      "step": 3055
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8146253641178626e-05,
+      "loss": 0.4396,
+      "step": 3056
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8144894667074828e-05,
+      "loss": 0.4486,
+      "step": 3057
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8143535245950306e-05,
+      "loss": 0.4474,
+      "step": 3058
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8142175377879665e-05,
+      "loss": 0.4568,
+      "step": 3059
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8140815062937545e-05,
+      "loss": 0.4461,
+      "step": 3060
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.81394543011986e-05,
+      "loss": 0.4446,
+      "step": 3061
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8138093092737517e-05,
+      "loss": 0.4023,
+      "step": 3062
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8136731437628998e-05,
+      "loss": 0.4502,
+      "step": 3063
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8135369335947783e-05,
+      "loss": 0.4354,
+      "step": 3064
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8134006787768626e-05,
+      "loss": 0.4439,
+      "step": 3065
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8132643793166305e-05,
+      "loss": 0.444,
+      "step": 3066
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813128035221563e-05,
+      "loss": 0.4354,
+      "step": 3067
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8129916464991433e-05,
+      "loss": 0.4404,
+      "step": 3068
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8128552131568564e-05,
+      "loss": 0.4494,
+      "step": 3069
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8127187352021908e-05,
+      "loss": 0.4222,
+      "step": 3070
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812582212642636e-05,
+      "loss": 0.4159,
+      "step": 3071
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812445645485686e-05,
+      "loss": 0.4121,
+      "step": 3072
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8123090337388353e-05,
+      "loss": 0.4639,
+      "step": 3073
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8121723774095817e-05,
+      "loss": 0.4465,
+      "step": 3074
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812035676505426e-05,
+      "loss": 0.4569,
+      "step": 3075
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.81189893103387e-05,
+      "loss": 0.4271,
+      "step": 3076
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8117621410024192e-05,
+      "loss": 0.4599,
+      "step": 3077
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8116253064185815e-05,
+      "loss": 0.4352,
+      "step": 3078
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8114884272898662e-05,
+      "loss": 0.4451,
+      "step": 3079
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8113515036237858e-05,
+      "loss": 0.4256,
+      "step": 3080
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8112145354278552e-05,
+      "loss": 0.4475,
+      "step": 3081
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811077522709592e-05,
+      "loss": 0.4407,
+      "step": 3082
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8109404654765156e-05,
+      "loss": 0.4438,
+      "step": 3083
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8108033637361485e-05,
+      "loss": 0.4616,
+      "step": 3084
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810666217496015e-05,
+      "loss": 0.436,
+      "step": 3085
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8105290267636428e-05,
+      "loss": 0.4702,
+      "step": 3086
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8103917915465604e-05,
+      "loss": 0.4437,
+      "step": 3087
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8102545118523008e-05,
+      "loss": 0.4285,
+      "step": 3088
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8101171876883977e-05,
+      "loss": 0.4398,
+      "step": 3089
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809979819062388e-05,
+      "loss": 0.4265,
+      "step": 3090
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8098424059818116e-05,
+      "loss": 0.4328,
+      "step": 3091
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8097049484542094e-05,
+      "loss": 0.4413,
+      "step": 3092
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809567446487126e-05,
+      "loss": 0.4256,
+      "step": 3093
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8094299000881078e-05,
+      "loss": 0.4577,
+      "step": 3094
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8092923092647038e-05,
+      "loss": 0.4489,
+      "step": 3095
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8091546740244658e-05,
+      "loss": 0.4576,
+      "step": 3096
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.4338,
+      "step": 3097
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8088792703237056e-05,
+      "loss": 0.4353,
+      "step": 3098
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8087415018782983e-05,
+      "loss": 0.4497,
+      "step": 3099
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8086036890462865e-05,
+      "loss": 0.4211,
+      "step": 3100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808465831835235e-05,
+      "loss": 0.3849,
+      "step": 3101
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8083279302527093e-05,
+      "loss": 0.449,
+      "step": 3102
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808189984306278e-05,
+      "loss": 0.4385,
+      "step": 3103
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8080519940035118e-05,
+      "loss": 0.4683,
+      "step": 3104
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8079139593519843e-05,
+      "loss": 0.3856,
+      "step": 3105
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8077758803592717e-05,
+      "loss": 0.4258,
+      "step": 3106
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807637757032952e-05,
+      "loss": 0.4128,
+      "step": 3107
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807499589380605e-05,
+      "loss": 0.4617,
+      "step": 3108
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8073613774098154e-05,
+      "loss": 0.4079,
+      "step": 3109
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8072231211281677e-05,
+      "loss": 0.4131,
+      "step": 3110
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.80708482054325e-05,
+      "loss": 0.4408,
+      "step": 3111
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8069464756626534e-05,
+      "loss": 0.4464,
+      "step": 3112
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8068080864939697e-05,
+      "loss": 0.4392,
+      "step": 3113
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8066696530447953e-05,
+      "loss": 0.4327,
+      "step": 3114
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8065311753227272e-05,
+      "loss": 0.4373,
+      "step": 3115
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8063926533353658e-05,
+      "loss": 0.4239,
+      "step": 3116
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8062540870903133e-05,
+      "loss": 0.4516,
+      "step": 3117
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8061154765951755e-05,
+      "loss": 0.4459,
+      "step": 3118
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8059768218575587e-05,
+      "loss": 0.4169,
+      "step": 3119
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805838122885074e-05,
+      "loss": 0.4284,
+      "step": 3120
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805699379685333e-05,
+      "loss": 0.4089,
+      "step": 3121
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8055605922659505e-05,
+      "loss": 0.4264,
+      "step": 3122
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805421760634544e-05,
+      "loss": 0.4094,
+      "step": 3123
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8052828847987324e-05,
+      "loss": 0.4316,
+      "step": 3124
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805143964766138e-05,
+      "loss": 0.4212,
+      "step": 3125
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8050050005443852e-05,
+      "loss": 0.4262,
+      "step": 3126
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8048659921411012e-05,
+      "loss": 0.4151,
+      "step": 3127
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804726939563915e-05,
+      "loss": 0.3996,
+      "step": 3128
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804587842820458e-05,
+      "loss": 0.436,
+      "step": 3129
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8044487019183647e-05,
+      "loss": 0.44,
+      "step": 3130
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804309516865272e-05,
+      "loss": 0.4228,
+      "step": 3131
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804170287668818e-05,
+      "loss": 0.4546,
+      "step": 3132
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8040310143366447e-05,
+      "loss": 0.4218,
+      "step": 3133
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8038916968763957e-05,
+      "loss": 0.4582,
+      "step": 3134
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803752335295717e-05,
+      "loss": 0.4367,
+      "step": 3135
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803612929602258e-05,
+      "loss": 0.4081,
+      "step": 3136
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8034734798036694e-05,
+      "loss": 0.4403,
+      "step": 3137
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803333985907604e-05,
+      "loss": 0.4499,
+      "step": 3138
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.803194447921719e-05,
+      "loss": 0.4565,
+      "step": 3139
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8030548658536716e-05,
+      "loss": 0.4178,
+      "step": 3140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802915239711123e-05,
+      "loss": 0.4442,
+      "step": 3141
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8027755695017368e-05,
+      "loss": 0.4552,
+      "step": 3142
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8026358552331783e-05,
+      "loss": 0.4567,
+      "step": 3143
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8024960969131152e-05,
+      "loss": 0.4368,
+      "step": 3144
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802356294549218e-05,
+      "loss": 0.4299,
+      "step": 3145
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8022164481491596e-05,
+      "loss": 0.4406,
+      "step": 3146
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802076557720616e-05,
+      "loss": 0.452,
+      "step": 3147
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8019366232712635e-05,
+      "loss": 0.4212,
+      "step": 3148
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801796644808784e-05,
+      "loss": 0.442,
+      "step": 3149
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801656622340858e-05,
+      "loss": 0.4621,
+      "step": 3150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801516555875172e-05,
+      "loss": 0.4403,
+      "step": 3151
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801376445419412e-05,
+      "loss": 0.4178,
+      "step": 3152
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801236290981269e-05,
+      "loss": 0.4464,
+      "step": 3153
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801096092568435e-05,
+      "loss": 0.4428,
+      "step": 3154
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800955850188604e-05,
+      "loss": 0.4309,
+      "step": 3155
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800815563849473e-05,
+      "loss": 0.4215,
+      "step": 3156
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800675233558742e-05,
+      "loss": 0.4127,
+      "step": 3157
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8005348593241124e-05,
+      "loss": 0.4458,
+      "step": 3158
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8003944411532884e-05,
+      "loss": 0.4242,
+      "step": 3159
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8002539790539772e-05,
+      "loss": 0.426,
+      "step": 3160
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800113473033887e-05,
+      "loss": 0.4219,
+      "step": 3161
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.79997292310073e-05,
+      "loss": 0.4549,
+      "step": 3162
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.79983232926222e-05,
+      "loss": 0.4238,
+      "step": 3163
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7996916915260727e-05,
+      "loss": 0.4228,
+      "step": 3164
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7995510099000074e-05,
+      "loss": 0.419,
+      "step": 3165
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7994102843917445e-05,
+      "loss": 0.4181,
+      "step": 3166
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7992695150090085e-05,
+      "loss": 0.4223,
+      "step": 3167
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7991287017595245e-05,
+      "loss": 0.4522,
+      "step": 3168
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7989878446510215e-05,
+      "loss": 0.4288,
+      "step": 3169
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.79884694369123e-05,
+      "loss": 0.4342,
+      "step": 3170
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7987059988878827e-05,
+      "loss": 0.4029,
+      "step": 3171
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7985650102487154e-05,
+      "loss": 0.4242,
+      "step": 3172
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7984239777814665e-05,
+      "loss": 0.4076,
+      "step": 3173
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7982829014938757e-05,
+      "loss": 0.4013,
+      "step": 3174
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7981417813936863e-05,
+      "loss": 0.4666,
+      "step": 3175
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7980006174886432e-05,
+      "loss": 0.4246,
+      "step": 3176
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797859409786494e-05,
+      "loss": 0.4423,
+      "step": 3177
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797718158294989e-05,
+      "loss": 0.4463,
+      "step": 3178
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.79757686302188e-05,
+      "loss": 0.4419,
+      "step": 3179
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797435523974922e-05,
+      "loss": 0.4469,
+      "step": 3180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7972941411618724e-05,
+      "loss": 0.4157,
+      "step": 3181
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7971527145904904e-05,
+      "loss": 0.4395,
+      "step": 3182
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7970112442685387e-05,
+      "loss": 0.4473,
+      "step": 3183
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796869730203781e-05,
+      "loss": 0.4226,
+      "step": 3184
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796728172403984e-05,
+      "loss": 0.4622,
+      "step": 3185
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796586570876918e-05,
+      "loss": 0.4186,
+      "step": 3186
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796444925630353e-05,
+      "loss": 0.4477,
+      "step": 3187
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7963032366720643e-05,
+      "loss": 0.4517,
+      "step": 3188
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796161504009828e-05,
+      "loss": 0.4028,
+      "step": 3189
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796019727651422e-05,
+      "loss": 0.4412,
+      "step": 3190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.795877907604629e-05,
+      "loss": 0.4273,
+      "step": 3191
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7957360438772312e-05,
+      "loss": 0.4549,
+      "step": 3192
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7955941364770155e-05,
+      "loss": 0.4323,
+      "step": 3193
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7954521854117697e-05,
+      "loss": 0.4223,
+      "step": 3194
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7953101906892852e-05,
+      "loss": 0.4119,
+      "step": 3195
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7951681523173543e-05,
+      "loss": 0.4386,
+      "step": 3196
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7950260703037733e-05,
+      "loss": 0.4299,
+      "step": 3197
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.79488394465634e-05,
+      "loss": 0.4481,
+      "step": 3198
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7947417753828545e-05,
+      "loss": 0.4452,
+      "step": 3199
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.79459956249112e-05,
+      "loss": 0.435,
+      "step": 3200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7944573059889414e-05,
+      "loss": 0.4349,
+      "step": 3201
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7943150058841262e-05,
+      "loss": 0.4603,
+      "step": 3202
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7941726621844843e-05,
+      "loss": 0.4317,
+      "step": 3203
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794030274897828e-05,
+      "loss": 0.4346,
+      "step": 3204
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7938878440319722e-05,
+      "loss": 0.4477,
+      "step": 3205
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793745369594734e-05,
+      "loss": 0.4577,
+      "step": 3206
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7936028515939327e-05,
+      "loss": 0.4591,
+      "step": 3207
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7934602900373903e-05,
+      "loss": 0.416,
+      "step": 3208
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7933176849329315e-05,
+      "loss": 0.4252,
+      "step": 3209
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7931750362883822e-05,
+      "loss": 0.4617,
+      "step": 3210
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793032344111572e-05,
+      "loss": 0.4458,
+      "step": 3211
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7928896084103324e-05,
+      "loss": 0.4379,
+      "step": 3212
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7927468291924962e-05,
+      "loss": 0.4179,
+      "step": 3213
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7926040064659012e-05,
+      "loss": 0.4298,
+      "step": 3214
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7924611402383853e-05,
+      "loss": 0.4329,
+      "step": 3215
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7923182305177896e-05,
+      "loss": 0.4389,
+      "step": 3216
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792175277311957e-05,
+      "loss": 0.4148,
+      "step": 3217
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792032280628734e-05,
+      "loss": 0.4295,
+      "step": 3218
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.791889240475968e-05,
+      "loss": 0.4433,
+      "step": 3219
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7917461568615104e-05,
+      "loss": 0.434,
+      "step": 3220
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7916030297932138e-05,
+      "loss": 0.4334,
+      "step": 3221
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.791459859278933e-05,
+      "loss": 0.4134,
+      "step": 3222
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7913166453265264e-05,
+      "loss": 0.4413,
+      "step": 3223
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.791173387943854e-05,
+      "loss": 0.4107,
+      "step": 3224
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7910300871387777e-05,
+      "loss": 0.4007,
+      "step": 3225
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7908867429191632e-05,
+      "loss": 0.4344,
+      "step": 3226
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790743355292877e-05,
+      "loss": 0.4169,
+      "step": 3227
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7905999242677895e-05,
+      "loss": 0.3989,
+      "step": 3228
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7904564498517717e-05,
+      "loss": 0.4348,
+      "step": 3229
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790312932052699e-05,
+      "loss": 0.4075,
+      "step": 3230
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7901693708784474e-05,
+      "loss": 0.4299,
+      "step": 3231
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790025766336896e-05,
+      "loss": 0.4279,
+      "step": 3232
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7898821184359272e-05,
+      "loss": 0.399,
+      "step": 3233
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789738427183424e-05,
+      "loss": 0.4501,
+      "step": 3234
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7895946925872733e-05,
+      "loss": 0.4437,
+      "step": 3235
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7894509146553635e-05,
+      "loss": 0.414,
+      "step": 3236
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7893070933955856e-05,
+      "loss": 0.428,
+      "step": 3237
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7891632288158333e-05,
+      "loss": 0.4149,
+      "step": 3238
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789019320924002e-05,
+      "loss": 0.4619,
+      "step": 3239
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.78887536972799e-05,
+      "loss": 0.4239,
+      "step": 3240
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788731375235698e-05,
+      "loss": 0.4311,
+      "step": 3241
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7885873374550285e-05,
+      "loss": 0.4251,
+      "step": 3242
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7884432563938875e-05,
+      "loss": 0.4416,
+      "step": 3243
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7882991320601825e-05,
+      "loss": 0.4167,
+      "step": 3244
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788154964461823e-05,
+      "loss": 0.4265,
+      "step": 3245
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788010753606722e-05,
+      "loss": 0.4164,
+      "step": 3246
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.787866499502794e-05,
+      "loss": 0.4574,
+      "step": 3247
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7877222021579566e-05,
+      "loss": 0.4224,
+      "step": 3248
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7875778615801286e-05,
+      "loss": 0.4054,
+      "step": 3249
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7874334777772326e-05,
+      "loss": 0.4296,
+      "step": 3250
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7872890507571926e-05,
+      "loss": 0.4338,
+      "step": 3251
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.787144580527936e-05,
+      "loss": 0.4456,
+      "step": 3252
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.78700006709739e-05,
+      "loss": 0.4244,
+      "step": 3253
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7868555104734876e-05,
+      "loss": 0.4143,
+      "step": 3254
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7867109106641626e-05,
+      "loss": 0.4305,
+      "step": 3255
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7865662676773504e-05,
+      "loss": 0.4421,
+      "step": 3256
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7864215815209894e-05,
+      "loss": 0.4524,
+      "step": 3257
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7862768522030217e-05,
+      "loss": 0.4329,
+      "step": 3258
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786132079731389e-05,
+      "loss": 0.4216,
+      "step": 3259
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7859872641140385e-05,
+      "loss": 0.4241,
+      "step": 3260
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7858424053589166e-05,
+      "loss": 0.4109,
+      "step": 3261
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785697503473975e-05,
+      "loss": 0.424,
+      "step": 3262
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7855525584671658e-05,
+      "loss": 0.4162,
+      "step": 3263
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785407570346444e-05,
+      "loss": 0.4338,
+      "step": 3264
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7852625391197675e-05,
+      "loss": 0.4336,
+      "step": 3265
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7851174647950957e-05,
+      "loss": 0.4312,
+      "step": 3266
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784972347380391e-05,
+      "loss": 0.4355,
+      "step": 3267
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784827186883618e-05,
+      "loss": 0.4295,
+      "step": 3268
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7846819833127436e-05,
+      "loss": 0.4229,
+      "step": 3269
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784536736675737e-05,
+      "loss": 0.4067,
+      "step": 3270
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7843914469805698e-05,
+      "loss": 0.4133,
+      "step": 3271
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784246114235216e-05,
+      "loss": 0.4657,
+      "step": 3272
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7841007384476522e-05,
+      "loss": 0.4264,
+      "step": 3273
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7839553196258573e-05,
+      "loss": 0.4119,
+      "step": 3274
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7838098577778116e-05,
+      "loss": 0.4577,
+      "step": 3275
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7836643529114993e-05,
+      "loss": 0.4403,
+      "step": 3276
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7835188050349064e-05,
+      "loss": 0.4301,
+      "step": 3277
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7833732141560204e-05,
+      "loss": 0.447,
+      "step": 3278
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7832275802828317e-05,
+      "loss": 0.4338,
+      "step": 3279
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7830819034233342e-05,
+      "loss": 0.4689,
+      "step": 3280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7829361835855222e-05,
+      "loss": 0.4209,
+      "step": 3281
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7827904207773943e-05,
+      "loss": 0.3967,
+      "step": 3282
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7826446150069493e-05,
+      "loss": 0.432,
+      "step": 3283
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7824987662821907e-05,
+      "loss": 0.4314,
+      "step": 3284
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7823528746111224e-05,
+      "loss": 0.4357,
+      "step": 3285
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7822069400017515e-05,
+      "loss": 0.4485,
+      "step": 3286
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782060962462088e-05,
+      "loss": 0.4248,
+      "step": 3287
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781914942000143e-05,
+      "loss": 0.4948,
+      "step": 3288
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7817688786239307e-05,
+      "loss": 0.4294,
+      "step": 3289
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781622772341468e-05,
+      "loss": 0.4364,
+      "step": 3290
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7814766231607734e-05,
+      "loss": 0.4059,
+      "step": 3291
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781330431089868e-05,
+      "loss": 0.4421,
+      "step": 3292
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781184196136776e-05,
+      "loss": 0.4335,
+      "step": 3293
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781037918309523e-05,
+      "loss": 0.4304,
+      "step": 3294
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7808915976161364e-05,
+      "loss": 0.4316,
+      "step": 3295
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7807452340646477e-05,
+      "loss": 0.4535,
+      "step": 3296
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7805988276630897e-05,
+      "loss": 0.4695,
+      "step": 3297
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7804523784194973e-05,
+      "loss": 0.4559,
+      "step": 3298
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780305886341909e-05,
+      "loss": 0.4176,
+      "step": 3299
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7801593514383637e-05,
+      "loss": 0.4254,
+      "step": 3300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7800127737169047e-05,
+      "loss": 0.4285,
+      "step": 3301
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.779866153185576e-05,
+      "loss": 0.4133,
+      "step": 3302
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7797194898524256e-05,
+      "loss": 0.4014,
+      "step": 3303
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7795727837255016e-05,
+      "loss": 0.431,
+      "step": 3304
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.779426034812857e-05,
+      "loss": 0.4358,
+      "step": 3305
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.779279243122545e-05,
+      "loss": 0.4407,
+      "step": 3306
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7791324086626228e-05,
+      "loss": 0.428,
+      "step": 3307
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7789855314411482e-05,
+      "loss": 0.4263,
+      "step": 3308
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7788386114661833e-05,
+      "loss": 0.4245,
+      "step": 3309
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778691648745791e-05,
+      "loss": 0.3967,
+      "step": 3310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778544643288038e-05,
+      "loss": 0.4473,
+      "step": 3311
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778397595100991e-05,
+      "loss": 0.4,
+      "step": 3312
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7782505041927218e-05,
+      "loss": 0.4566,
+      "step": 3313
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778103370571303e-05,
+      "loss": 0.4483,
+      "step": 3314
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7779561942448093e-05,
+      "loss": 0.4398,
+      "step": 3315
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7778089752213187e-05,
+      "loss": 0.4012,
+      "step": 3316
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7776617135089112e-05,
+      "loss": 0.4416,
+      "step": 3317
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7775144091156685e-05,
+      "loss": 0.3952,
+      "step": 3318
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.777367062049676e-05,
+      "loss": 0.4258,
+      "step": 3319
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77721967231902e-05,
+      "loss": 0.4654,
+      "step": 3320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77707223993179e-05,
+      "loss": 0.3984,
+      "step": 3321
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7769247648960775e-05,
+      "loss": 0.4339,
+      "step": 3322
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7767772472199763e-05,
+      "loss": 0.4696,
+      "step": 3323
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7766296869115833e-05,
+      "loss": 0.4269,
+      "step": 3324
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7764820839789966e-05,
+      "loss": 0.415,
+      "step": 3325
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7763344384303175e-05,
+      "loss": 0.4536,
+      "step": 3326
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7761867502736487e-05,
+      "loss": 0.4589,
+      "step": 3327
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7760390195170963e-05,
+      "loss": 0.4327,
+      "step": 3328
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7758912461687688e-05,
+      "loss": 0.4221,
+      "step": 3329
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7757434302367757e-05,
+      "loss": 0.4709,
+      "step": 3330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77559557172923e-05,
+      "loss": 0.4401,
+      "step": 3331
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7754476706542462e-05,
+      "loss": 0.42,
+      "step": 3332
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7752997270199427e-05,
+      "loss": 0.3894,
+      "step": 3333
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.775151740834438e-05,
+      "loss": 0.4059,
+      "step": 3334
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.775003712105855e-05,
+      "loss": 0.4425,
+      "step": 3335
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774855640842318e-05,
+      "loss": 0.4214,
+      "step": 3336
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774707527051953e-05,
+      "loss": 0.3855,
+      "step": 3337
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7745593707428893e-05,
+      "loss": 0.4305,
+      "step": 3338
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7744111719232588e-05,
+      "loss": 0.4163,
+      "step": 3339
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7742629306011944e-05,
+      "loss": 0.4215,
+      "step": 3340
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7741146467848326e-05,
+      "loss": 0.4375,
+      "step": 3341
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7739663204823115e-05,
+      "loss": 0.412,
+      "step": 3342
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7738179517017718e-05,
+      "loss": 0.4431,
+      "step": 3343
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7736695404513566e-05,
+      "loss": 0.3999,
+      "step": 3344
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7735210867392112e-05,
+      "loss": 0.4509,
+      "step": 3345
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7733725905734832e-05,
+      "loss": 0.4344,
+      "step": 3346
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7732240519623226e-05,
+      "loss": 0.4495,
+      "step": 3347
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773075470913882e-05,
+      "loss": 0.4696,
+      "step": 3348
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7729268474363156e-05,
+      "loss": 0.4018,
+      "step": 3349
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7727781815377806e-05,
+      "loss": 0.436,
+      "step": 3350
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7726294732264367e-05,
+      "loss": 0.4462,
+      "step": 3351
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7724807225104446e-05,
+      "loss": 0.4361,
+      "step": 3352
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7723319293979693e-05,
+      "loss": 0.4267,
+      "step": 3353
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7721830938971766e-05,
+      "loss": 0.4051,
+      "step": 3354
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7720342160162348e-05,
+      "loss": 0.4584,
+      "step": 3355
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7718852957633156e-05,
+      "loss": 0.4622,
+      "step": 3356
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7717363331465916e-05,
+      "loss": 0.4421,
+      "step": 3357
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.771587328174239e-05,
+      "loss": 0.4457,
+      "step": 3358
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.771438280854435e-05,
+      "loss": 0.3845,
+      "step": 3359
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7712891911953604e-05,
+      "loss": 0.4197,
+      "step": 3360
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7711400592051976e-05,
+      "loss": 0.4224,
+      "step": 3361
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7709908848921316e-05,
+      "loss": 0.4695,
+      "step": 3362
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770841668264349e-05,
+      "loss": 0.4677,
+      "step": 3363
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7706924093300405e-05,
+      "loss": 0.4251,
+      "step": 3364
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770543108097397e-05,
+      "loss": 0.4647,
+      "step": 3365
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7703937645746134e-05,
+      "loss": 0.4542,
+      "step": 3366
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7702443787698855e-05,
+      "loss": 0.4123,
+      "step": 3367
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7700949506914124e-05,
+      "loss": 0.4631,
+      "step": 3368
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769945480347395e-05,
+      "loss": 0.429,
+      "step": 3369
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7697959677460373e-05,
+      "loss": 0.4256,
+      "step": 3370
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7696464128955447e-05,
+      "loss": 0.4357,
+      "step": 3371
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7694968158041255e-05,
+      "loss": 0.4018,
+      "step": 3372
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7693471764799902e-05,
+      "loss": 0.4518,
+      "step": 3373
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7691974949313512e-05,
+      "loss": 0.4384,
+      "step": 3374
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7690477711664234e-05,
+      "loss": 0.4018,
+      "step": 3375
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768898005193425e-05,
+      "loss": 0.4085,
+      "step": 3376
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768748197020575e-05,
+      "loss": 0.4441,
+      "step": 3377
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7685983466560953e-05,
+      "loss": 0.4092,
+      "step": 3378
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768448454108211e-05,
+      "loss": 0.4317,
+      "step": 3379
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7682985193851478e-05,
+      "loss": 0.4347,
+      "step": 3380
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7681485424951353e-05,
+      "loss": 0.4337,
+      "step": 3381
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7679985234464045e-05,
+      "loss": 0.4167,
+      "step": 3382
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767848462247189e-05,
+      "loss": 0.4277,
+      "step": 3383
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767698358905725e-05,
+      "loss": 0.4368,
+      "step": 3384
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7675482134302503e-05,
+      "loss": 0.4697,
+      "step": 3385
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7673980258290052e-05,
+      "loss": 0.4166,
+      "step": 3386
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7672477961102333e-05,
+      "loss": 0.4122,
+      "step": 3387
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7670975242821787e-05,
+      "loss": 0.4307,
+      "step": 3388
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7669472103530903e-05,
+      "loss": 0.453,
+      "step": 3389
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7667968543312164e-05,
+      "loss": 0.466,
+      "step": 3390
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.76664645622481e-05,
+      "loss": 0.4162,
+      "step": 3391
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7664960160421252e-05,
+      "loss": 0.4354,
+      "step": 3392
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7663455337914184e-05,
+      "loss": 0.4475,
+      "step": 3393
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766195009480949e-05,
+      "loss": 0.4331,
+      "step": 3394
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.4538,
+      "step": 3395
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7658938347137695e-05,
+      "loss": 0.438,
+      "step": 3396
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765743184273589e-05,
+      "loss": 0.448,
+      "step": 3397
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765592491806705e-05,
+      "loss": 0.4303,
+      "step": 3398
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7654417573213875e-05,
+      "loss": 0.4404,
+      "step": 3399
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7652909808259102e-05,
+      "loss": 0.4404,
+      "step": 3400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7651401623285472e-05,
+      "loss": 0.4376,
+      "step": 3401
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7649893018375773e-05,
+      "loss": 0.4603,
+      "step": 3402
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764838399361279e-05,
+      "loss": 0.4278,
+      "step": 3403
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7646874549079348e-05,
+      "loss": 0.4457,
+      "step": 3404
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7645364684858295e-05,
+      "loss": 0.4282,
+      "step": 3405
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764385440103249e-05,
+      "loss": 0.4321,
+      "step": 3406
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7642343697684826e-05,
+      "loss": 0.4187,
+      "step": 3407
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764083257489822e-05,
+      "loss": 0.4173,
+      "step": 3408
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7639321032755604e-05,
+      "loss": 0.4398,
+      "step": 3409
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7637809071339934e-05,
+      "loss": 0.4302,
+      "step": 3410
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7636296690734198e-05,
+      "loss": 0.4136,
+      "step": 3411
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7634783891021394e-05,
+      "loss": 0.4068,
+      "step": 3412
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7633270672284555e-05,
+      "loss": 0.437,
+      "step": 3413
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7631757034606732e-05,
+      "loss": 0.4232,
+      "step": 3414
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7630242978071e-05,
+      "loss": 0.4279,
+      "step": 3415
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7628728502760444e-05,
+      "loss": 0.4422,
+      "step": 3416
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7627213608758197e-05,
+      "loss": 0.4169,
+      "step": 3417
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.76256982961474e-05,
+      "loss": 0.433,
+      "step": 3418
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762418256501121e-05,
+      "loss": 0.4227,
+      "step": 3419
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762266641543283e-05,
+      "loss": 0.4158,
+      "step": 3420
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7621149847495458e-05,
+      "loss": 0.4278,
+      "step": 3421
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7619632861282333e-05,
+      "loss": 0.436,
+      "step": 3422
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7618115456876718e-05,
+      "loss": 0.4085,
+      "step": 3423
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761659763436189e-05,
+      "loss": 0.4542,
+      "step": 3424
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7615079393821148e-05,
+      "loss": 0.4464,
+      "step": 3425
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7613560735337824e-05,
+      "loss": 0.4079,
+      "step": 3426
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7612041658995264e-05,
+      "loss": 0.3922,
+      "step": 3427
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761052216487684e-05,
+      "loss": 0.4452,
+      "step": 3428
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7609002253065956e-05,
+      "loss": 0.4262,
+      "step": 3429
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760748192364602e-05,
+      "loss": 0.43,
+      "step": 3430
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7605961176700472e-05,
+      "loss": 0.4253,
+      "step": 3431
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7604440012312786e-05,
+      "loss": 0.4013,
+      "step": 3432
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7602918430566437e-05,
+      "loss": 0.4274,
+      "step": 3433
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760139643154495e-05,
+      "loss": 0.4297,
+      "step": 3434
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759987401533184e-05,
+      "loss": 0.4138,
+      "step": 3435
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7598351182010677e-05,
+      "loss": 0.4156,
+      "step": 3436
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759682793166503e-05,
+      "loss": 0.4338,
+      "step": 3437
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7595304264378507e-05,
+      "loss": 0.4171,
+      "step": 3438
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759378018023473e-05,
+      "loss": 0.42,
+      "step": 3439
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759225567931734e-05,
+      "loss": 0.4621,
+      "step": 3440
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759073076171002e-05,
+      "loss": 0.4416,
+      "step": 3441
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7589205427496452e-05,
+      "loss": 0.4516,
+      "step": 3442
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7587679676760355e-05,
+      "loss": 0.4553,
+      "step": 3443
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758615350958547e-05,
+      "loss": 0.427,
+      "step": 3444
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7584626926055553e-05,
+      "loss": 0.4187,
+      "step": 3445
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7583099926254396e-05,
+      "loss": 0.4657,
+      "step": 3446
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7581572510265798e-05,
+      "loss": 0.4498,
+      "step": 3447
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7580044678173593e-05,
+      "loss": 0.4372,
+      "step": 3448
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7578516430061638e-05,
+      "loss": 0.4481,
+      "step": 3449
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.75769877660138e-05,
+      "loss": 0.4539,
+      "step": 3450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7575458686113983e-05,
+      "loss": 0.4239,
+      "step": 3451
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757392919044611e-05,
+      "loss": 0.4319,
+      "step": 3452
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757239927909412e-05,
+      "loss": 0.393,
+      "step": 3453
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757086895214198e-05,
+      "loss": 0.4416,
+      "step": 3454
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7569338209673684e-05,
+      "loss": 0.4272,
+      "step": 3455
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7567807051773246e-05,
+      "loss": 0.431,
+      "step": 3456
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7566275478524694e-05,
+      "loss": 0.4294,
+      "step": 3457
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7564743490012093e-05,
+      "loss": 0.4369,
+      "step": 3458
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7563211086319517e-05,
+      "loss": 0.4033,
+      "step": 3459
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756167826753108e-05,
+      "loss": 0.4447,
+      "step": 3460
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7560145033730896e-05,
+      "loss": 0.4287,
+      "step": 3461
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7558611385003127e-05,
+      "loss": 0.4221,
+      "step": 3462
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7557077321431936e-05,
+      "loss": 0.4305,
+      "step": 3463
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755554284310152e-05,
+      "loss": 0.4507,
+      "step": 3464
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7554007950096095e-05,
+      "loss": 0.4291,
+      "step": 3465
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755247264249991e-05,
+      "loss": 0.4203,
+      "step": 3466
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7550936920397215e-05,
+      "loss": 0.3996,
+      "step": 3467
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.754940078387231e-05,
+      "loss": 0.4931,
+      "step": 3468
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7547864233009494e-05,
+      "loss": 0.4261,
+      "step": 3469
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7546327267893098e-05,
+      "loss": 0.4459,
+      "step": 3470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7544789888607484e-05,
+      "loss": 0.425,
+      "step": 3471
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7543252095237017e-05,
+      "loss": 0.4141,
+      "step": 3472
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754171388786611e-05,
+      "loss": 0.5001,
+      "step": 3473
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7540175266579175e-05,
+      "loss": 0.4147,
+      "step": 3474
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753863623146066e-05,
+      "loss": 0.4222,
+      "step": 3475
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7537096782595032e-05,
+      "loss": 0.4258,
+      "step": 3476
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7535556920066786e-05,
+      "loss": 0.4107,
+      "step": 3477
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753401664396043e-05,
+      "loss": 0.447,
+      "step": 3478
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7532475954360502e-05,
+      "loss": 0.4022,
+      "step": 3479
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753093485135156e-05,
+      "loss": 0.4193,
+      "step": 3480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7529393335018186e-05,
+      "loss": 0.4301,
+      "step": 3481
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7527851405444983e-05,
+      "loss": 0.4127,
+      "step": 3482
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752630906271658e-05,
+      "loss": 0.4213,
+      "step": 3483
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7524766306917617e-05,
+      "loss": 0.4717,
+      "step": 3484
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7523223138132776e-05,
+      "loss": 0.4333,
+      "step": 3485
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7521679556446753e-05,
+      "loss": 0.454,
+      "step": 3486
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7520135561944257e-05,
+      "loss": 0.468,
+      "step": 3487
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7518591154710033e-05,
+      "loss": 0.4525,
+      "step": 3488
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751704633482884e-05,
+      "loss": 0.4482,
+      "step": 3489
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7515501102385468e-05,
+      "loss": 0.4384,
+      "step": 3490
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751395545746472e-05,
+      "loss": 0.4417,
+      "step": 3491
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7512409400151432e-05,
+      "loss": 0.4409,
+      "step": 3492
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751086293053045e-05,
+      "loss": 0.4036,
+      "step": 3493
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7509316048686657e-05,
+      "loss": 0.4227,
+      "step": 3494
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7507768754704944e-05,
+      "loss": 0.4193,
+      "step": 3495
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.750622104867024e-05,
+      "loss": 0.4384,
+      "step": 3496
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7504672930667478e-05,
+      "loss": 0.4353,
+      "step": 3497
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.750312440078164e-05,
+      "loss": 0.424,
+      "step": 3498
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.75015754590977e-05,
+      "loss": 0.4066,
+      "step": 3499
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7500026105700675e-05,
+      "loss": 0.4029,
+      "step": 3500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7498476340675597e-05,
+      "loss": 0.4646,
+      "step": 3501
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749692616410753e-05,
+      "loss": 0.4485,
+      "step": 3502
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7495375576081544e-05,
+      "loss": 0.4149,
+      "step": 3503
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749382457668275e-05,
+      "loss": 0.4161,
+      "step": 3504
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749227316599626e-05,
+      "loss": 0.454,
+      "step": 3505
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7490721344107232e-05,
+      "loss": 0.4326,
+      "step": 3506
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7489169111100826e-05,
+      "loss": 0.4587,
+      "step": 3507
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7487616467062243e-05,
+      "loss": 0.416,
+      "step": 3508
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7486063412076695e-05,
+      "loss": 0.4134,
+      "step": 3509
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748450994622942e-05,
+      "loss": 0.4354,
+      "step": 3510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748295606960567e-05,
+      "loss": 0.4876,
+      "step": 3511
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7481401782290737e-05,
+      "loss": 0.4664,
+      "step": 3512
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747984708436992e-05,
+      "loss": 0.4664,
+      "step": 3513
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747829197592855e-05,
+      "loss": 0.4152,
+      "step": 3514
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7476736457051975e-05,
+      "loss": 0.4458,
+      "step": 3515
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7475180527825566e-05,
+      "loss": 0.4683,
+      "step": 3516
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747362418833472e-05,
+      "loss": 0.3688,
+      "step": 3517
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7472067438664856e-05,
+      "loss": 0.4103,
+      "step": 3518
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747051027890141e-05,
+      "loss": 0.411,
+      "step": 3519
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7468952709129848e-05,
+      "loss": 0.4394,
+      "step": 3520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746739472943565e-05,
+      "loss": 0.4171,
+      "step": 3521
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746583633990433e-05,
+      "loss": 0.4126,
+      "step": 3522
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7464277540621413e-05,
+      "loss": 0.4084,
+      "step": 3523
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7462718331672455e-05,
+      "loss": 0.4522,
+      "step": 3524
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746115871314303e-05,
+      "loss": 0.4187,
+      "step": 3525
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745959868511873e-05,
+      "loss": 0.4293,
+      "step": 3526
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7458038247685182e-05,
+      "loss": 0.4552,
+      "step": 3527
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745647740092802e-05,
+      "loss": 0.4284,
+      "step": 3528
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7454916144932923e-05,
+      "loss": 0.4588,
+      "step": 3529
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7453354479785566e-05,
+      "loss": 0.4288,
+      "step": 3530
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745179240557166e-05,
+      "loss": 0.4498,
+      "step": 3531
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7450229922376945e-05,
+      "loss": 0.4459,
+      "step": 3532
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7448667030287165e-05,
+      "loss": 0.4064,
+      "step": 3533
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7447103729388107e-05,
+      "loss": 0.4788,
+      "step": 3534
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744554001976556e-05,
+      "loss": 0.4089,
+      "step": 3535
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7443975901505358e-05,
+      "loss": 0.4171,
+      "step": 3536
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7442411374693336e-05,
+      "loss": 0.4206,
+      "step": 3537
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7440846439415363e-05,
+      "loss": 0.4715,
+      "step": 3538
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743928109575733e-05,
+      "loss": 0.4307,
+      "step": 3539
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743771534380515e-05,
+      "loss": 0.3987,
+      "step": 3540
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7436149183644754e-05,
+      "loss": 0.4381,
+      "step": 3541
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7434582615362096e-05,
+      "loss": 0.4126,
+      "step": 3542
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7433015639043162e-05,
+      "loss": 0.431,
+      "step": 3543
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7431448254773943e-05,
+      "loss": 0.4277,
+      "step": 3544
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7429880462640474e-05,
+      "loss": 0.4242,
+      "step": 3545
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7428312262728793e-05,
+      "loss": 0.4262,
+      "step": 3546
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7426743655124973e-05,
+      "loss": 0.4584,
+      "step": 3547
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7425174639915105e-05,
+      "loss": 0.4528,
+      "step": 3548
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7423605217185296e-05,
+      "loss": 0.4484,
+      "step": 3549
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742203538702169e-05,
+      "loss": 0.4045,
+      "step": 3550
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7420465149510437e-05,
+      "loss": 0.4276,
+      "step": 3551
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.741889450473772e-05,
+      "loss": 0.3961,
+      "step": 3552
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7417323452789746e-05,
+      "loss": 0.42,
+      "step": 3553
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7415751993752734e-05,
+      "loss": 0.4146,
+      "step": 3554
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7414180127712933e-05,
+      "loss": 0.4412,
+      "step": 3555
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7412607854756613e-05,
+      "loss": 0.4178,
+      "step": 3556
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7411035174970067e-05,
+      "loss": 0.4265,
+      "step": 3557
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7409462088439607e-05,
+      "loss": 0.4124,
+      "step": 3558
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7407888595251567e-05,
+      "loss": 0.4305,
+      "step": 3559
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7406314695492313e-05,
+      "loss": 0.4619,
+      "step": 3560
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7404740389248223e-05,
+      "loss": 0.4457,
+      "step": 3561
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.74031656766057e-05,
+      "loss": 0.4494,
+      "step": 3562
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7401590557651173e-05,
+      "loss": 0.4424,
+      "step": 3563
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7400015032471086e-05,
+      "loss": 0.4581,
+      "step": 3564
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7398439101151908e-05,
+      "loss": 0.4602,
+      "step": 3565
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7396862763780134e-05,
+      "loss": 0.4147,
+      "step": 3566
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739528602044228e-05,
+      "loss": 0.4388,
+      "step": 3567
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7393708871224887e-05,
+      "loss": 0.427,
+      "step": 3568
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7392131316214503e-05,
+      "loss": 0.4142,
+      "step": 3569
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7390553355497723e-05,
+      "loss": 0.4224,
+      "step": 3570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738897498916114e-05,
+      "loss": 0.4127,
+      "step": 3571
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738739621729139e-05,
+      "loss": 0.4368,
+      "step": 3572
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7385817039975115e-05,
+      "loss": 0.4599,
+      "step": 3573
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7384237457298987e-05,
+      "loss": 0.4356,
+      "step": 3574
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7382657469349702e-05,
+      "loss": 0.42,
+      "step": 3575
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738107707621397e-05,
+      "loss": 0.4351,
+      "step": 3576
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7379496277978535e-05,
+      "loss": 0.4272,
+      "step": 3577
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7377915074730152e-05,
+      "loss": 0.4615,
+      "step": 3578
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7376333466555605e-05,
+      "loss": 0.426,
+      "step": 3579
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.73747514535417e-05,
+      "loss": 0.4425,
+      "step": 3580
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7373169035775257e-05,
+      "loss": 0.4093,
+      "step": 3581
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7371586213343134e-05,
+      "loss": 0.437,
+      "step": 3582
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7370002986332195e-05,
+      "loss": 0.4137,
+      "step": 3583
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7368419354829335e-05,
+      "loss": 0.4523,
+      "step": 3584
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736683531892147e-05,
+      "loss": 0.46,
+      "step": 3585
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7365250878695538e-05,
+      "loss": 0.4094,
+      "step": 3586
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7363666034238496e-05,
+      "loss": 0.4092,
+      "step": 3587
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736208078563733e-05,
+      "loss": 0.4066,
+      "step": 3588
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7360495132979037e-05,
+      "loss": 0.4171,
+      "step": 3589
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735890907635065e-05,
+      "loss": 0.4025,
+      "step": 3590
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735732261583922e-05,
+      "loss": 0.4318,
+      "step": 3591
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7355735751531808e-05,
+      "loss": 0.4718,
+      "step": 3592
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7354148483515515e-05,
+      "loss": 0.4113,
+      "step": 3593
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735256081187745e-05,
+      "loss": 0.4723,
+      "step": 3594
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735097273670475e-05,
+      "loss": 0.4571,
+      "step": 3595
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7349384258084583e-05,
+      "loss": 0.4501,
+      "step": 3596
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734779537610412e-05,
+      "loss": 0.3987,
+      "step": 3597
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7346206090850572e-05,
+      "loss": 0.4332,
+      "step": 3598
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7344616402411157e-05,
+      "loss": 0.4278,
+      "step": 3599
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734302631087313e-05,
+      "loss": 0.4401,
+      "step": 3600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7341435816323756e-05,
+      "loss": 0.4283,
+      "step": 3601
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7339844918850333e-05,
+      "loss": 0.4146,
+      "step": 3602
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7338253618540166e-05,
+      "loss": 0.449,
+      "step": 3603
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7336661915480602e-05,
+      "loss": 0.4561,
+      "step": 3604
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733506980975899e-05,
+      "loss": 0.4029,
+      "step": 3605
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7333477301462712e-05,
+      "loss": 0.4175,
+      "step": 3606
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7331884390679173e-05,
+      "loss": 0.409,
+      "step": 3607
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.73302910774958e-05,
+      "loss": 0.436,
+      "step": 3608
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732869736200004e-05,
+      "loss": 0.4191,
+      "step": 3609
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732710324427935e-05,
+      "loss": 0.4364,
+      "step": 3610
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7325508724421232e-05,
+      "loss": 0.429,
+      "step": 3611
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7323913802513198e-05,
+      "loss": 0.4305,
+      "step": 3612
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732231847864278e-05,
+      "loss": 0.4611,
+      "step": 3613
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732072275289754e-05,
+      "loss": 0.4443,
+      "step": 3614
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7319126625365048e-05,
+      "loss": 0.4554,
+      "step": 3615
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7317530096132914e-05,
+      "loss": 0.4174,
+      "step": 3616
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7315933165288754e-05,
+      "loss": 0.4274,
+      "step": 3617
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731433583292022e-05,
+      "loss": 0.4597,
+      "step": 3618
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7312738099114973e-05,
+      "loss": 0.4543,
+      "step": 3619
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731113996396071e-05,
+      "loss": 0.4277,
+      "step": 3620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7309541427545135e-05,
+      "loss": 0.4372,
+      "step": 3621
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7307942489955985e-05,
+      "loss": 0.4457,
+      "step": 3622
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7306343151281014e-05,
+      "loss": 0.4373,
+      "step": 3623
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7304743411608e-05,
+      "loss": 0.4413,
+      "step": 3624
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7303143271024744e-05,
+      "loss": 0.4296,
+      "step": 3625
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7301542729619065e-05,
+      "loss": 0.4635,
+      "step": 3626
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7299941787478805e-05,
+      "loss": 0.4596,
+      "step": 3627
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7298340444691836e-05,
+      "loss": 0.4204,
+      "step": 3628
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7296738701346035e-05,
+      "loss": 0.4549,
+      "step": 3629
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729513655752932e-05,
+      "loss": 0.4227,
+      "step": 3630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7293534013329618e-05,
+      "loss": 0.4526,
+      "step": 3631
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729193106883489e-05,
+      "loss": 0.4209,
+      "step": 3632
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7290327724133098e-05,
+      "loss": 0.4312,
+      "step": 3633
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.728872397931225e-05,
+      "loss": 0.4059,
+      "step": 3634
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.728711983446036e-05,
+      "loss": 0.4287,
+      "step": 3635
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.728551528966547e-05,
+      "loss": 0.4267,
+      "step": 3636
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7283910345015645e-05,
+      "loss": 0.4426,
+      "step": 3637
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.728230500059897e-05,
+      "loss": 0.4364,
+      "step": 3638
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.728069925650355e-05,
+      "loss": 0.4489,
+      "step": 3639
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7279093112817517e-05,
+      "loss": 0.4188,
+      "step": 3640
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7277486569629017e-05,
+      "loss": 0.4385,
+      "step": 3641
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7275879627026223e-05,
+      "loss": 0.4485,
+      "step": 3642
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7274272285097336e-05,
+      "loss": 0.4424,
+      "step": 3643
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727266454393057e-05,
+      "loss": 0.4509,
+      "step": 3644
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7271056403614158e-05,
+      "loss": 0.4442,
+      "step": 3645
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726944786423637e-05,
+      "loss": 0.405,
+      "step": 3646
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7267838925885482e-05,
+      "loss": 0.4093,
+      "step": 3647
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7266229588649796e-05,
+      "loss": 0.4324,
+      "step": 3648
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7264619852617645e-05,
+      "loss": 0.4297,
+      "step": 3649
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7263009717877372e-05,
+      "loss": 0.4673,
+      "step": 3650
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726139918451735e-05,
+      "loss": 0.4548,
+      "step": 3651
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7259788252625968e-05,
+      "loss": 0.4484,
+      "step": 3652
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725817692229164e-05,
+      "loss": 0.4455,
+      "step": 3653
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7256565193602805e-05,
+      "loss": 0.3983,
+      "step": 3654
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7254953066647915e-05,
+      "loss": 0.4373,
+      "step": 3655
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7253340541515453e-05,
+      "loss": 0.4359,
+      "step": 3656
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7251727618293923e-05,
+      "loss": 0.4245,
+      "step": 3657
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725011429707184e-05,
+      "loss": 0.4251,
+      "step": 3658
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7248500577937755e-05,
+      "loss": 0.419,
+      "step": 3659
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7246886460980234e-05,
+      "loss": 0.4361,
+      "step": 3660
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7245271946287858e-05,
+      "loss": 0.4589,
+      "step": 3661
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7243657033949248e-05,
+      "loss": 0.3981,
+      "step": 3662
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724204172405303e-05,
+      "loss": 0.4673,
+      "step": 3663
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7240426016687863e-05,
+      "loss": 0.4134,
+      "step": 3664
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7238809911942417e-05,
+      "loss": 0.4499,
+      "step": 3665
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7237193409905392e-05,
+      "loss": 0.4368,
+      "step": 3666
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7235576510665507e-05,
+      "loss": 0.4206,
+      "step": 3667
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7233959214311498e-05,
+      "loss": 0.427,
+      "step": 3668
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723234152093214e-05,
+      "loss": 0.4325,
+      "step": 3669
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723072343061621e-05,
+      "loss": 0.4647,
+      "step": 3670
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7229104943452512e-05,
+      "loss": 0.4242,
+      "step": 3671
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722748605952988e-05,
+      "loss": 0.4137,
+      "step": 3672
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7225866778937168e-05,
+      "loss": 0.4324,
+      "step": 3673
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7224247101763235e-05,
+      "loss": 0.4085,
+      "step": 3674
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722262702809698e-05,
+      "loss": 0.4374,
+      "step": 3675
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7221006558027325e-05,
+      "loss": 0.4038,
+      "step": 3676
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.72193856916432e-05,
+      "loss": 0.4341,
+      "step": 3677
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7217764429033566e-05,
+      "loss": 0.3959,
+      "step": 3678
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7216142770287403e-05,
+      "loss": 0.463,
+      "step": 3679
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7214520715493712e-05,
+      "loss": 0.4148,
+      "step": 3680
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7212898264741524e-05,
+      "loss": 0.409,
+      "step": 3681
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7211275418119875e-05,
+      "loss": 0.4237,
+      "step": 3682
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720965217571784e-05,
+      "loss": 0.432,
+      "step": 3683
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7208028537624505e-05,
+      "loss": 0.4302,
+      "step": 3684
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7206404503928983e-05,
+      "loss": 0.4666,
+      "step": 3685
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7204780074720405e-05,
+      "loss": 0.4345,
+      "step": 3686
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7203155250087926e-05,
+      "loss": 0.4148,
+      "step": 3687
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7201530030120722e-05,
+      "loss": 0.4336,
+      "step": 3688
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7199904414907992e-05,
+      "loss": 0.3863,
+      "step": 3689
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7198278404538952e-05,
+      "loss": 0.4637,
+      "step": 3690
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.719665199910285e-05,
+      "loss": 0.4241,
+      "step": 3691
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.719502519868894e-05,
+      "loss": 0.4155,
+      "step": 3692
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7193398003386514e-05,
+      "loss": 0.4246,
+      "step": 3693
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7191770413284873e-05,
+      "loss": 0.4167,
+      "step": 3694
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.719014242847335e-05,
+      "loss": 0.399,
+      "step": 3695
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718851404904129e-05,
+      "loss": 0.4328,
+      "step": 3696
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7186885275078068e-05,
+      "loss": 0.4839,
+      "step": 3697
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7185256106673076e-05,
+      "loss": 0.4836,
+      "step": 3698
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7183626543915726e-05,
+      "loss": 0.4326,
+      "step": 3699
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7181996586895456e-05,
+      "loss": 0.5219,
+      "step": 3700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7180366235701723e-05,
+      "loss": 0.4655,
+      "step": 3701
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7178735490424007e-05,
+      "loss": 0.399,
+      "step": 3702
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7177104351151812e-05,
+      "loss": 0.4205,
+      "step": 3703
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7175472817974654e-05,
+      "loss": 0.4347,
+      "step": 3704
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7173840890982086e-05,
+      "loss": 0.4507,
+      "step": 3705
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7172208570263664e-05,
+      "loss": 0.4341,
+      "step": 3706
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717057585590898e-05,
+      "loss": 0.3997,
+      "step": 3707
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716894274800765e-05,
+      "loss": 0.4286,
+      "step": 3708
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7167309246649298e-05,
+      "loss": 0.4378,
+      "step": 3709
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716567535192357e-05,
+      "loss": 0.4376,
+      "step": 3710
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7164041063920155e-05,
+      "loss": 0.4265,
+      "step": 3711
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7162406382728737e-05,
+      "loss": 0.4398,
+      "step": 3712
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716077130843904e-05,
+      "loss": 0.4311,
+      "step": 3713
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7159135841140795e-05,
+      "loss": 0.4485,
+      "step": 3714
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7157499980923767e-05,
+      "loss": 0.4009,
+      "step": 3715
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7155863727877738e-05,
+      "loss": 0.4194,
+      "step": 3716
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7154227082092512e-05,
+      "loss": 0.4195,
+      "step": 3717
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715259004365791e-05,
+      "loss": 0.4467,
+      "step": 3718
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7150952612663784e-05,
+      "loss": 0.4399,
+      "step": 3719
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7149314789199998e-05,
+      "loss": 0.4236,
+      "step": 3720
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7147676573356446e-05,
+      "loss": 0.4255,
+      "step": 3721
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.714603796522303e-05,
+      "loss": 0.4613,
+      "step": 3722
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.714439896488969e-05,
+      "loss": 0.4307,
+      "step": 3723
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7142759572446383e-05,
+      "loss": 0.3927,
+      "step": 3724
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7141119787983075e-05,
+      "loss": 0.4224,
+      "step": 3725
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7139479611589772e-05,
+      "loss": 0.4305,
+      "step": 3726
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7137839043356486e-05,
+      "loss": 0.4541,
+      "step": 3727
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713619808337326e-05,
+      "loss": 0.4713,
+      "step": 3728
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713455673173016e-05,
+      "loss": 0.4341,
+      "step": 3729
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7132914988517263e-05,
+      "loss": 0.4411,
+      "step": 3730
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7131272853824678e-05,
+      "loss": 0.4079,
+      "step": 3731
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7129630327742524e-05,
+      "loss": 0.4236,
+      "step": 3732
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712798741036096e-05,
+      "loss": 0.4522,
+      "step": 3733
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7126344101770147e-05,
+      "loss": 0.4319,
+      "step": 3734
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7124700402060278e-05,
+      "loss": 0.448,
+      "step": 3735
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712305631132156e-05,
+      "loss": 0.3875,
+      "step": 3736
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7121411829644237e-05,
+      "loss": 0.4304,
+      "step": 3737
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711976695711856e-05,
+      "loss": 0.4289,
+      "step": 3738
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7118121693834796e-05,
+      "loss": 0.4443,
+      "step": 3739
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7116476039883253e-05,
+      "loss": 0.4613,
+      "step": 3740
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711482999535425e-05,
+      "loss": 0.3937,
+      "step": 3741
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7113183560338123e-05,
+      "loss": 0.4448,
+      "step": 3742
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711153673492524e-05,
+      "loss": 0.4367,
+      "step": 3743
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.710988951920598e-05,
+      "loss": 0.4127,
+      "step": 3744
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.710824191327075e-05,
+      "loss": 0.4556,
+      "step": 3745
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7106593917209973e-05,
+      "loss": 0.4196,
+      "step": 3746
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7104945531114103e-05,
+      "loss": 0.4315,
+      "step": 3747
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7103296755073608e-05,
+      "loss": 0.4098,
+      "step": 3748
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7101647589178972e-05,
+      "loss": 0.4517,
+      "step": 3749
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7099998033520716e-05,
+      "loss": 0.4495,
+      "step": 3750
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7098348088189364e-05,
+      "loss": 0.454,
+      "step": 3751
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709669775327548e-05,
+      "loss": 0.4324,
+      "step": 3752
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7095047028869634e-05,
+      "loss": 0.4301,
+      "step": 3753
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709339591506243e-05,
+      "loss": 0.4531,
+      "step": 3754
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709174441194448e-05,
+      "loss": 0.4269,
+      "step": 3755
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7090092519606428e-05,
+      "loss": 0.4412,
+      "step": 3756
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7088440238138936e-05,
+      "loss": 0.4467,
+      "step": 3757
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7086787567632687e-05,
+      "loss": 0.4425,
+      "step": 3758
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708513450817838e-05,
+      "loss": 0.4454,
+      "step": 3759
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708348105986675e-05,
+      "loss": 0.4223,
+      "step": 3760
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7081827222788537e-05,
+      "loss": 0.4499,
+      "step": 3761
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7080172997034513e-05,
+      "loss": 0.4407,
+      "step": 3762
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7078518382695467e-05,
+      "loss": 0.4373,
+      "step": 3763
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7076863379862206e-05,
+      "loss": 0.4183,
+      "step": 3764
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707520798862557e-05,
+      "loss": 0.4573,
+      "step": 3765
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707355220907641e-05,
+      "loss": 0.4339,
+      "step": 3766
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7071896041305596e-05,
+      "loss": 0.4332,
+      "step": 3767
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7070239485404033e-05,
+      "loss": 0.4391,
+      "step": 3768
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7068582541462632e-05,
+      "loss": 0.4356,
+      "step": 3769
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7066925209572332e-05,
+      "loss": 0.4539,
+      "step": 3770
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7065267489824097e-05,
+      "loss": 0.467,
+      "step": 3771
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706360938230891e-05,
+      "loss": 0.4605,
+      "step": 3772
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7061950887117765e-05,
+      "loss": 0.4128,
+      "step": 3773
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7060292004341696e-05,
+      "loss": 0.4298,
+      "step": 3774
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.705863273407174e-05,
+      "loss": 0.439,
+      "step": 3775
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.705697307639897e-05,
+      "loss": 0.4537,
+      "step": 3776
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7055313031414475e-05,
+      "loss": 0.4183,
+      "step": 3777
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7053652599209353e-05,
+      "loss": 0.4131,
+      "step": 3778
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7051991779874748e-05,
+      "loss": 0.4524,
+      "step": 3779
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7050330573501804e-05,
+      "loss": 0.4471,
+      "step": 3780
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7048668980181696e-05,
+      "loss": 0.4309,
+      "step": 3781
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704700700000562e-05,
+      "loss": 0.4208,
+      "step": 3782
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704534463306479e-05,
+      "loss": 0.4122,
+      "step": 3783
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704368187945044e-05,
+      "loss": 0.4536,
+      "step": 3784
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704201873925383e-05,
+      "loss": 0.4254,
+      "step": 3785
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7040355212566232e-05,
+      "loss": 0.4553,
+      "step": 3786
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7038691299478963e-05,
+      "loss": 0.4611,
+      "step": 3787
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7037027000083328e-05,
+      "loss": 0.4167,
+      "step": 3788
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703536231447068e-05,
+      "loss": 0.4359,
+      "step": 3789
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7033697242732376e-05,
+      "loss": 0.4454,
+      "step": 3790
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7032031784959805e-05,
+      "loss": 0.4254,
+      "step": 3791
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7030365941244373e-05,
+      "loss": 0.4317,
+      "step": 3792
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7028699711677508e-05,
+      "loss": 0.4156,
+      "step": 3793
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7027033096350657e-05,
+      "loss": 0.4484,
+      "step": 3794
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7025366095355288e-05,
+      "loss": 0.4345,
+      "step": 3795
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7023698708782895e-05,
+      "loss": 0.4195,
+      "step": 3796
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7022030936724988e-05,
+      "loss": 0.4645,
+      "step": 3797
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7020362779273103e-05,
+      "loss": 0.4756,
+      "step": 3798
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7018694236518792e-05,
+      "loss": 0.398,
+      "step": 3799
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7017025308553633e-05,
+      "loss": 0.4431,
+      "step": 3800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7015355995469217e-05,
+      "loss": 0.4458,
+      "step": 3801
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.701368629735717e-05,
+      "loss": 0.421,
+      "step": 3802
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7012016214309123e-05,
+      "loss": 0.4327,
+      "step": 3803
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.701034574641674e-05,
+      "loss": 0.4235,
+      "step": 3804
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7008674893771708e-05,
+      "loss": 0.4391,
+      "step": 3805
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7007003656465716e-05,
+      "loss": 0.4308,
+      "step": 3806
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7005332034590497e-05,
+      "loss": 0.4263,
+      "step": 3807
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7003660028237793e-05,
+      "loss": 0.4182,
+      "step": 3808
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700198763749937e-05,
+      "loss": 0.4375,
+      "step": 3809
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7000314862467018e-05,
+      "loss": 0.4396,
+      "step": 3810
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6998641703232535e-05,
+      "loss": 0.418,
+      "step": 3811
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6996968159887765e-05,
+      "loss": 0.4028,
+      "step": 3812
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6995294232524544e-05,
+      "loss": 0.4156,
+      "step": 3813
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699361992123475e-05,
+      "loss": 0.4246,
+      "step": 3814
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6991945226110274e-05,
+      "loss": 0.4609,
+      "step": 3815
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6990270147243027e-05,
+      "loss": 0.4665,
+      "step": 3816
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6988594684724947e-05,
+      "loss": 0.4262,
+      "step": 3817
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6986918838647986e-05,
+      "loss": 0.4321,
+      "step": 3818
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698524260910412e-05,
+      "loss": 0.4323,
+      "step": 3819
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698356599618535e-05,
+      "loss": 0.4214,
+      "step": 3820
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6981888999983697e-05,
+      "loss": 0.4024,
+      "step": 3821
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698021162059119e-05,
+      "loss": 0.4201,
+      "step": 3822
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.69785338580999e-05,
+      "loss": 0.4696,
+      "step": 3823
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6976855712601904e-05,
+      "loss": 0.4206,
+      "step": 3824
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6975177184189303e-05,
+      "loss": 0.4408,
+      "step": 3825
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6973498272954223e-05,
+      "loss": 0.4192,
+      "step": 3826
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6971818978988807e-05,
+      "loss": 0.4383,
+      "step": 3827
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6970139302385225e-05,
+      "loss": 0.4532,
+      "step": 3828
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.696845924323566e-05,
+      "loss": 0.4469,
+      "step": 3829
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6966778801632318e-05,
+      "loss": 0.3828,
+      "step": 3830
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.696509797766743e-05,
+      "loss": 0.4193,
+      "step": 3831
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6963416771433246e-05,
+      "loss": 0.4681,
+      "step": 3832
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6961735183022034e-05,
+      "loss": 0.425,
+      "step": 3833
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6960053212526088e-05,
+      "loss": 0.4287,
+      "step": 3834
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695837086003772e-05,
+      "loss": 0.4363,
+      "step": 3835
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6956688125649264e-05,
+      "loss": 0.4412,
+      "step": 3836
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695500500945307e-05,
+      "loss": 0.4195,
+      "step": 3837
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6953321511541523e-05,
+      "loss": 0.4302,
+      "step": 3838
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695163763200701e-05,
+      "loss": 0.4263,
+      "step": 3839
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6949953370941952e-05,
+      "loss": 0.4224,
+      "step": 3840
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6948268728438784e-05,
+      "loss": 0.4507,
+      "step": 3841
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6946583704589973e-05,
+      "loss": 0.4165,
+      "step": 3842
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6944898299487996e-05,
+      "loss": 0.4399,
+      "step": 3843
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6943212513225347e-05,
+      "loss": 0.4325,
+      "step": 3844
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6941526345894552e-05,
+      "loss": 0.4279,
+      "step": 3845
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693983979758816e-05,
+      "loss": 0.4302,
+      "step": 3846
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693815286839873e-05,
+      "loss": 0.4294,
+      "step": 3847
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693646555841884e-05,
+      "loss": 0.4308,
+      "step": 3848
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6934777867741107e-05,
+      "loss": 0.4534,
+      "step": 3849
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693308979645815e-05,
+      "loss": 0.4252,
+      "step": 3850
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693140134466262e-05,
+      "loss": 0.443,
+      "step": 3851
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6929712512447185e-05,
+      "loss": 0.4375,
+      "step": 3852
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.692802329990453e-05,
+      "loss": 0.3964,
+      "step": 3853
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.692633370712737e-05,
+      "loss": 0.4285,
+      "step": 3854
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6924643734208435e-05,
+      "loss": 0.4279,
+      "step": 3855
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.692295338124047e-05,
+      "loss": 0.4165,
+      "step": 3856
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6921262648316263e-05,
+      "loss": 0.4191,
+      "step": 3857
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691957153552859e-05,
+      "loss": 0.4147,
+      "step": 3858
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6917880042970278e-05,
+      "loss": 0.4555,
+      "step": 3859
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6916188170734154e-05,
+      "loss": 0.4382,
+      "step": 3860
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6914495918913076e-05,
+      "loss": 0.4419,
+      "step": 3861
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691280328759992e-05,
+      "loss": 0.3946,
+      "step": 3862
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6911110276887588e-05,
+      "loss": 0.4111,
+      "step": 3863
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6909416886868996e-05,
+      "loss": 0.4589,
+      "step": 3864
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6907723117637085e-05,
+      "loss": 0.4738,
+      "step": 3865
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690602896928481e-05,
+      "loss": 0.4311,
+      "step": 3866
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6904334441905155e-05,
+      "loss": 0.4547,
+      "step": 3867
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6902639535591126e-05,
+      "loss": 0.4272,
+      "step": 3868
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6900944250435736e-05,
+      "loss": 0.4391,
+      "step": 3869
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6899248586532034e-05,
+      "loss": 0.4623,
+      "step": 3870
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6897552543973084e-05,
+      "loss": 0.4356,
+      "step": 3871
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6895856122851973e-05,
+      "loss": 0.425,
+      "step": 3872
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.68941593232618e-05,
+      "loss": 0.4281,
+      "step": 3873
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.68924621452957e-05,
+      "loss": 0.4407,
+      "step": 3874
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6890764589046812e-05,
+      "loss": 0.4228,
+      "step": 3875
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6889066654608306e-05,
+      "loss": 0.4288,
+      "step": 3876
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6887368342073374e-05,
+      "loss": 0.443,
+      "step": 3877
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6885669651535224e-05,
+      "loss": 0.4123,
+      "step": 3878
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6883970583087084e-05,
+      "loss": 0.4206,
+      "step": 3879
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6882271136822204e-05,
+      "loss": 0.4069,
+      "step": 3880
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6880571312833868e-05,
+      "loss": 0.4236,
+      "step": 3881
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687887111121535e-05,
+      "loss": 0.4224,
+      "step": 3882
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687717053205997e-05,
+      "loss": 0.4671,
+      "step": 3883
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687546957546107e-05,
+      "loss": 0.3968,
+      "step": 3884
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6873768241511994e-05,
+      "loss": 0.4123,
+      "step": 3885
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6872066530306123e-05,
+      "loss": 0.4546,
+      "step": 3886
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.687036444193685e-05,
+      "loss": 0.4466,
+      "step": 3887
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6868661976497593e-05,
+      "loss": 0.4541,
+      "step": 3888
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686695913408179e-05,
+      "loss": 0.4518,
+      "step": 3889
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6865255914782897e-05,
+      "loss": 0.4128,
+      "step": 3890
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6863552318694396e-05,
+      "loss": 0.4605,
+      "step": 3891
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686184834590978e-05,
+      "loss": 0.4408,
+      "step": 3892
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686014399652258e-05,
+      "loss": 0.4347,
+      "step": 3893
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6858439270626323e-05,
+      "loss": 0.4543,
+      "step": 3894
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6856734168314585e-05,
+      "loss": 0.4334,
+      "step": 3895
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685502868968093e-05,
+      "loss": 0.4333,
+      "step": 3896
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6853322834818983e-05,
+      "loss": 0.4064,
+      "step": 3897
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6851616603822348e-05,
+      "loss": 0.4299,
+      "step": 3898
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684990999678468e-05,
+      "loss": 0.4171,
+      "step": 3899
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684820301379964e-05,
+      "loss": 0.4431,
+      "step": 3900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6846495654960913e-05,
+      "loss": 0.4331,
+      "step": 3901
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6844787920362205e-05,
+      "loss": 0.4054,
+      "step": 3902
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6843079810097246e-05,
+      "loss": 0.4489,
+      "step": 3903
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6841371324259776e-05,
+      "loss": 0.4535,
+      "step": 3904
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683966246294357e-05,
+      "loss": 0.4226,
+      "step": 3905
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683795322624241e-05,
+      "loss": 0.452,
+      "step": 3906
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6836243614250113e-05,
+      "loss": 0.4185,
+      "step": 3907
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6834533627060503e-05,
+      "loss": 0.4534,
+      "step": 3908
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6832823264767428e-05,
+      "loss": 0.4494,
+      "step": 3909
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6831112527464765e-05,
+      "loss": 0.4257,
+      "step": 3910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.68294014152464e-05,
+      "loss": 0.4291,
+      "step": 3911
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6827689928206247e-05,
+      "loss": 0.422,
+      "step": 3912
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6825978066438243e-05,
+      "loss": 0.4181,
+      "step": 3913
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6824265830036332e-05,
+      "loss": 0.4307,
+      "step": 3914
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6822553219094493e-05,
+      "loss": 0.4056,
+      "step": 3915
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.682084023370672e-05,
+      "loss": 0.4273,
+      "step": 3916
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6819126873967027e-05,
+      "loss": 0.4116,
+      "step": 3917
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6817413139969448e-05,
+      "loss": 0.4285,
+      "step": 3918
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6815699031808045e-05,
+      "loss": 0.4368,
+      "step": 3919
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6813984549576882e-05,
+      "loss": 0.4257,
+      "step": 3920
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6812269693370068e-05,
+      "loss": 0.4485,
+      "step": 3921
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6810554463281715e-05,
+      "loss": 0.4343,
+      "step": 3922
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6808838859405965e-05,
+      "loss": 0.4255,
+      "step": 3923
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680712288183697e-05,
+      "loss": 0.4532,
+      "step": 3924
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680540653066891e-05,
+      "loss": 0.3998,
+      "step": 3925
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680368980599599e-05,
+      "loss": 0.4267,
+      "step": 3926
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6801972707912427e-05,
+      "loss": 0.4307,
+      "step": 3927
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6800255236512456e-05,
+      "loss": 0.4573,
+      "step": 3928
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6798537391890347e-05,
+      "loss": 0.4746,
+      "step": 3929
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679681917414038e-05,
+      "loss": 0.4233,
+      "step": 3930
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679510058335685e-05,
+      "loss": 0.3963,
+      "step": 3931
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679338161963408e-05,
+      "loss": 0.3782,
+      "step": 3932
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679166228306642e-05,
+      "loss": 0.4423,
+      "step": 3933
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678994257374823e-05,
+      "loss": 0.4153,
+      "step": 3934
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6788222491773895e-05,
+      "loss": 0.4044,
+      "step": 3935
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6786502037237816e-05,
+      "loss": 0.4206,
+      "step": 3936
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6784781210234418e-05,
+      "loss": 0.4063,
+      "step": 3937
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678306001085815e-05,
+      "loss": 0.4206,
+      "step": 3938
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6781338439203475e-05,
+      "loss": 0.443,
+      "step": 3939
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6779616495364873e-05,
+      "loss": 0.3808,
+      "step": 3940
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6777894179436863e-05,
+      "loss": 0.4463,
+      "step": 3941
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677617149151396e-05,
+      "loss": 0.3947,
+      "step": 3942
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677444843169072e-05,
+      "loss": 0.4273,
+      "step": 3943
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6772725000061706e-05,
+      "loss": 0.4383,
+      "step": 3944
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6771001196721503e-05,
+      "loss": 0.4678,
+      "step": 3945
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6769277021764728e-05,
+      "loss": 0.441,
+      "step": 3946
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6767552475286003e-05,
+      "loss": 0.4518,
+      "step": 3947
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676582755737998e-05,
+      "loss": 0.3945,
+      "step": 3948
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676410226814133e-05,
+      "loss": 0.4275,
+      "step": 3949
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6762376607664738e-05,
+      "loss": 0.417,
+      "step": 3950
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676065057604492e-05,
+      "loss": 0.4149,
+      "step": 3951
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6758924173376604e-05,
+      "loss": 0.4431,
+      "step": 3952
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6757197399754538e-05,
+      "loss": 0.4163,
+      "step": 3953
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6755470255273498e-05,
+      "loss": 0.4587,
+      "step": 3954
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6753742740028278e-05,
+      "loss": 0.4314,
+      "step": 3955
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6752014854113682e-05,
+      "loss": 0.4178,
+      "step": 3956
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675028659762455e-05,
+      "loss": 0.4376,
+      "step": 3957
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6748557970655732e-05,
+      "loss": 0.3998,
+      "step": 3958
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.67468289733021e-05,
+      "loss": 0.3989,
+      "step": 3959
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674509960565855e-05,
+      "loss": 0.4311,
+      "step": 3960
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674336986781999e-05,
+      "loss": 0.4033,
+      "step": 3961
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6741639759881367e-05,
+      "loss": 0.4322,
+      "step": 3962
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6739909281937624e-05,
+      "loss": 0.4445,
+      "step": 3963
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6738178434083736e-05,
+      "loss": 0.4324,
+      "step": 3964
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6736447216414702e-05,
+      "loss": 0.449,
+      "step": 3965
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673471562902554e-05,
+      "loss": 0.4599,
+      "step": 3966
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673298367201128e-05,
+      "loss": 0.4558,
+      "step": 3967
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6731251345466978e-05,
+      "loss": 0.4169,
+      "step": 3968
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6729518649487712e-05,
+      "loss": 0.3923,
+      "step": 3969
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6727785584168583e-05,
+      "loss": 0.4472,
+      "step": 3970
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.67260521496047e-05,
+      "loss": 0.4434,
+      "step": 3971
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6724318345891207e-05,
+      "loss": 0.4376,
+      "step": 3972
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6722584173123256e-05,
+      "loss": 0.4041,
+      "step": 3973
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6720849631396026e-05,
+      "loss": 0.4207,
+      "step": 3974
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6719114720804715e-05,
+      "loss": 0.4282,
+      "step": 3975
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6717379441444542e-05,
+      "loss": 0.4255,
+      "step": 3976
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6715643793410745e-05,
+      "loss": 0.4283,
+      "step": 3977
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.671390777679858e-05,
+      "loss": 0.4306,
+      "step": 3978
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6712171391703332e-05,
+      "loss": 0.4176,
+      "step": 3979
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6710434638220293e-05,
+      "loss": 0.4134,
+      "step": 3980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6708697516444783e-05,
+      "loss": 0.4407,
+      "step": 3981
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6706960026472146e-05,
+      "loss": 0.4326,
+      "step": 3982
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670522216839774e-05,
+      "loss": 0.4353,
+      "step": 3983
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670348394231694e-05,
+      "loss": 0.4408,
+      "step": 3984
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6701745348325155e-05,
+      "loss": 0.4341,
+      "step": 3985
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.67000063865178e-05,
+      "loss": 0.4136,
+      "step": 3986
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669826705699031e-05,
+      "loss": 0.4102,
+      "step": 3987
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6696527359838155e-05,
+      "loss": 0.4686,
+      "step": 3988
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6694787295156812e-05,
+      "loss": 0.4236,
+      "step": 3989
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669304686304178e-05,
+      "loss": 0.4643,
+      "step": 3990
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.4144,
+      "step": 3991
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.668956489689276e-05,
+      "loss": 0.4294,
+      "step": 3992
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6687823363049874e-05,
+      "loss": 0.4128,
+      "step": 3993
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.668608146215551e-05,
+      "loss": 0.4401,
+      "step": 3994
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.668433919430526e-05,
+      "loss": 0.4328,
+      "step": 3995
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6682596559594757e-05,
+      "loss": 0.469,
+      "step": 3996
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.668085355811963e-05,
+      "loss": 0.4039,
+      "step": 3997
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6679110189975556e-05,
+      "loss": 0.4127,
+      "step": 3998
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6677366455258208e-05,
+      "loss": 0.4364,
+      "step": 3999
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667562235406329e-05,
+      "loss": 0.4279,
+      "step": 4000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6673877886486522e-05,
+      "loss": 0.4402,
+      "step": 4001
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6672133052623653e-05,
+      "loss": 0.4003,
+      "step": 4002
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6670387852570442e-05,
+      "loss": 0.4568,
+      "step": 4003
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666864228642267e-05,
+      "loss": 0.4125,
+      "step": 4004
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6666896354276142e-05,
+      "loss": 0.4168,
+      "step": 4005
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666515005622668e-05,
+      "loss": 0.4185,
+      "step": 4006
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666340339237013e-05,
+      "loss": 0.4634,
+      "step": 4007
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666165636280235e-05,
+      "loss": 0.4274,
+      "step": 4008
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6659908967619226e-05,
+      "loss": 0.4151,
+      "step": 4009
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665816120691666e-05,
+      "loss": 0.4329,
+      "step": 4010
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665641308079058e-05,
+      "loss": 0.4483,
+      "step": 4011
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6654664589336925e-05,
+      "loss": 0.4236,
+      "step": 4012
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665291573265166e-05,
+      "loss": 0.4231,
+      "step": 4013
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6651166510830765e-05,
+      "loss": 0.4351,
+      "step": 4014
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6649416923970248e-05,
+      "loss": 0.454,
+      "step": 4015
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664766697216613e-05,
+      "loss": 0.4174,
+      "step": 4016
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6645916655514457e-05,
+      "loss": 0.4437,
+      "step": 4017
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6644165974111288e-05,
+      "loss": 0.435,
+      "step": 4018
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6642414928052713e-05,
+      "loss": 0.4145,
+      "step": 4019
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664066351743483e-05,
+      "loss": 0.4261,
+      "step": 4020
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6638911742353764e-05,
+      "loss": 0.4242,
+      "step": 4021
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.663715960290566e-05,
+      "loss": 0.4067,
+      "step": 4022
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6635407099186683e-05,
+      "loss": 0.4536,
+      "step": 4023
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6633654231293013e-05,
+      "loss": 0.4685,
+      "step": 4024
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6631900999320854e-05,
+      "loss": 0.3997,
+      "step": 4025
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.663014740336643e-05,
+      "loss": 0.4271,
+      "step": 4026
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6628393443525993e-05,
+      "loss": 0.406,
+      "step": 4027
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6626639119895794e-05,
+      "loss": 0.3929,
+      "step": 4028
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662488443257212e-05,
+      "loss": 0.4521,
+      "step": 4029
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6623129381651277e-05,
+      "loss": 0.4281,
+      "step": 4030
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6621373967229588e-05,
+      "loss": 0.4337,
+      "step": 4031
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6619618189403396e-05,
+      "loss": 0.3955,
+      "step": 4032
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6617862048269065e-05,
+      "loss": 0.3995,
+      "step": 4033
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6616105543922975e-05,
+      "loss": 0.4341,
+      "step": 4034
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661434867646153e-05,
+      "loss": 0.44,
+      "step": 4035
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661259144598116e-05,
+      "loss": 0.429,
+      "step": 4036
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.66108338525783e-05,
+      "loss": 0.4558,
+      "step": 4037
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6609075896349413e-05,
+      "loss": 0.4702,
+      "step": 4038
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660731757739099e-05,
+      "loss": 0.4539,
+      "step": 4039
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6605558895799525e-05,
+      "loss": 0.4332,
+      "step": 4040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6603799851671543e-05,
+      "loss": 0.432,
+      "step": 4041
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660204044510359e-05,
+      "loss": 0.4521,
+      "step": 4042
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6600280676192226e-05,
+      "loss": 0.4321,
+      "step": 4043
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659852054503403e-05,
+      "loss": 0.4333,
+      "step": 4044
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659676005172561e-05,
+      "loss": 0.4126,
+      "step": 4045
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6594999196363586e-05,
+      "loss": 0.4356,
+      "step": 4046
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6593237979044598e-05,
+      "loss": 0.4285,
+      "step": 4047
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6591476399865314e-05,
+      "loss": 0.4093,
+      "step": 4048
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6589714458922406e-05,
+      "loss": 0.4451,
+      "step": 4049
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6587952156312585e-05,
+      "loss": 0.4458,
+      "step": 4050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6586189492132567e-05,
+      "loss": 0.4353,
+      "step": 4051
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6584426466479097e-05,
+      "loss": 0.4183,
+      "step": 4052
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6582663079448928e-05,
+      "loss": 0.4178,
+      "step": 4053
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.658089933113885e-05,
+      "loss": 0.4121,
+      "step": 4054
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657913522164566e-05,
+      "loss": 0.4311,
+      "step": 4055
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657737075106618e-05,
+      "loss": 0.4239,
+      "step": 4056
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657560591949725e-05,
+      "loss": 0.4321,
+      "step": 4057
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6573840727035727e-05,
+      "loss": 0.4394,
+      "step": 4058
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6572075173778498e-05,
+      "loss": 0.4288,
+      "step": 4059
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6570309259822455e-05,
+      "loss": 0.4292,
+      "step": 4060
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6568542985264524e-05,
+      "loss": 0.4492,
+      "step": 4061
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6566776350201638e-05,
+      "loss": 0.4747,
+      "step": 4062
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656500935473076e-05,
+      "loss": 0.4322,
+      "step": 4063
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656324199894887e-05,
+      "loss": 0.4176,
+      "step": 4064
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6561474282952965e-05,
+      "loss": 0.4316,
+      "step": 4065
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6559706206840065e-05,
+      "loss": 0.4366,
+      "step": 4066
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6557937770707205e-05,
+      "loss": 0.4219,
+      "step": 4067
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.655616897465145e-05,
+      "loss": 0.4277,
+      "step": 4068
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.655439981876987e-05,
+      "loss": 0.428,
+      "step": 4069
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6552630303159563e-05,
+      "loss": 0.4315,
+      "step": 4070
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6550860427917652e-05,
+      "loss": 0.4201,
+      "step": 4071
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6549090193141272e-05,
+      "loss": 0.4236,
+      "step": 4072
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6547319598927577e-05,
+      "loss": 0.3871,
+      "step": 4073
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.654554864537375e-05,
+      "loss": 0.4229,
+      "step": 4074
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6543777332576976e-05,
+      "loss": 0.4168,
+      "step": 4075
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6542005660634483e-05,
+      "loss": 0.4253,
+      "step": 4076
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6540233629643495e-05,
+      "loss": 0.4058,
+      "step": 4077
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.653846123970128e-05,
+      "loss": 0.4156,
+      "step": 4078
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6536688490905098e-05,
+      "loss": 0.4669,
+      "step": 4079
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6534915383352263e-05,
+      "loss": 0.438,
+      "step": 4080
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653314191714007e-05,
+      "loss": 0.4466,
+      "step": 4081
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6531368092365868e-05,
+      "loss": 0.4428,
+      "step": 4082
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6529593909127e-05,
+      "loss": 0.4184,
+      "step": 4083
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6527819367520847e-05,
+      "loss": 0.4348,
+      "step": 4084
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6526044467644794e-05,
+      "loss": 0.4344,
+      "step": 4085
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6524269209596265e-05,
+      "loss": 0.4457,
+      "step": 4086
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6522493593472683e-05,
+      "loss": 0.443,
+      "step": 4087
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6520717619371507e-05,
+      "loss": 0.4457,
+      "step": 4088
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6518941287390206e-05,
+      "loss": 0.4295,
+      "step": 4089
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6517164597626268e-05,
+      "loss": 0.4444,
+      "step": 4090
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6515387550177205e-05,
+      "loss": 0.4515,
+      "step": 4091
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651361014514055e-05,
+      "loss": 0.4204,
+      "step": 4092
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6511832382613853e-05,
+      "loss": 0.4236,
+      "step": 4093
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651005426269469e-05,
+      "loss": 0.4251,
+      "step": 4094
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6508275785480636e-05,
+      "loss": 0.4452,
+      "step": 4095
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6506496951069313e-05,
+      "loss": 0.412,
+      "step": 4096
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6504717759558345e-05,
+      "loss": 0.4048,
+      "step": 4097
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650293821104538e-05,
+      "loss": 0.4188,
+      "step": 4098
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6501158305628084e-05,
+      "loss": 0.3999,
+      "step": 4099
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649937804340415e-05,
+      "loss": 0.4387,
+      "step": 4100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6497597424471277e-05,
+      "loss": 0.4128,
+      "step": 4101
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6495816448927204e-05,
+      "loss": 0.419,
+      "step": 4102
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6494035116869663e-05,
+      "loss": 0.4255,
+      "step": 4103
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649225342839643e-05,
+      "loss": 0.4094,
+      "step": 4104
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649047138360529e-05,
+      "loss": 0.4232,
+      "step": 4105
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6488688982594042e-05,
+      "loss": 0.4059,
+      "step": 4106
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6486906225460512e-05,
+      "loss": 0.441,
+      "step": 4107
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6485123112302547e-05,
+      "loss": 0.3929,
+      "step": 4108
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.648333964321801e-05,
+      "loss": 0.4546,
+      "step": 4109
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6481555818304783e-05,
+      "loss": 0.4307,
+      "step": 4110
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647977163766077e-05,
+      "loss": 0.4493,
+      "step": 4111
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647798710138389e-05,
+      "loss": 0.4688,
+      "step": 4112
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647620220957209e-05,
+      "loss": 0.431,
+      "step": 4113
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6474416962323327e-05,
+      "loss": 0.4213,
+      "step": 4114
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647263135973558e-05,
+      "loss": 0.4173,
+      "step": 4115
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6470845401906855e-05,
+      "loss": 0.4671,
+      "step": 4116
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646905908893517e-05,
+      "loss": 0.4223,
+      "step": 4117
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646727242091856e-05,
+      "loss": 0.4145,
+      "step": 4118
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6465485397955085e-05,
+      "loss": 0.4357,
+      "step": 4119
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6463698020142827e-05,
+      "loss": 0.4631,
+      "step": 4120
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646191028757988e-05,
+      "loss": 0.4304,
+      "step": 4121
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6460122200364365e-05,
+      "loss": 0.4433,
+      "step": 4122
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6458333758594415e-05,
+      "loss": 0.3827,
+      "step": 4123
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6456544962368187e-05,
+      "loss": 0.4399,
+      "step": 4124
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6454755811783855e-05,
+      "loss": 0.4326,
+      "step": 4125
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6452966306939617e-05,
+      "loss": 0.4633,
+      "step": 4126
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6451176447933686e-05,
+      "loss": 0.405,
+      "step": 4127
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6449386234864297e-05,
+      "loss": 0.4276,
+      "step": 4128
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6447595667829698e-05,
+      "loss": 0.4102,
+      "step": 4129
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6445804746928166e-05,
+      "loss": 0.4015,
+      "step": 4130
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6444013472257995e-05,
+      "loss": 0.433,
+      "step": 4131
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6442221843917496e-05,
+      "loss": 0.4353,
+      "step": 4132
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6440429862004995e-05,
+      "loss": 0.3967,
+      "step": 4133
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6438637526618845e-05,
+      "loss": 0.4471,
+      "step": 4134
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6436844837857417e-05,
+      "loss": 0.3819,
+      "step": 4135
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6435051795819103e-05,
+      "loss": 0.4464,
+      "step": 4136
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6433258400602304e-05,
+      "loss": 0.4347,
+      "step": 4137
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643146465230545e-05,
+      "loss": 0.4221,
+      "step": 4138
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642967055102699e-05,
+      "loss": 0.4262,
+      "step": 4139
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.4577,
+      "step": 4140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6426081289919144e-05,
+      "loss": 0.4497,
+      "step": 4141
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6424286130286745e-05,
+      "loss": 0.4261,
+      "step": 4142
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642249061806672e-05,
+      "loss": 0.4577,
+      "step": 4143
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6420694753357622e-05,
+      "loss": 0.4264,
+      "step": 4144
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6418898536258006e-05,
+      "loss": 0.4134,
+      "step": 4145
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6417101966866454e-05,
+      "loss": 0.4275,
+      "step": 4146
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6415305045281574e-05,
+      "loss": 0.4282,
+      "step": 4147
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6413507771601984e-05,
+      "loss": 0.3956,
+      "step": 4148
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6411710145926326e-05,
+      "loss": 0.4461,
+      "step": 4149
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640991216835326e-05,
+      "loss": 0.4635,
+      "step": 4150
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640811383898146e-05,
+      "loss": 0.4226,
+      "step": 4151
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6406315157909635e-05,
+      "loss": 0.4034,
+      "step": 4152
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6404516125236493e-05,
+      "loss": 0.4304,
+      "step": 4153
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640271674106078e-05,
+      "loss": 0.4481,
+      "step": 4154
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6400917005481246e-05,
+      "loss": 0.3869,
+      "step": 4155
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6399116918596668e-05,
+      "loss": 0.4283,
+      "step": 4156
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6397316480505847e-05,
+      "loss": 0.4388,
+      "step": 4157
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639551569130759e-05,
+      "loss": 0.4443,
+      "step": 4158
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6393714551100734e-05,
+      "loss": 0.3914,
+      "step": 4159
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639191305998413e-05,
+      "loss": 0.4777,
+      "step": 4160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6390111218056653e-05,
+      "loss": 0.4636,
+      "step": 4161
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6388309025417194e-05,
+      "loss": 0.4246,
+      "step": 4162
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638650648216466e-05,
+      "loss": 0.3996,
+      "step": 4163
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6384703588397987e-05,
+      "loss": 0.4715,
+      "step": 4164
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6382900344216116e-05,
+      "loss": 0.4253,
+      "step": 4165
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6381096749718027e-05,
+      "loss": 0.4683,
+      "step": 4166
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.63792928050027e-05,
+      "loss": 0.4651,
+      "step": 4167
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637748851016914e-05,
+      "loss": 0.4761,
+      "step": 4168
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6375683865316377e-05,
+      "loss": 0.4561,
+      "step": 4169
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6373878870543457e-05,
+      "loss": 0.4219,
+      "step": 4170
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637207352594944e-05,
+      "loss": 0.4059,
+      "step": 4171
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637026783163342e-05,
+      "loss": 0.44,
+      "step": 4172
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6368461787694485e-05,
+      "loss": 0.4336,
+      "step": 4173
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6366655394231767e-05,
+      "loss": 0.4345,
+      "step": 4174
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6364848651344407e-05,
+      "loss": 0.4141,
+      "step": 4175
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6363041559131562e-05,
+      "loss": 0.4106,
+      "step": 4176
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6361234117692414e-05,
+      "loss": 0.4704,
+      "step": 4177
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6359426327126163e-05,
+      "loss": 0.4219,
+      "step": 4178
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6357618187532017e-05,
+      "loss": 0.4308,
+      "step": 4179
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635580969900923e-05,
+      "loss": 0.496,
+      "step": 4180
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6354000861657045e-05,
+      "loss": 0.4332,
+      "step": 4181
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6352191675574743e-05,
+      "loss": 0.4365,
+      "step": 4182
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6350382140861616e-05,
+      "loss": 0.4131,
+      "step": 4183
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6348572257616978e-05,
+      "loss": 0.4358,
+      "step": 4184
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6346762025940167e-05,
+      "loss": 0.4298,
+      "step": 4185
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6344951445930527e-05,
+      "loss": 0.4139,
+      "step": 4186
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6343140517687434e-05,
+      "loss": 0.427,
+      "step": 4187
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6341329241310276e-05,
+      "loss": 0.4562,
+      "step": 4188
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6339517616898464e-05,
+      "loss": 0.44,
+      "step": 4189
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6337705644551427e-05,
+      "loss": 0.4648,
+      "step": 4190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633589332436861e-05,
+      "loss": 0.4831,
+      "step": 4191
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633408065644948e-05,
+      "loss": 0.4533,
+      "step": 4192
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6332267640893523e-05,
+      "loss": 0.4211,
+      "step": 4193
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6330454277800244e-05,
+      "loss": 0.4588,
+      "step": 4194
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632864056726917e-05,
+      "loss": 0.4414,
+      "step": 4195
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6326826509399836e-05,
+      "loss": 0.417,
+      "step": 4196
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6325012104291812e-05,
+      "loss": 0.4465,
+      "step": 4197
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6323197352044675e-05,
+      "loss": 0.4201,
+      "step": 4198
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6321382252758023e-05,
+      "loss": 0.4326,
+      "step": 4199
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6319566806531483e-05,
+      "loss": 0.4777,
+      "step": 4200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6317751013464684e-05,
+      "loss": 0.4129,
+      "step": 4201
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6315934873657287e-05,
+      "loss": 0.4422,
+      "step": 4202
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631411838720897e-05,
+      "loss": 0.4456,
+      "step": 4203
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6312301554219426e-05,
+      "loss": 0.4102,
+      "step": 4204
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6310484374788368e-05,
+      "loss": 0.439,
+      "step": 4205
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6308666849015532e-05,
+      "loss": 0.4106,
+      "step": 4206
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6306848977000672e-05,
+      "loss": 0.4521,
+      "step": 4207
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6305030758843552e-05,
+      "loss": 0.4373,
+      "step": 4208
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630321219464397e-05,
+      "loss": 0.4498,
+      "step": 4209
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630139328450173e-05,
+      "loss": 0.4413,
+      "step": 4210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6299574028516666e-05,
+      "loss": 0.4452,
+      "step": 4211
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6297754426788615e-05,
+      "loss": 0.403,
+      "step": 4212
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6295934479417454e-05,
+      "loss": 0.4206,
+      "step": 4213
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6294114186503064e-05,
+      "loss": 0.4311,
+      "step": 4214
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6292293548145348e-05,
+      "loss": 0.4273,
+      "step": 4215
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629047256444423e-05,
+      "loss": 0.4388,
+      "step": 4216
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6288651235499654e-05,
+      "loss": 0.4036,
+      "step": 4217
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6286829561411575e-05,
+      "loss": 0.4349,
+      "step": 4218
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.628500754227998e-05,
+      "loss": 0.4262,
+      "step": 4219
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6283185178204866e-05,
+      "loss": 0.4109,
+      "step": 4220
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6281362469286247e-05,
+      "loss": 0.4259,
+      "step": 4221
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6279539415624165e-05,
+      "loss": 0.4527,
+      "step": 4222
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627771601731867e-05,
+      "loss": 0.416,
+      "step": 4223
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6275892274469845e-05,
+      "loss": 0.4222,
+      "step": 4224
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627406818717777e-05,
+      "loss": 0.4416,
+      "step": 4225
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627224375554257e-05,
+      "loss": 0.4529,
+      "step": 4226
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6270418979664377e-05,
+      "loss": 0.4069,
+      "step": 4227
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6268593859643333e-05,
+      "loss": 0.4322,
+      "step": 4228
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6266768395579603e-05,
+      "loss": 0.443,
+      "step": 4229
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6264942587573388e-05,
+      "loss": 0.4514,
+      "step": 4230
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626311643572489e-05,
+      "loss": 0.4133,
+      "step": 4231
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6261289940134333e-05,
+      "loss": 0.4272,
+      "step": 4232
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625946310090196e-05,
+      "loss": 0.4287,
+      "step": 4233
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625763591812804e-05,
+      "loss": 0.4206,
+      "step": 4234
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6255808391912847e-05,
+      "loss": 0.4331,
+      "step": 4235
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625398052235669e-05,
+      "loss": 0.4107,
+      "step": 4236
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6252152309559885e-05,
+      "loss": 0.4297,
+      "step": 4237
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6250323753622773e-05,
+      "loss": 0.3752,
+      "step": 4238
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6248494854645712e-05,
+      "loss": 0.4564,
+      "step": 4239
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6246665612729074e-05,
+      "loss": 0.4213,
+      "step": 4240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6244836027973255e-05,
+      "loss": 0.419,
+      "step": 4241
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6243006100478673e-05,
+      "loss": 0.4227,
+      "step": 4242
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6241175830345762e-05,
+      "loss": 0.4184,
+      "step": 4243
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6239345217674967e-05,
+      "loss": 0.4402,
+      "step": 4244
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6237514262566766e-05,
+      "loss": 0.4587,
+      "step": 4245
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623568296512164e-05,
+      "loss": 0.4146,
+      "step": 4246
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6233851325440107e-05,
+      "loss": 0.4172,
+      "step": 4247
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6232019343622688e-05,
+      "loss": 0.4096,
+      "step": 4248
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623018701976993e-05,
+      "loss": 0.4297,
+      "step": 4249
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6228354353982393e-05,
+      "loss": 0.4154,
+      "step": 4250
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6226521346360665e-05,
+      "loss": 0.4492,
+      "step": 4251
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6224687997005347e-05,
+      "loss": 0.4629,
+      "step": 4252
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622285430601706e-05,
+      "loss": 0.4265,
+      "step": 4253
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6221020273496443e-05,
+      "loss": 0.4419,
+      "step": 4254
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6219185899544153e-05,
+      "loss": 0.4207,
+      "step": 4255
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621735118426087e-05,
+      "loss": 0.4646,
+      "step": 4256
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621551612774729e-05,
+      "loss": 0.4296,
+      "step": 4257
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6213680730104124e-05,
+      "loss": 0.4407,
+      "step": 4258
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6211844991432105e-05,
+      "loss": 0.4053,
+      "step": 4259
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621000891183199e-05,
+      "loss": 0.4367,
+      "step": 4260
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620817249140454e-05,
+      "loss": 0.4434,
+      "step": 4261
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6206335730250553e-05,
+      "loss": 0.458,
+      "step": 4262
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620449862847084e-05,
+      "loss": 0.4035,
+      "step": 4263
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6202661186166214e-05,
+      "loss": 0.4179,
+      "step": 4264
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620082340343753e-05,
+      "loss": 0.4935,
+      "step": 4265
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619898528038565e-05,
+      "loss": 0.454,
+      "step": 4266
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619714681711146e-05,
+      "loss": 0.435,
+      "step": 4267
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6195308013715858e-05,
+      "loss": 0.4559,
+      "step": 4268
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619346887029976e-05,
+      "loss": 0.4472,
+      "step": 4269
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6191629386964114e-05,
+      "loss": 0.4168,
+      "step": 4270
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6189789563809872e-05,
+      "loss": 0.4209,
+      "step": 4271
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.618794940093801e-05,
+      "loss": 0.4265,
+      "step": 4272
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.618610889844952e-05,
+      "loss": 0.3966,
+      "step": 4273
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.618426805644542e-05,
+      "loss": 0.4438,
+      "step": 4274
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6182426875026743e-05,
+      "loss": 0.4372,
+      "step": 4275
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6180585354294537e-05,
+      "loss": 0.4366,
+      "step": 4276
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617874349434987e-05,
+      "loss": 0.4314,
+      "step": 4277
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617690129529383e-05,
+      "loss": 0.4395,
+      "step": 4278
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6175058757227528e-05,
+      "loss": 0.4331,
+      "step": 4279
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6173215880252083e-05,
+      "loss": 0.436,
+      "step": 4280
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617137266446864e-05,
+      "loss": 0.4577,
+      "step": 4281
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6169529109978362e-05,
+      "loss": 0.4328,
+      "step": 4282
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616768521688243e-05,
+      "loss": 0.4301,
+      "step": 4283
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6165840985282046e-05,
+      "loss": 0.4348,
+      "step": 4284
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6163996415278423e-05,
+      "loss": 0.4336,
+      "step": 4285
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6162151506972802e-05,
+      "loss": 0.4019,
+      "step": 4286
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6160306260466438e-05,
+      "loss": 0.4471,
+      "step": 4287
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.61584606758606e-05,
+      "loss": 0.4458,
+      "step": 4288
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6156614753256583e-05,
+      "loss": 0.4432,
+      "step": 4289
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.61547684927557e-05,
+      "loss": 0.4372,
+      "step": 4290
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6152921894459277e-05,
+      "loss": 0.4327,
+      "step": 4291
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6151074958468664e-05,
+      "loss": 0.4438,
+      "step": 4292
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614922768488523e-05,
+      "loss": 0.406,
+      "step": 4293
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6147380073810345e-05,
+      "loss": 0.4484,
+      "step": 4294
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6145532125345434e-05,
+      "loss": 0.4273,
+      "step": 4295
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6143683839591905e-05,
+      "loss": 0.4205,
+      "step": 4296
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.61418352166512e-05,
+      "loss": 0.4292,
+      "step": 4297
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6139986256624786e-05,
+      "loss": 0.4322,
+      "step": 4298
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6138136959614135e-05,
+      "loss": 0.4143,
+      "step": 4299
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6136287325720737e-05,
+      "loss": 0.4475,
+      "step": 4300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613443735504612e-05,
+      "loss": 0.4647,
+      "step": 4301
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.61325870476918e-05,
+      "loss": 0.4735,
+      "step": 4302
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6130736403759342e-05,
+      "loss": 0.4466,
+      "step": 4303
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6128885423350313e-05,
+      "loss": 0.4578,
+      "step": 4304
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.61270341065663e-05,
+      "loss": 0.4441,
+      "step": 4305
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6125182453508908e-05,
+      "loss": 0.4499,
+      "step": 4306
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6123330464279762e-05,
+      "loss": 0.4364,
+      "step": 4307
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6121478138980513e-05,
+      "loss": 0.4452,
+      "step": 4308
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6119625477712815e-05,
+      "loss": 0.4178,
+      "step": 4309
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611777248057835e-05,
+      "loss": 0.4067,
+      "step": 4310
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6115919147678823e-05,
+      "loss": 0.4593,
+      "step": 4311
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6114065479115946e-05,
+      "loss": 0.4353,
+      "step": 4312
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6112211474991453e-05,
+      "loss": 0.4493,
+      "step": 4313
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6110357135407104e-05,
+      "loss": 0.4078,
+      "step": 4314
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6108502460464666e-05,
+      "loss": 0.4608,
+      "step": 4315
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6106647450265938e-05,
+      "loss": 0.4503,
+      "step": 4316
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6104792104912722e-05,
+      "loss": 0.4442,
+      "step": 4317
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6102936424506848e-05,
+      "loss": 0.413,
+      "step": 4318
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6101080409150164e-05,
+      "loss": 0.4208,
+      "step": 4319
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6099224058944537e-05,
+      "loss": 0.4389,
+      "step": 4320
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6097367373991842e-05,
+      "loss": 0.4358,
+      "step": 4321
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6095510354393987e-05,
+      "loss": 0.4137,
+      "step": 4322
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609365300025289e-05,
+      "loss": 0.4206,
+      "step": 4323
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6091795311670493e-05,
+      "loss": 0.4372,
+      "step": 4324
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6089937288748745e-05,
+      "loss": 0.401,
+      "step": 4325
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6088078931589626e-05,
+      "loss": 0.4193,
+      "step": 4326
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6086220240295126e-05,
+      "loss": 0.3833,
+      "step": 4327
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608436121496726e-05,
+      "loss": 0.4199,
+      "step": 4328
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608250185570806e-05,
+      "loss": 0.4358,
+      "step": 4329
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6080642162619567e-05,
+      "loss": 0.4408,
+      "step": 4330
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607878213580385e-05,
+      "loss": 0.427,
+      "step": 4331
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6076921775363e-05,
+      "loss": 0.4593,
+      "step": 4332
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6075061081399112e-05,
+      "loss": 0.4474,
+      "step": 4333
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607320005401431e-05,
+      "loss": 0.4265,
+      "step": 4334
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6071338693310735e-05,
+      "loss": 0.4112,
+      "step": 4335
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6069476999390546e-05,
+      "loss": 0.3987,
+      "step": 4336
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6067614972355917e-05,
+      "loss": 0.4402,
+      "step": 4337
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6065752612309045e-05,
+      "loss": 0.4456,
+      "step": 4338
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606388991935214e-05,
+      "loss": 0.4547,
+      "step": 4339
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6062026893587437e-05,
+      "loss": 0.4222,
+      "step": 4340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6060163535117182e-05,
+      "loss": 0.43,
+      "step": 4341
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605829984404364e-05,
+      "loss": 0.4358,
+      "step": 4342
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605643582046911e-05,
+      "loss": 0.4213,
+      "step": 4343
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6054571464495885e-05,
+      "loss": 0.4066,
+      "step": 4344
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6052706776226285e-05,
+      "loss": 0.4701,
+      "step": 4345
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6050841755762658e-05,
+      "loss": 0.4081,
+      "step": 4346
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604897640320736e-05,
+      "loss": 0.4194,
+      "step": 4347
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604711071866277e-05,
+      "loss": 0.4137,
+      "step": 4348
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6045244702231283e-05,
+      "loss": 0.4352,
+      "step": 4349
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6043378354015305e-05,
+      "loss": 0.4298,
+      "step": 4350
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6041511674117282e-05,
+      "loss": 0.4498,
+      "step": 4351
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6039644662639653e-05,
+      "loss": 0.3977,
+      "step": 4352
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.603777731968489e-05,
+      "loss": 0.4073,
+      "step": 4353
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6035909645355478e-05,
+      "loss": 0.4293,
+      "step": 4354
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6034041639753924e-05,
+      "loss": 0.4331,
+      "step": 4355
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6032173302982748e-05,
+      "loss": 0.4166,
+      "step": 4356
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6030304635144493e-05,
+      "loss": 0.4358,
+      "step": 4357
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.602843563634172e-05,
+      "loss": 0.4338,
+      "step": 4358
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6026566306677e-05,
+      "loss": 0.409,
+      "step": 4359
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6024696646252933e-05,
+      "loss": 0.3915,
+      "step": 4360
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6022826655172134e-05,
+      "loss": 0.4435,
+      "step": 4361
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.602095633353723e-05,
+      "loss": 0.4195,
+      "step": 4362
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6019085681450874e-05,
+      "loss": 0.4354,
+      "step": 4363
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6017214699015733e-05,
+      "loss": 0.4339,
+      "step": 4364
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6015343386334495e-05,
+      "loss": 0.4231,
+      "step": 4365
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6013471743509864e-05,
+      "loss": 0.4218,
+      "step": 4366
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6011599770644556e-05,
+      "loss": 0.4442,
+      "step": 4367
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6009727467841322e-05,
+      "loss": 0.4492,
+      "step": 4368
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6007854835202916e-05,
+      "loss": 0.4071,
+      "step": 4369
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600598187283211e-05,
+      "loss": 0.4142,
+      "step": 4370
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6004108580831705e-05,
+      "loss": 0.387,
+      "step": 4371
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6002234959304512e-05,
+      "loss": 0.4271,
+      "step": 4372
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6000361008353365e-05,
+      "loss": 0.4402,
+      "step": 4373
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5998486728081103e-05,
+      "loss": 0.4414,
+      "step": 4374
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5996612118590604e-05,
+      "loss": 0.4039,
+      "step": 4375
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5994737179984753e-05,
+      "loss": 0.4536,
+      "step": 4376
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5992861912366446e-05,
+      "loss": 0.4299,
+      "step": 4377
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5990986315838604e-05,
+      "loss": 0.4206,
+      "step": 4378
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5989110390504174e-05,
+      "loss": 0.4137,
+      "step": 4379
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5987234136466113e-05,
+      "loss": 0.4048,
+      "step": 4380
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.598535755382739e-05,
+      "loss": 0.4543,
+      "step": 4381
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5983480642691e-05,
+      "loss": 0.3908,
+      "step": 4382
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.598160340315996e-05,
+      "loss": 0.4313,
+      "step": 4383
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5979725835337293e-05,
+      "loss": 0.3989,
+      "step": 4384
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5977847939326054e-05,
+      "loss": 0.4133,
+      "step": 4385
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.59759697152293e-05,
+      "loss": 0.4311,
+      "step": 4386
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597409116315012e-05,
+      "loss": 0.4342,
+      "step": 4387
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5972212283191614e-05,
+      "loss": 0.3826,
+      "step": 4388
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5970333075456902e-05,
+      "loss": 0.4253,
+      "step": 4389
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.596845354004912e-05,
+      "loss": 0.4381,
+      "step": 4390
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.596657367707143e-05,
+      "loss": 0.4114,
+      "step": 4391
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5964693486626995e-05,
+      "loss": 0.4052,
+      "step": 4392
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5962812968819018e-05,
+      "loss": 0.3994,
+      "step": 4393
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.59609321237507e-05,
+      "loss": 0.4212,
+      "step": 4394
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5959050951525267e-05,
+      "loss": 0.4212,
+      "step": 4395
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5957169452245973e-05,
+      "loss": 0.4156,
+      "step": 4396
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5955287626016075e-05,
+      "loss": 0.415,
+      "step": 4397
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5953405472938856e-05,
+      "loss": 0.4377,
+      "step": 4398
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5951522993117618e-05,
+      "loss": 0.435,
+      "step": 4399
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5949640186655673e-05,
+      "loss": 0.4236,
+      "step": 4400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594775705365636e-05,
+      "loss": 0.4369,
+      "step": 4401
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594587359422303e-05,
+      "loss": 0.4327,
+      "step": 4402
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5943989808459055e-05,
+      "loss": 0.4208,
+      "step": 4403
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594210569646782e-05,
+      "loss": 0.4199,
+      "step": 4404
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594022125835274e-05,
+      "loss": 0.4466,
+      "step": 4405
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5938336494217238e-05,
+      "loss": 0.4494,
+      "step": 4406
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5936451404164747e-05,
+      "loss": 0.4301,
+      "step": 4407
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5934565988298736e-05,
+      "loss": 0.4343,
+      "step": 4408
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593268024672268e-05,
+      "loss": 0.4289,
+      "step": 4409
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593079417954008e-05,
+      "loss": 0.4338,
+      "step": 4410
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592890778685444e-05,
+      "loss": 0.4486,
+      "step": 4411
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5927021068769303e-05,
+      "loss": 0.4591,
+      "step": 4412
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5925134025388214e-05,
+      "loss": 0.4277,
+      "step": 4413
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592324665681474e-05,
+      "loss": 0.4374,
+      "step": 4414
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5921358963152465e-05,
+      "loss": 0.4272,
+      "step": 4415
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5919470944504996e-05,
+      "loss": 0.4303,
+      "step": 4416
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5917582600975957e-05,
+      "loss": 0.3933,
+      "step": 4417
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591569393266898e-05,
+      "loss": 0.4277,
+      "step": 4418
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5913804939687724e-05,
+      "loss": 0.4367,
+      "step": 4419
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5911915622135864e-05,
+      "loss": 0.4186,
+      "step": 4420
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591002598011709e-05,
+      "loss": 0.4373,
+      "step": 4421
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5908136013735123e-05,
+      "loss": 0.4478,
+      "step": 4422
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5906245723093677e-05,
+      "loss": 0.4452,
+      "step": 4423
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5904355108296506e-05,
+      "loss": 0.4277,
+      "step": 4424
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.590246416944737e-05,
+      "loss": 0.4318,
+      "step": 4425
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.590057290665005e-05,
+      "loss": 0.4412,
+      "step": 4426
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589868132000835e-05,
+      "loss": 0.423,
+      "step": 4427
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5896789409626082e-05,
+      "loss": 0.4088,
+      "step": 4428
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589489717560709e-05,
+      "loss": 0.4529,
+      "step": 4429
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589300461805521e-05,
+      "loss": 0.4273,
+      "step": 4430
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5891111737074326e-05,
+      "loss": 0.4121,
+      "step": 4431
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.588921853276832e-05,
+      "loss": 0.4458,
+      "step": 4432
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5887325005241102e-05,
+      "loss": 0.4386,
+      "step": 4433
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5885431154596595e-05,
+      "loss": 0.401,
+      "step": 4434
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5883536980938736e-05,
+      "loss": 0.4311,
+      "step": 4435
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5881642484371483e-05,
+      "loss": 0.4259,
+      "step": 4436
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5879747664998816e-05,
+      "loss": 0.4435,
+      "step": 4437
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.4347,
+      "step": 4438
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.587595705825324e-05,
+      "loss": 0.4326,
+      "step": 4439
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.587406127108837e-05,
+      "loss": 0.4185,
+      "step": 4440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.587216516153417e-05,
+      "loss": 0.4508,
+      "step": 4441
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.587026872969471e-05,
+      "loss": 0.3972,
+      "step": 4442
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5868371975674063e-05,
+      "loss": 0.4438,
+      "step": 4443
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586647489957633e-05,
+      "loss": 0.4415,
+      "step": 4444
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5864577501505637e-05,
+      "loss": 0.4179,
+      "step": 4445
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5862679781566122e-05,
+      "loss": 0.4496,
+      "step": 4446
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5860781739861928e-05,
+      "loss": 0.4219,
+      "step": 4447
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5858883376497233e-05,
+      "loss": 0.4361,
+      "step": 4448
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5856984691576225e-05,
+      "loss": 0.4173,
+      "step": 4449
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585508568520311e-05,
+      "loss": 0.4386,
+      "step": 4450
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5853186357482115e-05,
+      "loss": 0.4375,
+      "step": 4451
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5851286708517473e-05,
+      "loss": 0.4154,
+      "step": 4452
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5849386738413453e-05,
+      "loss": 0.4151,
+      "step": 4453
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584748644727433e-05,
+      "loss": 0.4275,
+      "step": 4454
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5845585835204395e-05,
+      "loss": 0.4186,
+      "step": 4455
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5843684902307964e-05,
+      "loss": 0.4255,
+      "step": 4456
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5841783648689366e-05,
+      "loss": 0.3899,
+      "step": 4457
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5839882074452947e-05,
+      "loss": 0.4158,
+      "step": 4458
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5837980179703073e-05,
+      "loss": 0.4094,
+      "step": 4459
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583607796454413e-05,
+      "loss": 0.4149,
+      "step": 4460
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5834175429080514e-05,
+      "loss": 0.45,
+      "step": 4461
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5832272573416644e-05,
+      "loss": 0.4238,
+      "step": 4462
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5830369397656956e-05,
+      "loss": 0.4092,
+      "step": 4463
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5828465901905903e-05,
+      "loss": 0.4339,
+      "step": 4464
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5826562086267956e-05,
+      "loss": 0.4317,
+      "step": 4465
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5824657950847603e-05,
+      "loss": 0.4197,
+      "step": 4466
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.582275349574935e-05,
+      "loss": 0.4252,
+      "step": 4467
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5820848721077716e-05,
+      "loss": 0.4156,
+      "step": 4468
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.581894362693725e-05,
+      "loss": 0.4222,
+      "step": 4469
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.58170382134325e-05,
+      "loss": 0.4145,
+      "step": 4470
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5815132480668055e-05,
+      "loss": 0.4302,
+      "step": 4471
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5813226428748495e-05,
+      "loss": 0.4335,
+      "step": 4472
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.581132005777844e-05,
+      "loss": 0.4133,
+      "step": 4473
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5809413367862513e-05,
+      "loss": 0.4488,
+      "step": 4474
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5807506359105365e-05,
+      "loss": 0.3952,
+      "step": 4475
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5805599031611653e-05,
+      "loss": 0.4243,
+      "step": 4476
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.580369138548606e-05,
+      "loss": 0.4968,
+      "step": 4477
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5801783420833295e-05,
+      "loss": 0.4145,
+      "step": 4478
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5799875137758056e-05,
+      "loss": 0.4403,
+      "step": 4479
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5797966536365088e-05,
+      "loss": 0.3987,
+      "step": 4480
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5796057616759138e-05,
+      "loss": 0.4349,
+      "step": 4481
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5794148379044977e-05,
+      "loss": 0.4626,
+      "step": 4482
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5792238823327387e-05,
+      "loss": 0.4682,
+      "step": 4483
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5790328949711176e-05,
+      "loss": 0.4659,
+      "step": 4484
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578841875830116e-05,
+      "loss": 0.4451,
+      "step": 4485
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578650824920218e-05,
+      "loss": 0.4307,
+      "step": 4486
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578459742251909e-05,
+      "loss": 0.4304,
+      "step": 4487
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578268627835676e-05,
+      "loss": 0.4741,
+      "step": 4488
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5780774816820087e-05,
+      "loss": 0.4494,
+      "step": 4489
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5778863038013972e-05,
+      "loss": 0.438,
+      "step": 4490
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.577695094204335e-05,
+      "loss": 0.4473,
+      "step": 4491
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5775038529013154e-05,
+      "loss": 0.4397,
+      "step": 4492
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.577312579902835e-05,
+      "loss": 0.3982,
+      "step": 4493
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5771212752193906e-05,
+      "loss": 0.4387,
+      "step": 4494
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5769299388614825e-05,
+      "loss": 0.4334,
+      "step": 4495
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5767385708396124e-05,
+      "loss": 0.4125,
+      "step": 4496
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5765471711642822e-05,
+      "loss": 0.4229,
+      "step": 4497
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.576355739845997e-05,
+      "loss": 0.4412,
+      "step": 4498
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5761642768952637e-05,
+      "loss": 0.4062,
+      "step": 4499
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5759727823225895e-05,
+      "loss": 0.4155,
+      "step": 4500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5757812561384848e-05,
+      "loss": 0.4139,
+      "step": 4501
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5755896983534616e-05,
+      "loss": 0.4058,
+      "step": 4502
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575398108978033e-05,
+      "loss": 0.4269,
+      "step": 4503
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5752064880227134e-05,
+      "loss": 0.4257,
+      "step": 4504
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5750148354980207e-05,
+      "loss": 0.4346,
+      "step": 4505
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574823151414473e-05,
+      "loss": 0.4013,
+      "step": 4506
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5746314357825908e-05,
+      "loss": 0.4382,
+      "step": 4507
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5744396886128956e-05,
+      "loss": 0.4214,
+      "step": 4508
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5742479099159118e-05,
+      "loss": 0.4175,
+      "step": 4509
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574056099702165e-05,
+      "loss": 0.435,
+      "step": 4510
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5738642579821815e-05,
+      "loss": 0.4476,
+      "step": 4511
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573672384766491e-05,
+      "loss": 0.3973,
+      "step": 4512
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5734804800656238e-05,
+      "loss": 0.4329,
+      "step": 4513
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573288543890113e-05,
+      "loss": 0.4353,
+      "step": 4514
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573096576250492e-05,
+      "loss": 0.4208,
+      "step": 4515
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572904577157297e-05,
+      "loss": 0.3945,
+      "step": 4516
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572712546621065e-05,
+      "loss": 0.4534,
+      "step": 4517
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5725204846523363e-05,
+      "loss": 0.4375,
+      "step": 4518
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5723283912616515e-05,
+      "loss": 0.4131,
+      "step": 4519
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5721362664595533e-05,
+      "loss": 0.4083,
+      "step": 4520
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571944110256586e-05,
+      "loss": 0.4061,
+      "step": 4521
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5717519226632964e-05,
+      "loss": 0.4888,
+      "step": 4522
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5715597036902316e-05,
+      "loss": 0.4403,
+      "step": 4523
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5713674533479423e-05,
+      "loss": 0.4499,
+      "step": 4524
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5711751716469788e-05,
+      "loss": 0.4165,
+      "step": 4525
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570982858597895e-05,
+      "loss": 0.409,
+      "step": 4526
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5707905142112453e-05,
+      "loss": 0.4009,
+      "step": 4527
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5705981384975868e-05,
+      "loss": 0.4248,
+      "step": 4528
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570405731467477e-05,
+      "loss": 0.4259,
+      "step": 4529
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5702132931314763e-05,
+      "loss": 0.4548,
+      "step": 4530
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5700208235001462e-05,
+      "loss": 0.4293,
+      "step": 4531
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5698283225840504e-05,
+      "loss": 0.4043,
+      "step": 4532
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.569635790393754e-05,
+      "loss": 0.4378,
+      "step": 4533
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5694432269398233e-05,
+      "loss": 0.4291,
+      "step": 4534
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5692506322328277e-05,
+      "loss": 0.4556,
+      "step": 4535
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.569058006283337e-05,
+      "loss": 0.4456,
+      "step": 4536
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568865349101923e-05,
+      "loss": 0.4481,
+      "step": 4537
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5686726606991597e-05,
+      "loss": 0.4415,
+      "step": 4538
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5684799410856228e-05,
+      "loss": 0.4205,
+      "step": 4539
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568287190271889e-05,
+      "loss": 0.4151,
+      "step": 4540
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5680944082685376e-05,
+      "loss": 0.4371,
+      "step": 4541
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5679015950861485e-05,
+      "loss": 0.4006,
+      "step": 4542
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5677087507353044e-05,
+      "loss": 0.4187,
+      "step": 4543
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5675158752265892e-05,
+      "loss": 0.4639,
+      "step": 4544
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5673229685705884e-05,
+      "loss": 0.435,
+      "step": 4545
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5671300307778898e-05,
+      "loss": 0.4689,
+      "step": 4546
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5669370618590822e-05,
+      "loss": 0.4328,
+      "step": 4547
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.566744061824757e-05,
+      "loss": 0.4018,
+      "step": 4548
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5665510306855058e-05,
+      "loss": 0.4198,
+      "step": 4549
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5663579684519234e-05,
+      "loss": 0.4141,
+      "step": 4550
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5661648751346055e-05,
+      "loss": 0.3978,
+      "step": 4551
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56597175074415e-05,
+      "loss": 0.4491,
+      "step": 4552
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565778595291156e-05,
+      "loss": 0.4481,
+      "step": 4553
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5655854087862248e-05,
+      "loss": 0.4454,
+      "step": 4554
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565392191239959e-05,
+      "loss": 0.4025,
+      "step": 4555
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5651989426629633e-05,
+      "loss": 0.4547,
+      "step": 4556
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5650056630658436e-05,
+      "loss": 0.4343,
+      "step": 4557
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5648123524592075e-05,
+      "loss": 0.4136,
+      "step": 4558
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5646190108536652e-05,
+      "loss": 0.4032,
+      "step": 4559
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564425638259828e-05,
+      "loss": 0.427,
+      "step": 4560
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564232234688308e-05,
+      "loss": 0.4031,
+      "step": 4561
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564038800149721e-05,
+      "loss": 0.3809,
+      "step": 4562
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5638453346546824e-05,
+      "loss": 0.3968,
+      "step": 4563
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563651838213811e-05,
+      "loss": 0.4466,
+      "step": 4564
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5634583108377257e-05,
+      "loss": 0.4314,
+      "step": 4565
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563264752537049e-05,
+      "loss": 0.4046,
+      "step": 4566
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5630711633224036e-05,
+      "loss": 0.4072,
+      "step": 4567
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5628775432044144e-05,
+      "loss": 0.422,
+      "step": 4568
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562683892193708e-05,
+      "loss": 0.4603,
+      "step": 4569
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562490210300912e-05,
+      "loss": 0.4333,
+      "step": 4570
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562296497536657e-05,
+      "loss": 0.4169,
+      "step": 4571
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562102753911575e-05,
+      "loss": 0.4597,
+      "step": 4572
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.561908979436299e-05,
+      "loss": 0.4001,
+      "step": 4573
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5617151741214635e-05,
+      "loss": 0.4318,
+      "step": 4574
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.561521337977706e-05,
+      "loss": 0.4165,
+      "step": 4575
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5613274710156642e-05,
+      "loss": 0.4083,
+      "step": 4576
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.561133573245979e-05,
+      "loss": 0.4094,
+      "step": 4577
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5609396446792918e-05,
+      "loss": 0.4294,
+      "step": 4578
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5607456853262453e-05,
+      "loss": 0.4484,
+      "step": 4579
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5605516951974863e-05,
+      "loss": 0.445,
+      "step": 4580
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.560357674303661e-05,
+      "loss": 0.4109,
+      "step": 4581
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.560163622655417e-05,
+      "loss": 0.4382,
+      "step": 4582
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5599695402634054e-05,
+      "loss": 0.4349,
+      "step": 4583
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5597754271382782e-05,
+      "loss": 0.3996,
+      "step": 4584
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.559581283290689e-05,
+      "loss": 0.4187,
+      "step": 4585
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.559387108731293e-05,
+      "loss": 0.4535,
+      "step": 4586
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5591929034707468e-05,
+      "loss": 0.4178,
+      "step": 4587
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5589986675197098e-05,
+      "loss": 0.4341,
+      "step": 4588
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558804400888842e-05,
+      "loss": 0.4152,
+      "step": 4589
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558610103588805e-05,
+      "loss": 0.4369,
+      "step": 4590
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5584157756302635e-05,
+      "loss": 0.42,
+      "step": 4591
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558221417023882e-05,
+      "loss": 0.4149,
+      "step": 4592
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558027027780328e-05,
+      "loss": 0.4078,
+      "step": 4593
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.55783260791027e-05,
+      "loss": 0.4243,
+      "step": 4594
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557638157424379e-05,
+      "loss": 0.4145,
+      "step": 4595
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557443676333327e-05,
+      "loss": 0.4086,
+      "step": 4596
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5572491646477872e-05,
+      "loss": 0.4464,
+      "step": 4597
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5570546223784355e-05,
+      "loss": 0.4038,
+      "step": 4598
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5568600495359493e-05,
+      "loss": 0.4829,
+      "step": 4599
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5566654461310073e-05,
+      "loss": 0.3936,
+      "step": 4600
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.55647081217429e-05,
+      "loss": 0.4066,
+      "step": 4601
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5562761476764793e-05,
+      "loss": 0.4016,
+      "step": 4602
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556081452648259e-05,
+      "loss": 0.4146,
+      "step": 4603
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5558867271003158e-05,
+      "loss": 0.4319,
+      "step": 4604
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5556919710433355e-05,
+      "loss": 0.4185,
+      "step": 4605
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.555497184488008e-05,
+      "loss": 0.4308,
+      "step": 4606
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5553023674450233e-05,
+      "loss": 0.3984,
+      "step": 4607
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5551075199250737e-05,
+      "loss": 0.4029,
+      "step": 4608
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5549126419388538e-05,
+      "loss": 0.4441,
+      "step": 4609
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5547177334970583e-05,
+      "loss": 0.4434,
+      "step": 4610
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5545227946103847e-05,
+      "loss": 0.3824,
+      "step": 4611
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554327825289532e-05,
+      "loss": 0.3978,
+      "step": 4612
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5541328255452018e-05,
+      "loss": 0.4039,
+      "step": 4613
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5539377953880945e-05,
+      "loss": 0.4097,
+      "step": 4614
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5537427348289155e-05,
+      "loss": 0.4156,
+      "step": 4615
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5535476438783695e-05,
+      "loss": 0.3938,
+      "step": 4616
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5533525225471647e-05,
+      "loss": 0.4216,
+      "step": 4617
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.553157370846009e-05,
+      "loss": 0.4173,
+      "step": 4618
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.552962188785614e-05,
+      "loss": 0.3687,
+      "step": 4619
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5527669763766914e-05,
+      "loss": 0.4336,
+      "step": 4620
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5525717336299554e-05,
+      "loss": 0.3988,
+      "step": 4621
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5523764605561215e-05,
+      "loss": 0.397,
+      "step": 4622
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5521811571659068e-05,
+      "loss": 0.4639,
+      "step": 4623
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5519858234700306e-05,
+      "loss": 0.4417,
+      "step": 4624
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5517904594792136e-05,
+      "loss": 0.4469,
+      "step": 4625
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5515950652041777e-05,
+      "loss": 0.428,
+      "step": 4626
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5513996406556465e-05,
+      "loss": 0.4478,
+      "step": 4627
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5512041858443466e-05,
+      "loss": 0.451,
+      "step": 4628
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5510087007810044e-05,
+      "loss": 0.4172,
+      "step": 4629
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.550813185476349e-05,
+      "loss": 0.437,
+      "step": 4630
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5506176399411117e-05,
+      "loss": 0.4227,
+      "step": 4631
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5504220641860237e-05,
+      "loss": 0.4303,
+      "step": 4632
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5502264582218193e-05,
+      "loss": 0.4404,
+      "step": 4633
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5500308220592343e-05,
+      "loss": 0.4598,
+      "step": 4634
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5498351557090052e-05,
+      "loss": 0.4398,
+      "step": 4635
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5496394591818718e-05,
+      "loss": 0.436,
+      "step": 4636
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.549443732488574e-05,
+      "loss": 0.4477,
+      "step": 4637
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5492479756398542e-05,
+      "loss": 0.4026,
+      "step": 4638
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.549052188646456e-05,
+      "loss": 0.4173,
+      "step": 4639
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.548856371519125e-05,
+      "loss": 0.4457,
+      "step": 4640
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5486605242686085e-05,
+      "loss": 0.4265,
+      "step": 4641
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5484646469056553e-05,
+      "loss": 0.4112,
+      "step": 4642
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5482687394410153e-05,
+      "loss": 0.4459,
+      "step": 4643
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5480728018854417e-05,
+      "loss": 0.4092,
+      "step": 4644
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5478768342496872e-05,
+      "loss": 0.4387,
+      "step": 4645
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5476808365445073e-05,
+      "loss": 0.4152,
+      "step": 4646
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5474848087806594e-05,
+      "loss": 0.4149,
+      "step": 4647
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5472887509689025e-05,
+      "loss": 0.4231,
+      "step": 4648
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5470926631199966e-05,
+      "loss": 0.4142,
+      "step": 4649
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.546896545244703e-05,
+      "loss": 0.4046,
+      "step": 4650
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5467003973537864e-05,
+      "loss": 0.4306,
+      "step": 4651
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.546504219458012e-05,
+      "loss": 0.4386,
+      "step": 4652
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.546308011568146e-05,
+      "loss": 0.4182,
+      "step": 4653
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5461117736949577e-05,
+      "loss": 0.4292,
+      "step": 4654
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5459155058492168e-05,
+      "loss": 0.4144,
+      "step": 4655
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5457192080416955e-05,
+      "loss": 0.4676,
+      "step": 4656
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5455228802831675e-05,
+      "loss": 0.4276,
+      "step": 4657
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5453265225844077e-05,
+      "loss": 0.4131,
+      "step": 4658
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5451301349561924e-05,
+      "loss": 0.4428,
+      "step": 4659
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5449337174093012e-05,
+      "loss": 0.4171,
+      "step": 4660
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.544737269954513e-05,
+      "loss": 0.4101,
+      "step": 4661
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5445407926026107e-05,
+      "loss": 0.4345,
+      "step": 4662
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5443442853643765e-05,
+      "loss": 0.4709,
+      "step": 4663
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5441477482505964e-05,
+      "loss": 0.424,
+      "step": 4664
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5439511812720564e-05,
+      "loss": 0.4392,
+      "step": 4665
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543754584439545e-05,
+      "loss": 0.4255,
+      "step": 4666
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543557957763852e-05,
+      "loss": 0.392,
+      "step": 4667
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5433613012557697e-05,
+      "loss": 0.4292,
+      "step": 4668
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5431646149260906e-05,
+      "loss": 0.4012,
+      "step": 4669
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5429678987856097e-05,
+      "loss": 0.4225,
+      "step": 4670
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542771152845123e-05,
+      "loss": 0.471,
+      "step": 4671
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5425743771154294e-05,
+      "loss": 0.4266,
+      "step": 4672
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5423775716073283e-05,
+      "loss": 0.4363,
+      "step": 4673
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5421807363316214e-05,
+      "loss": 0.4565,
+      "step": 4674
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541983871299111e-05,
+      "loss": 0.4193,
+      "step": 4675
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5417869765206023e-05,
+      "loss": 0.4235,
+      "step": 4676
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5415900520069017e-05,
+      "loss": 0.4785,
+      "step": 4677
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5413930977688166e-05,
+      "loss": 0.4103,
+      "step": 4678
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541196113817157e-05,
+      "loss": 0.4403,
+      "step": 4679
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.540999100162734e-05,
+      "loss": 0.4141,
+      "step": 4680
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5408020568163602e-05,
+      "loss": 0.4346,
+      "step": 4681
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.54060498378885e-05,
+      "loss": 0.4342,
+      "step": 4682
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5404078810910198e-05,
+      "loss": 0.3989,
+      "step": 4683
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.540210748733687e-05,
+      "loss": 0.4325,
+      "step": 4684
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5400135867276714e-05,
+      "loss": 0.4439,
+      "step": 4685
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5398163950837936e-05,
+      "loss": 0.4286,
+      "step": 4686
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5396191738128758e-05,
+      "loss": 0.4359,
+      "step": 4687
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5394219229257427e-05,
+      "loss": 0.4232,
+      "step": 4688
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.53922464243322e-05,
+      "loss": 0.4144,
+      "step": 4689
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5390273323461354e-05,
+      "loss": 0.4206,
+      "step": 4690
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5388299926753175e-05,
+      "loss": 0.4331,
+      "step": 4691
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5386326234315967e-05,
+      "loss": 0.4704,
+      "step": 4692
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.538435224625807e-05,
+      "loss": 0.4268,
+      "step": 4693
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5382377962687806e-05,
+      "loss": 0.4305,
+      "step": 4694
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5380403383713536e-05,
+      "loss": 0.413,
+      "step": 4695
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5378428509443634e-05,
+      "loss": 0.42,
+      "step": 4696
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5376453339986488e-05,
+      "loss": 0.4099,
+      "step": 4697
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.53744778754505e-05,
+      "loss": 0.4515,
+      "step": 4698
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537250211594409e-05,
+      "loss": 0.4714,
+      "step": 4699
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.53705260615757e-05,
+      "loss": 0.4209,
+      "step": 4700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5368549712453775e-05,
+      "loss": 0.4647,
+      "step": 4701
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536657306868679e-05,
+      "loss": 0.4305,
+      "step": 4702
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536459613038323e-05,
+      "loss": 0.3946,
+      "step": 4703
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5362618897651596e-05,
+      "loss": 0.4073,
+      "step": 4704
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.53606413706004e-05,
+      "loss": 0.416,
+      "step": 4705
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.535866354933818e-05,
+      "loss": 0.4277,
+      "step": 4706
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5356685433973492e-05,
+      "loss": 0.4442,
+      "step": 4707
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5354707024614892e-05,
+      "loss": 0.4421,
+      "step": 4708
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5352728321370964e-05,
+      "loss": 0.4336,
+      "step": 4709
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.535074932435031e-05,
+      "loss": 0.4352,
+      "step": 4710
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5348770033661546e-05,
+      "loss": 0.4338,
+      "step": 4711
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5346790449413296e-05,
+      "loss": 0.4347,
+      "step": 4712
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5344810571714212e-05,
+      "loss": 0.434,
+      "step": 4713
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534283040067295e-05,
+      "loss": 0.4374,
+      "step": 4714
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5340849936398197e-05,
+      "loss": 0.4421,
+      "step": 4715
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5338869178998646e-05,
+      "loss": 0.4143,
+      "step": 4716
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5336888128583002e-05,
+      "loss": 0.4495,
+      "step": 4717
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.533490678526e-05,
+      "loss": 0.4125,
+      "step": 4718
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5332925149138377e-05,
+      "loss": 0.4336,
+      "step": 4719
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5330943220326893e-05,
+      "loss": 0.4074,
+      "step": 4720
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532896099893433e-05,
+      "loss": 0.4297,
+      "step": 4721
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532697848506947e-05,
+      "loss": 0.4264,
+      "step": 4722
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5324995678841128e-05,
+      "loss": 0.4358,
+      "step": 4723
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5323012580358124e-05,
+      "loss": 0.4407,
+      "step": 4724
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5321029189729296e-05,
+      "loss": 0.4048,
+      "step": 4725
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.53190455070635e-05,
+      "loss": 0.4229,
+      "step": 4726
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5317061532469613e-05,
+      "loss": 0.4655,
+      "step": 4727
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5315077266056514e-05,
+      "loss": 0.4491,
+      "step": 4728
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5313092707933114e-05,
+      "loss": 0.4109,
+      "step": 4729
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5311107858208326e-05,
+      "loss": 0.4084,
+      "step": 4730
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5309122716991095e-05,
+      "loss": 0.4373,
+      "step": 4731
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530713728439036e-05,
+      "loss": 0.4116,
+      "step": 4732
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5305151560515098e-05,
+      "loss": 0.4118,
+      "step": 4733
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5303165545474294e-05,
+      "loss": 0.4117,
+      "step": 4734
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5301179239376936e-05,
+      "loss": 0.4533,
+      "step": 4735
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529919264233205e-05,
+      "loss": 0.4133,
+      "step": 4736
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529720575444867e-05,
+      "loss": 0.4124,
+      "step": 4737
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5295218575835828e-05,
+      "loss": 0.4119,
+      "step": 4738
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5293231106602602e-05,
+      "loss": 0.429,
+      "step": 4739
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5291243346858065e-05,
+      "loss": 0.443,
+      "step": 4740
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5289255296711318e-05,
+      "loss": 0.4341,
+      "step": 4741
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5287266956271468e-05,
+      "loss": 0.4411,
+      "step": 4742
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5285278325647638e-05,
+      "loss": 0.4442,
+      "step": 4743
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.528328940494898e-05,
+      "loss": 0.4068,
+      "step": 4744
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5281300194284644e-05,
+      "loss": 0.4193,
+      "step": 4745
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5279310693763815e-05,
+      "loss": 0.4213,
+      "step": 4746
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.527732090349567e-05,
+      "loss": 0.4311,
+      "step": 4747
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5275330823589428e-05,
+      "loss": 0.4479,
+      "step": 4748
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5273340454154314e-05,
+      "loss": 0.4113,
+      "step": 4749
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5271349795299554e-05,
+      "loss": 0.4148,
+      "step": 4750
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5269358847134407e-05,
+      "loss": 0.4345,
+      "step": 4751
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5267367609768148e-05,
+      "loss": 0.4371,
+      "step": 4752
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5265376083310063e-05,
+      "loss": 0.4337,
+      "step": 4753
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5263384267869446e-05,
+      "loss": 0.4277,
+      "step": 4754
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5261392163555625e-05,
+      "loss": 0.4546,
+      "step": 4755
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5259399770477922e-05,
+      "loss": 0.4188,
+      "step": 4756
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5257407088745702e-05,
+      "loss": 0.4351,
+      "step": 4757
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5255414118468316e-05,
+      "loss": 0.4601,
+      "step": 4758
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5253420859755155e-05,
+      "loss": 0.453,
+      "step": 4759
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.525142731271561e-05,
+      "loss": 0.4003,
+      "step": 4760
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5249433477459095e-05,
+      "loss": 0.3873,
+      "step": 4761
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5247439354095041e-05,
+      "loss": 0.4254,
+      "step": 4762
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5245444942732891e-05,
+      "loss": 0.4382,
+      "step": 4763
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5243450243482108e-05,
+      "loss": 0.4347,
+      "step": 4764
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5241455256452162e-05,
+      "loss": 0.4348,
+      "step": 4765
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5239459981752553e-05,
+      "loss": 0.4122,
+      "step": 4766
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5237464419492787e-05,
+      "loss": 0.4187,
+      "step": 4767
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.523546856978238e-05,
+      "loss": 0.4034,
+      "step": 4768
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5233472432730878e-05,
+      "loss": 0.4363,
+      "step": 4769
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5231476008447836e-05,
+      "loss": 0.4334,
+      "step": 4770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5229479297042825e-05,
+      "loss": 0.4155,
+      "step": 4771
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5227482298625427e-05,
+      "loss": 0.3998,
+      "step": 4772
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5225485013305249e-05,
+      "loss": 0.4123,
+      "step": 4773
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5223487441191909e-05,
+      "loss": 0.4332,
+      "step": 4774
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5221489582395036e-05,
+      "loss": 0.4341,
+      "step": 4775
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5219491437024287e-05,
+      "loss": 0.4282,
+      "step": 4776
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5217493005189319e-05,
+      "loss": 0.424,
+      "step": 4777
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5215494286999822e-05,
+      "loss": 0.4053,
+      "step": 4778
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5213495282565487e-05,
+      "loss": 0.4297,
+      "step": 4779
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5211495991996029e-05,
+      "loss": 0.4524,
+      "step": 4780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5209496415401171e-05,
+      "loss": 0.4113,
+      "step": 4781
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5207496552890662e-05,
+      "loss": 0.4655,
+      "step": 4782
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5205496404574263e-05,
+      "loss": 0.4163,
+      "step": 4783
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5203495970561743e-05,
+      "loss": 0.4082,
+      "step": 4784
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5201495250962899e-05,
+      "loss": 0.436,
+      "step": 4785
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5199494245887531e-05,
+      "loss": 0.4339,
+      "step": 4786
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5197492955445471e-05,
+      "loss": 0.4012,
+      "step": 4787
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.519549137974655e-05,
+      "loss": 0.4471,
+      "step": 4788
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.519348951890062e-05,
+      "loss": 0.4074,
+      "step": 4789
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5191487373017553e-05,
+      "loss": 0.4311,
+      "step": 4790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5189484942207235e-05,
+      "loss": 0.4124,
+      "step": 4791
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5187482226579566e-05,
+      "loss": 0.4275,
+      "step": 4792
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5185479226244462e-05,
+      "loss": 0.4442,
+      "step": 4793
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5183475941311852e-05,
+      "loss": 0.4161,
+      "step": 4794
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5181472371891686e-05,
+      "loss": 0.4175,
+      "step": 4795
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5179468518093928e-05,
+      "loss": 0.4342,
+      "step": 4796
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5177464380028553e-05,
+      "loss": 0.4096,
+      "step": 4797
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.517545995780556e-05,
+      "loss": 0.4243,
+      "step": 4798
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5173455251534956e-05,
+      "loss": 0.4251,
+      "step": 4799
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5171450261326768e-05,
+      "loss": 0.4242,
+      "step": 4800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.516944498729103e-05,
+      "loss": 0.4487,
+      "step": 4801
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5167439429537809e-05,
+      "loss": 0.4445,
+      "step": 4802
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5165433588177172e-05,
+      "loss": 0.4382,
+      "step": 4803
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5163427463319207e-05,
+      "loss": 0.4268,
+      "step": 4804
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5161421055074015e-05,
+      "loss": 0.4475,
+      "step": 4805
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.515941436355172e-05,
+      "loss": 0.4665,
+      "step": 4806
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5157407388862452e-05,
+      "loss": 0.4545,
+      "step": 4807
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5155400131116364e-05,
+      "loss": 0.4126,
+      "step": 4808
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5153392590423617e-05,
+      "loss": 0.445,
+      "step": 4809
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5151384766894398e-05,
+      "loss": 0.415,
+      "step": 4810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5149376660638897e-05,
+      "loss": 0.4164,
+      "step": 4811
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.514736827176733e-05,
+      "loss": 0.4574,
+      "step": 4812
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5145359600389925e-05,
+      "loss": 0.4227,
+      "step": 4813
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5143350646616923e-05,
+      "loss": 0.4448,
+      "step": 4814
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5141341410558581e-05,
+      "loss": 0.4267,
+      "step": 4815
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.513933189232518e-05,
+      "loss": 0.4269,
+      "step": 4816
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5137322092027001e-05,
+      "loss": 0.4523,
+      "step": 4817
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5135312009774354e-05,
+      "loss": 0.4541,
+      "step": 4818
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5133301645677558e-05,
+      "loss": 0.4297,
+      "step": 4819
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.513129099984695e-05,
+      "loss": 0.4226,
+      "step": 4820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.512928007239288e-05,
+      "loss": 0.4386,
+      "step": 4821
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5127268863425716e-05,
+      "loss": 0.4266,
+      "step": 4822
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5125257373055839e-05,
+      "loss": 0.4389,
+      "step": 4823
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5123245601393644e-05,
+      "loss": 0.436,
+      "step": 4824
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.512123354854955e-05,
+      "loss": 0.4196,
+      "step": 4825
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5119221214633983e-05,
+      "loss": 0.4004,
+      "step": 4826
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5117208599757391e-05,
+      "loss": 0.4146,
+      "step": 4827
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5115195704030225e-05,
+      "loss": 0.4261,
+      "step": 4828
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5113182527562964e-05,
+      "loss": 0.4055,
+      "step": 4829
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5111169070466101e-05,
+      "loss": 0.4392,
+      "step": 4830
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5109155332850138e-05,
+      "loss": 0.4183,
+      "step": 4831
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5107141314825597e-05,
+      "loss": 0.4492,
+      "step": 4832
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5105127016503017e-05,
+      "loss": 0.4022,
+      "step": 4833
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5103112437992948e-05,
+      "loss": 0.4447,
+      "step": 4834
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5101097579405957e-05,
+      "loss": 0.4136,
+      "step": 4835
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5099082440852622e-05,
+      "loss": 0.4515,
+      "step": 4836
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5097067022443549e-05,
+      "loss": 0.4207,
+      "step": 4837
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5095051324289347e-05,
+      "loss": 0.3891,
+      "step": 4838
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5093035346500648e-05,
+      "loss": 0.414,
+      "step": 4839
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.509101908918809e-05,
+      "loss": 0.4124,
+      "step": 4840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5089002552462337e-05,
+      "loss": 0.4423,
+      "step": 4841
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5086985736434062e-05,
+      "loss": 0.4289,
+      "step": 4842
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5084968641213958e-05,
+      "loss": 0.435,
+      "step": 4843
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5082951266912727e-05,
+      "loss": 0.4494,
+      "step": 4844
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.508093361364109e-05,
+      "loss": 0.4185,
+      "step": 4845
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5078915681509782e-05,
+      "loss": 0.4413,
+      "step": 4846
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5076897470629557e-05,
+      "loss": 0.4543,
+      "step": 4847
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5074878981111183e-05,
+      "loss": 0.4145,
+      "step": 4848
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5072860213065434e-05,
+      "loss": 0.458,
+      "step": 4849
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5070841166603116e-05,
+      "loss": 0.4321,
+      "step": 4850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5068821841835038e-05,
+      "loss": 0.4538,
+      "step": 4851
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5066802238872024e-05,
+      "loss": 0.4037,
+      "step": 4852
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5064782357824922e-05,
+      "loss": 0.4385,
+      "step": 4853
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5062762198804591e-05,
+      "loss": 0.4067,
+      "step": 4854
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5060741761921903e-05,
+      "loss": 0.4617,
+      "step": 4855
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5058721047287743e-05,
+      "loss": 0.4292,
+      "step": 4856
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.505670005501302e-05,
+      "loss": 0.4421,
+      "step": 4857
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.505467878520865e-05,
+      "loss": 0.4069,
+      "step": 4858
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5052657237985568e-05,
+      "loss": 0.4235,
+      "step": 4859
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5050635413454729e-05,
+      "loss": 0.439,
+      "step": 4860
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5048613311727089e-05,
+      "loss": 0.4072,
+      "step": 4861
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5046590932913632e-05,
+      "loss": 0.4642,
+      "step": 4862
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5044568277125356e-05,
+      "loss": 0.4065,
+      "step": 4863
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.504254534447327e-05,
+      "loss": 0.4026,
+      "step": 4864
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5040522135068397e-05,
+      "loss": 0.4337,
+      "step": 4865
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503849864902178e-05,
+      "loss": 0.388,
+      "step": 4866
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5036474886444472e-05,
+      "loss": 0.4102,
+      "step": 4867
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5034450847447554e-05,
+      "loss": 0.4016,
+      "step": 4868
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5032426532142101e-05,
+      "loss": 0.4139,
+      "step": 4869
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5030401940639221e-05,
+      "loss": 0.4409,
+      "step": 4870
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5028377073050026e-05,
+      "loss": 0.4276,
+      "step": 4871
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5026351929485657e-05,
+      "loss": 0.5289,
+      "step": 4872
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5024326510057253e-05,
+      "loss": 0.4622,
+      "step": 4873
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5022300814875977e-05,
+      "loss": 0.4355,
+      "step": 4874
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5020274844053004e-05,
+      "loss": 0.4519,
+      "step": 4875
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5018248597699537e-05,
+      "loss": 0.4304,
+      "step": 4876
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5016222075926772e-05,
+      "loss": 0.4387,
+      "step": 4877
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5014195278845937e-05,
+      "loss": 0.4436,
+      "step": 4878
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.501216820656827e-05,
+      "loss": 0.434,
+      "step": 4879
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5010140859205024e-05,
+      "loss": 0.4443,
+      "step": 4880
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5008113236867465e-05,
+      "loss": 0.4293,
+      "step": 4881
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5006085339666879e-05,
+      "loss": 0.4284,
+      "step": 4882
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5004057167714563e-05,
+      "loss": 0.4442,
+      "step": 4883
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5002028721121825e-05,
+      "loss": 0.4184,
+      "step": 4884
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.4268,
+      "step": 4885
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4997971004460432e-05,
+      "loss": 0.4532,
+      "step": 4886
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4995941734614476e-05,
+      "loss": 0.3976,
+      "step": 4887
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4993912190573506e-05,
+      "loss": 0.4648,
+      "step": 4888
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4991882372448912e-05,
+      "loss": 0.4222,
+      "step": 4889
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4989852280352094e-05,
+      "loss": 0.4455,
+      "step": 4890
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4987821914394475e-05,
+      "loss": 0.4454,
+      "step": 4891
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4985791274687488e-05,
+      "loss": 0.4282,
+      "step": 4892
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.498376036134258e-05,
+      "loss": 0.3887,
+      "step": 4893
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4981729174471213e-05,
+      "loss": 0.4253,
+      "step": 4894
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4979697714184872e-05,
+      "loss": 0.4089,
+      "step": 4895
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4977665980595047e-05,
+      "loss": 0.4501,
+      "step": 4896
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4975633973813244e-05,
+      "loss": 0.4137,
+      "step": 4897
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.497360169395099e-05,
+      "loss": 0.4205,
+      "step": 4898
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4971569141119827e-05,
+      "loss": 0.402,
+      "step": 4899
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4969536315431298e-05,
+      "loss": 0.4137,
+      "step": 4900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4967503216996987e-05,
+      "loss": 0.4417,
+      "step": 4901
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4965469845928462e-05,
+      "loss": 0.4453,
+      "step": 4902
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4963436202337333e-05,
+      "loss": 0.4449,
+      "step": 4903
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.496140228633521e-05,
+      "loss": 0.4194,
+      "step": 4904
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4959368098033718e-05,
+      "loss": 0.4445,
+      "step": 4905
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4957333637544505e-05,
+      "loss": 0.4321,
+      "step": 4906
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4955298904979227e-05,
+      "loss": 0.4291,
+      "step": 4907
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4953263900449562e-05,
+      "loss": 0.454,
+      "step": 4908
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4951228624067189e-05,
+      "loss": 0.4375,
+      "step": 4909
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.494919307594382e-05,
+      "loss": 0.4206,
+      "step": 4910
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.494715725619117e-05,
+      "loss": 0.4333,
+      "step": 4911
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.494512116492097e-05,
+      "loss": 0.4008,
+      "step": 4912
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.494308480224497e-05,
+      "loss": 0.4162,
+      "step": 4913
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4941048168274934e-05,
+      "loss": 0.4214,
+      "step": 4914
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4939011263122635e-05,
+      "loss": 0.4055,
+      "step": 4915
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4936974086899873e-05,
+      "loss": 0.4002,
+      "step": 4916
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4934936639718449e-05,
+      "loss": 0.4461,
+      "step": 4917
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4932898921690189e-05,
+      "loss": 0.4339,
+      "step": 4918
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4930860932926927e-05,
+      "loss": 0.4564,
+      "step": 4919
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.492882267354052e-05,
+      "loss": 0.4332,
+      "step": 4920
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4926784143642827e-05,
+      "loss": 0.4284,
+      "step": 4921
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.492474534334574e-05,
+      "loss": 0.4247,
+      "step": 4922
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4922706272761145e-05,
+      "loss": 0.4341,
+      "step": 4923
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4920666932000963e-05,
+      "loss": 0.4648,
+      "step": 4924
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4918627321177115e-05,
+      "loss": 0.4266,
+      "step": 4925
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.491658744040154e-05,
+      "loss": 0.4463,
+      "step": 4926
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4914547289786197e-05,
+      "loss": 0.4333,
+      "step": 4927
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.491250686944306e-05,
+      "loss": 0.4392,
+      "step": 4928
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4910466179484109e-05,
+      "loss": 0.4614,
+      "step": 4929
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4908425220021346e-05,
+      "loss": 0.4281,
+      "step": 4930
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4906383991166786e-05,
+      "loss": 0.3956,
+      "step": 4931
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.490434249303246e-05,
+      "loss": 0.4263,
+      "step": 4932
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4902300725730414e-05,
+      "loss": 0.4196,
+      "step": 4933
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4900258689372701e-05,
+      "loss": 0.4236,
+      "step": 4934
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4898216384071402e-05,
+      "loss": 0.4234,
+      "step": 4935
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4896173809938606e-05,
+      "loss": 0.4368,
+      "step": 4936
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.489413096708641e-05,
+      "loss": 0.4657,
+      "step": 4937
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.489208785562694e-05,
+      "loss": 0.4481,
+      "step": 4938
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4890044475672322e-05,
+      "loss": 0.4512,
+      "step": 4939
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4888000827334712e-05,
+      "loss": 0.4358,
+      "step": 4940
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.488595691072627e-05,
+      "loss": 0.449,
+      "step": 4941
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4883912725959168e-05,
+      "loss": 0.4025,
+      "step": 4942
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4881868273145601e-05,
+      "loss": 0.4432,
+      "step": 4943
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.487982355239778e-05,
+      "loss": 0.4259,
+      "step": 4944
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4877778563827925e-05,
+      "loss": 0.4483,
+      "step": 4945
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4875733307548269e-05,
+      "loss": 0.4427,
+      "step": 4946
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.487368778367106e-05,
+      "loss": 0.4355,
+      "step": 4947
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4871641992308575e-05,
+      "loss": 0.4162,
+      "step": 4948
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4869595933573084e-05,
+      "loss": 0.4344,
+      "step": 4949
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4867549607576887e-05,
+      "loss": 0.4504,
+      "step": 4950
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4865503014432291e-05,
+      "loss": 0.4273,
+      "step": 4951
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.486345615425162e-05,
+      "loss": 0.4395,
+      "step": 4952
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4861409027147217e-05,
+      "loss": 0.4068,
+      "step": 4953
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4859361633231435e-05,
+      "loss": 0.4105,
+      "step": 4954
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4857313972616635e-05,
+      "loss": 0.4372,
+      "step": 4955
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4855266045415207e-05,
+      "loss": 0.4167,
+      "step": 4956
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4853217851739547e-05,
+      "loss": 0.3999,
+      "step": 4957
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.485116939170207e-05,
+      "loss": 0.42,
+      "step": 4958
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4849120665415193e-05,
+      "loss": 0.4177,
+      "step": 4959
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4847071672991366e-05,
+      "loss": 0.4867,
+      "step": 4960
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4845022414543047e-05,
+      "loss": 0.4137,
+      "step": 4961
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4842972890182698e-05,
+      "loss": 0.4115,
+      "step": 4962
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.484092310002281e-05,
+      "loss": 0.4384,
+      "step": 4963
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4838873044175881e-05,
+      "loss": 0.4318,
+      "step": 4964
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4836822722754427e-05,
+      "loss": 0.4325,
+      "step": 4965
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4834772135870977e-05,
+      "loss": 0.4217,
+      "step": 4966
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4832721283638068e-05,
+      "loss": 0.4284,
+      "step": 4967
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4830670166168265e-05,
+      "loss": 0.4418,
+      "step": 4968
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.482861878357414e-05,
+      "loss": 0.4601,
+      "step": 4969
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.482656713596828e-05,
+      "loss": 0.4289,
+      "step": 4970
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4824515223463283e-05,
+      "loss": 0.4199,
+      "step": 4971
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4822463046171769e-05,
+      "loss": 0.4292,
+      "step": 4972
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4820410604206367e-05,
+      "loss": 0.4591,
+      "step": 4973
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4818357897679725e-05,
+      "loss": 0.4387,
+      "step": 4974
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4816304926704499e-05,
+      "loss": 0.4564,
+      "step": 4975
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4814251691393365e-05,
+      "loss": 0.4508,
+      "step": 4976
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.481219819185901e-05,
+      "loss": 0.4246,
+      "step": 4977
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4810144428214143e-05,
+      "loss": 0.4377,
+      "step": 4978
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4808090400571478e-05,
+      "loss": 0.4086,
+      "step": 4979
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4806036109043747e-05,
+      "loss": 0.4228,
+      "step": 4980
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4803981553743694e-05,
+      "loss": 0.4271,
+      "step": 4981
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4801926734784086e-05,
+      "loss": 0.4207,
+      "step": 4982
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4799871652277697e-05,
+      "loss": 0.4477,
+      "step": 4983
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4797816306337315e-05,
+      "loss": 0.436,
+      "step": 4984
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4795760697075746e-05,
+      "loss": 0.4645,
+      "step": 4985
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.479370482460581e-05,
+      "loss": 0.452,
+      "step": 4986
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.479164868904034e-05,
+      "loss": 0.4861,
+      "step": 4987
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4789592290492185e-05,
+      "loss": 0.4305,
+      "step": 4988
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4787535629074205e-05,
+      "loss": 0.4668,
+      "step": 4989
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4785478704899276e-05,
+      "loss": 0.4406,
+      "step": 4990
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4783421518080296e-05,
+      "loss": 0.4274,
+      "step": 4991
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4781364068730165e-05,
+      "loss": 0.4085,
+      "step": 4992
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4779306356961802e-05,
+      "loss": 0.4286,
+      "step": 4993
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4777248382888145e-05,
+      "loss": 0.4313,
+      "step": 4994
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4775190146622143e-05,
+      "loss": 0.4376,
+      "step": 4995
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4773131648276757e-05,
+      "loss": 0.437,
+      "step": 4996
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4771072887964967e-05,
+      "loss": 0.4072,
+      "step": 4997
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4769013865799761e-05,
+      "loss": 0.42,
+      "step": 4998
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4766954581894152e-05,
+      "loss": 0.4355,
+      "step": 4999
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4764895036361155e-05,
+      "loss": 0.4031,
+      "step": 5000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.476283522931381e-05,
+      "loss": 0.4241,
+      "step": 5001
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.476077516086516e-05,
+      "loss": 0.4101,
+      "step": 5002
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4758714831128275e-05,
+      "loss": 0.443,
+      "step": 5003
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4756654240216231e-05,
+      "loss": 0.4563,
+      "step": 5004
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4754593388242117e-05,
+      "loss": 0.4303,
+      "step": 5005
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4752532275319048e-05,
+      "loss": 0.4394,
+      "step": 5006
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.475047090156014e-05,
+      "loss": 0.4265,
+      "step": 5007
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4748409267078525e-05,
+      "loss": 0.4392,
+      "step": 5008
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.474634737198736e-05,
+      "loss": 0.4463,
+      "step": 5009
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4744285216399807e-05,
+      "loss": 0.4511,
+      "step": 5010
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4742222800429043e-05,
+      "loss": 0.4221,
+      "step": 5011
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4740160124188256e-05,
+      "loss": 0.4143,
+      "step": 5012
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4738097187790663e-05,
+      "loss": 0.4381,
+      "step": 5013
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4736033991349482e-05,
+      "loss": 0.4412,
+      "step": 5014
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4733970534977942e-05,
+      "loss": 0.4195,
+      "step": 5015
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.47319068187893e-05,
+      "loss": 0.4378,
+      "step": 5016
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4729842842896818e-05,
+      "loss": 0.4384,
+      "step": 5017
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4727778607413774e-05,
+      "loss": 0.4197,
+      "step": 5018
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.472571411245346e-05,
+      "loss": 0.3985,
+      "step": 5019
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4723649358129184e-05,
+      "loss": 0.415,
+      "step": 5020
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4721584344554268e-05,
+      "loss": 0.4609,
+      "step": 5021
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4719519071842044e-05,
+      "loss": 0.4152,
+      "step": 5022
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4717453540105862e-05,
+      "loss": 0.4334,
+      "step": 5023
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4715387749459088e-05,
+      "loss": 0.4287,
+      "step": 5024
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4713321700015101e-05,
+      "loss": 0.4386,
+      "step": 5025
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4711255391887292e-05,
+      "loss": 0.4256,
+      "step": 5026
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4709188825189062e-05,
+      "loss": 0.4264,
+      "step": 5027
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4707122000033836e-05,
+      "loss": 0.4497,
+      "step": 5028
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4705054916535052e-05,
+      "loss": 0.4278,
+      "step": 5029
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4702987574806154e-05,
+      "loss": 0.4058,
+      "step": 5030
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4700919974960608e-05,
+      "loss": 0.4059,
+      "step": 5031
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4698852117111886e-05,
+      "loss": 0.4174,
+      "step": 5032
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4696784001373485e-05,
+      "loss": 0.4054,
+      "step": 5033
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.469471562785891e-05,
+      "loss": 0.43,
+      "step": 5034
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4692646996681678e-05,
+      "loss": 0.4354,
+      "step": 5035
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4690578107955324e-05,
+      "loss": 0.4287,
+      "step": 5036
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4688508961793394e-05,
+      "loss": 0.4678,
+      "step": 5037
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4686439558309457e-05,
+      "loss": 0.4209,
+      "step": 5038
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4684369897617084e-05,
+      "loss": 0.4251,
+      "step": 5039
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4682299979829862e-05,
+      "loss": 0.4387,
+      "step": 5040
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4680229805061399e-05,
+      "loss": 0.4061,
+      "step": 5041
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4678159373425318e-05,
+      "loss": 0.4154,
+      "step": 5042
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4676088685035245e-05,
+      "loss": 0.4261,
+      "step": 5043
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4674017740004829e-05,
+      "loss": 0.4108,
+      "step": 5044
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.467194653844773e-05,
+      "loss": 0.4844,
+      "step": 5045
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4669875080477625e-05,
+      "loss": 0.407,
+      "step": 5046
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4667803366208202e-05,
+      "loss": 0.4104,
+      "step": 5047
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4665731395753164e-05,
+      "loss": 0.4251,
+      "step": 5048
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4663659169226225e-05,
+      "loss": 0.3737,
+      "step": 5049
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.466158668674112e-05,
+      "loss": 0.4243,
+      "step": 5050
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4659513948411596e-05,
+      "loss": 0.4215,
+      "step": 5051
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4657440954351409e-05,
+      "loss": 0.397,
+      "step": 5052
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.465536770467433e-05,
+      "loss": 0.4182,
+      "step": 5053
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.465329419949415e-05,
+      "loss": 0.4316,
+      "step": 5054
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4651220438924671e-05,
+      "loss": 0.4163,
+      "step": 5055
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4649146423079705e-05,
+      "loss": 0.4244,
+      "step": 5056
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4647072152073083e-05,
+      "loss": 0.4526,
+      "step": 5057
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.464499762601865e-05,
+      "loss": 0.4173,
+      "step": 5058
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4642922845030258e-05,
+      "loss": 0.4136,
+      "step": 5059
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4640847809221787e-05,
+      "loss": 0.4295,
+      "step": 5060
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4638772518707113e-05,
+      "loss": 0.4279,
+      "step": 5061
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4636696973600141e-05,
+      "loss": 0.4741,
+      "step": 5062
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4634621174014785e-05,
+      "loss": 0.465,
+      "step": 5063
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.463254512006497e-05,
+      "loss": 0.4325,
+      "step": 5064
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4630468811864633e-05,
+      "loss": 0.3972,
+      "step": 5065
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4628392249527736e-05,
+      "loss": 0.417,
+      "step": 5066
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.462631543316825e-05,
+      "loss": 0.4086,
+      "step": 5067
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.462423836290015e-05,
+      "loss": 0.4157,
+      "step": 5068
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4622161038837437e-05,
+      "loss": 0.4763,
+      "step": 5069
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4620083461094119e-05,
+      "loss": 0.413,
+      "step": 5070
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.461800562978423e-05,
+      "loss": 0.4156,
+      "step": 5071
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4615927545021797e-05,
+      "loss": 0.3987,
+      "step": 5072
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4613849206920879e-05,
+      "loss": 0.4417,
+      "step": 5073
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4611770615595542e-05,
+      "loss": 0.3918,
+      "step": 5074
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4609691771159866e-05,
+      "loss": 0.3997,
+      "step": 5075
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4607612673727946e-05,
+      "loss": 0.4136,
+      "step": 5076
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4605533323413887e-05,
+      "loss": 0.4072,
+      "step": 5077
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4603453720331817e-05,
+      "loss": 0.4428,
+      "step": 5078
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4601373864595867e-05,
+      "loss": 0.4251,
+      "step": 5079
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.459929375632019e-05,
+      "loss": 0.4181,
+      "step": 5080
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4597213395618946e-05,
+      "loss": 0.408,
+      "step": 5081
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4595132782606318e-05,
+      "loss": 0.4086,
+      "step": 5082
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.459305191739649e-05,
+      "loss": 0.4121,
+      "step": 5083
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4590970800103676e-05,
+      "loss": 0.422,
+      "step": 5084
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4588889430842086e-05,
+      "loss": 0.4495,
+      "step": 5085
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4586807809725962e-05,
+      "loss": 0.4266,
+      "step": 5086
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4584725936869544e-05,
+      "loss": 0.4306,
+      "step": 5087
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4582643812387097e-05,
+      "loss": 0.4351,
+      "step": 5088
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4580561436392891e-05,
+      "loss": 0.4666,
+      "step": 5089
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4578478809001218e-05,
+      "loss": 0.4228,
+      "step": 5090
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4576395930326377e-05,
+      "loss": 0.434,
+      "step": 5091
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4574312800482686e-05,
+      "loss": 0.417,
+      "step": 5092
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4572229419584474e-05,
+      "loss": 0.3969,
+      "step": 5093
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4570145787746083e-05,
+      "loss": 0.4118,
+      "step": 5094
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4568061905081874e-05,
+      "loss": 0.4331,
+      "step": 5095
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4565977771706212e-05,
+      "loss": 0.4182,
+      "step": 5096
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4563893387733488e-05,
+      "loss": 0.4112,
+      "step": 5097
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4561808753278094e-05,
+      "loss": 0.4335,
+      "step": 5098
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4559723868454448e-05,
+      "loss": 0.4452,
+      "step": 5099
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.455763873337697e-05,
+      "loss": 0.4394,
+      "step": 5100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4555553348160106e-05,
+      "loss": 0.4536,
+      "step": 5101
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4553467712918306e-05,
+      "loss": 0.4568,
+      "step": 5102
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4551381827766035e-05,
+      "loss": 0.4045,
+      "step": 5103
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4549295692817778e-05,
+      "loss": 0.4132,
+      "step": 5104
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4547209308188023e-05,
+      "loss": 0.4402,
+      "step": 5105
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4545122673991288e-05,
+      "loss": 0.4172,
+      "step": 5106
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4543035790342091e-05,
+      "loss": 0.4316,
+      "step": 5107
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.454094865735496e-05,
+      "loss": 0.4331,
+      "step": 5108
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.453886127514445e-05,
+      "loss": 0.4535,
+      "step": 5109
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4536773643825131e-05,
+      "loss": 0.4097,
+      "step": 5110
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.453468576351157e-05,
+      "loss": 0.4173,
+      "step": 5111
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4532597634318362e-05,
+      "loss": 0.4014,
+      "step": 5112
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4530509256360104e-05,
+      "loss": 0.4476,
+      "step": 5113
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4528420629751421e-05,
+      "loss": 0.442,
+      "step": 5114
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4526331754606943e-05,
+      "loss": 0.4399,
+      "step": 5115
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4524242631041317e-05,
+      "loss": 0.4313,
+      "step": 5116
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4522153259169194e-05,
+      "loss": 0.4243,
+      "step": 5117
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4520063639105248e-05,
+      "loss": 0.429,
+      "step": 5118
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4517973770964173e-05,
+      "loss": 0.4228,
+      "step": 5119
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4515883654860663e-05,
+      "loss": 0.4494,
+      "step": 5120
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4513793290909428e-05,
+      "loss": 0.4062,
+      "step": 5121
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4511702679225194e-05,
+      "loss": 0.4197,
+      "step": 5122
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4509611819922712e-05,
+      "loss": 0.4114,
+      "step": 5123
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4507520713116722e-05,
+      "loss": 0.4322,
+      "step": 5124
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4505429358922001e-05,
+      "loss": 0.4471,
+      "step": 5125
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4503337757453325e-05,
+      "loss": 0.4311,
+      "step": 5126
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4501245908825492e-05,
+      "loss": 0.4101,
+      "step": 5127
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4499153813153308e-05,
+      "loss": 0.4054,
+      "step": 5128
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4497061470551592e-05,
+      "loss": 0.4323,
+      "step": 5129
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4494968881135183e-05,
+      "loss": 0.4461,
+      "step": 5130
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.449287604501893e-05,
+      "loss": 0.4586,
+      "step": 5131
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4490782962317693e-05,
+      "loss": 0.4466,
+      "step": 5132
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4488689633146351e-05,
+      "loss": 0.4292,
+      "step": 5133
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4486596057619786e-05,
+      "loss": 0.4004,
+      "step": 5134
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4484502235852911e-05,
+      "loss": 0.4083,
+      "step": 5135
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4482408167960635e-05,
+      "loss": 0.4281,
+      "step": 5136
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4480313854057893e-05,
+      "loss": 0.4236,
+      "step": 5137
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.447821929425962e-05,
+      "loss": 0.447,
+      "step": 5138
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4476124488680782e-05,
+      "loss": 0.4386,
+      "step": 5139
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4474029437436347e-05,
+      "loss": 0.4141,
+      "step": 5140
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4471934140641297e-05,
+      "loss": 0.4395,
+      "step": 5141
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4469838598410628e-05,
+      "loss": 0.4204,
+      "step": 5142
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4467742810859355e-05,
+      "loss": 0.4246,
+      "step": 5143
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.44656467781025e-05,
+      "loss": 0.4517,
+      "step": 5144
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4463550500255102e-05,
+      "loss": 0.4414,
+      "step": 5145
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4461453977432208e-05,
+      "loss": 0.4398,
+      "step": 5146
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4459357209748883e-05,
+      "loss": 0.4425,
+      "step": 5147
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4457260197320213e-05,
+      "loss": 0.4277,
+      "step": 5148
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4455162940261284e-05,
+      "loss": 0.4189,
+      "step": 5149
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4453065438687198e-05,
+      "loss": 0.4302,
+      "step": 5150
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4450967692713076e-05,
+      "loss": 0.4145,
+      "step": 5151
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4448869702454054e-05,
+      "loss": 0.4103,
+      "step": 5152
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4446771468025271e-05,
+      "loss": 0.3954,
+      "step": 5153
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4444672989541886e-05,
+      "loss": 0.4387,
+      "step": 5154
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4442574267119074e-05,
+      "loss": 0.431,
+      "step": 5155
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4440475300872022e-05,
+      "loss": 0.4337,
+      "step": 5156
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4438376090915924e-05,
+      "loss": 0.4158,
+      "step": 5157
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4436276637365991e-05,
+      "loss": 0.4177,
+      "step": 5158
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4434176940337454e-05,
+      "loss": 0.4254,
+      "step": 5159
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.443207699994555e-05,
+      "loss": 0.4225,
+      "step": 5160
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4429976816305531e-05,
+      "loss": 0.387,
+      "step": 5161
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4427876389532659e-05,
+      "loss": 0.4518,
+      "step": 5162
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4425775719742217e-05,
+      "loss": 0.4114,
+      "step": 5163
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4423674807049495e-05,
+      "loss": 0.4171,
+      "step": 5164
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.44215736515698e-05,
+      "loss": 0.4592,
+      "step": 5165
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4419472253418451e-05,
+      "loss": 0.426,
+      "step": 5166
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.441737061271078e-05,
+      "loss": 0.3949,
+      "step": 5167
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.441526872956213e-05,
+      "loss": 0.4026,
+      "step": 5168
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4413166604087865e-05,
+      "loss": 0.4149,
+      "step": 5169
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4411064236403351e-05,
+      "loss": 0.4482,
+      "step": 5170
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4408961626623979e-05,
+      "loss": 0.4581,
+      "step": 5171
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4406858774865141e-05,
+      "loss": 0.4163,
+      "step": 5172
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4404755681242257e-05,
+      "loss": 0.4455,
+      "step": 5173
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4402652345870746e-05,
+      "loss": 0.3959,
+      "step": 5174
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4400548768866052e-05,
+      "loss": 0.4167,
+      "step": 5175
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4398444950343624e-05,
+      "loss": 0.4655,
+      "step": 5176
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4396340890418923e-05,
+      "loss": 0.4261,
+      "step": 5177
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4394236589207431e-05,
+      "loss": 0.4021,
+      "step": 5178
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.439213204682464e-05,
+      "loss": 0.4247,
+      "step": 5179
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4390027263386057e-05,
+      "loss": 0.4686,
+      "step": 5180
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4387922239007192e-05,
+      "loss": 0.4312,
+      "step": 5181
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4385816973803584e-05,
+      "loss": 0.4182,
+      "step": 5182
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4383711467890776e-05,
+      "loss": 0.4275,
+      "step": 5183
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4381605721384324e-05,
+      "loss": 0.4492,
+      "step": 5184
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4379499734399797e-05,
+      "loss": 0.3806,
+      "step": 5185
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4377393507052784e-05,
+      "loss": 0.4496,
+      "step": 5186
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4375287039458878e-05,
+      "loss": 0.4079,
+      "step": 5187
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4373180331733691e-05,
+      "loss": 0.4345,
+      "step": 5188
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4371073383992845e-05,
+      "loss": 0.4061,
+      "step": 5189
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4368966196351983e-05,
+      "loss": 0.4169,
+      "step": 5190
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4366858768926746e-05,
+      "loss": 0.4197,
+      "step": 5191
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4364751101832804e-05,
+      "loss": 0.4217,
+      "step": 5192
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4362643195185828e-05,
+      "loss": 0.4267,
+      "step": 5193
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4360535049101512e-05,
+      "loss": 0.4426,
+      "step": 5194
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4358426663695555e-05,
+      "loss": 0.4481,
+      "step": 5195
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4356318039083675e-05,
+      "loss": 0.4291,
+      "step": 5196
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.43542091753816e-05,
+      "loss": 0.4304,
+      "step": 5197
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4352100072705072e-05,
+      "loss": 0.4391,
+      "step": 5198
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4349990731169846e-05,
+      "loss": 0.4214,
+      "step": 5199
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4347881150891691e-05,
+      "loss": 0.3971,
+      "step": 5200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4345771331986386e-05,
+      "loss": 0.41,
+      "step": 5201
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4343661274569727e-05,
+      "loss": 0.4432,
+      "step": 5202
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4341550978757521e-05,
+      "loss": 0.4472,
+      "step": 5203
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4339440444665588e-05,
+      "loss": 0.4422,
+      "step": 5204
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4337329672409765e-05,
+      "loss": 0.4163,
+      "step": 5205
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4335218662105894e-05,
+      "loss": 0.4509,
+      "step": 5206
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4333107413869835e-05,
+      "loss": 0.442,
+      "step": 5207
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4330995927817466e-05,
+      "loss": 0.4473,
+      "step": 5208
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4328884204064668e-05,
+      "loss": 0.4024,
+      "step": 5209
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.432677224272734e-05,
+      "loss": 0.4645,
+      "step": 5210
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4324660043921395e-05,
+      "loss": 0.4396,
+      "step": 5211
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4322547607762763e-05,
+      "loss": 0.4165,
+      "step": 5212
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4320434934367373e-05,
+      "loss": 0.4498,
+      "step": 5213
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4318322023851179e-05,
+      "loss": 0.4269,
+      "step": 5214
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4316208876330147e-05,
+      "loss": 0.4679,
+      "step": 5215
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4314095491920254e-05,
+      "loss": 0.4606,
+      "step": 5216
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.431198187073749e-05,
+      "loss": 0.4544,
+      "step": 5217
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4309868012897858e-05,
+      "loss": 0.4348,
+      "step": 5218
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4307753918517368e-05,
+      "loss": 0.4409,
+      "step": 5219
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.430563958771206e-05,
+      "loss": 0.4064,
+      "step": 5220
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.430352502059797e-05,
+      "loss": 0.4084,
+      "step": 5221
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4301410217291152e-05,
+      "loss": 0.4168,
+      "step": 5222
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4299295177907673e-05,
+      "loss": 0.4197,
+      "step": 5223
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4297179902563618e-05,
+      "loss": 0.4293,
+      "step": 5224
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.429506439137508e-05,
+      "loss": 0.4381,
+      "step": 5225
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4292948644458165e-05,
+      "loss": 0.437,
+      "step": 5226
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4290832661928989e-05,
+      "loss": 0.4383,
+      "step": 5227
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4288716443903687e-05,
+      "loss": 0.4274,
+      "step": 5228
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.428659999049841e-05,
+      "loss": 0.4336,
+      "step": 5229
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4284483301829308e-05,
+      "loss": 0.4372,
+      "step": 5230
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.428236637801256e-05,
+      "loss": 0.4174,
+      "step": 5231
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4280249219164342e-05,
+      "loss": 0.4253,
+      "step": 5232
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.427813182540086e-05,
+      "loss": 0.4322,
+      "step": 5233
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4276014196838319e-05,
+      "loss": 0.4085,
+      "step": 5234
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4273896333592943e-05,
+      "loss": 0.4416,
+      "step": 5235
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4271778235780965e-05,
+      "loss": 0.4281,
+      "step": 5236
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.426965990351864e-05,
+      "loss": 0.4545,
+      "step": 5237
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4267541336922226e-05,
+      "loss": 0.3876,
+      "step": 5238
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4265422536107994e-05,
+      "loss": 0.4323,
+      "step": 5239
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4263303501192237e-05,
+      "loss": 0.4702,
+      "step": 5240
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4261184232291252e-05,
+      "loss": 0.433,
+      "step": 5241
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4259064729521355e-05,
+      "loss": 0.4245,
+      "step": 5242
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4256944992998868e-05,
+      "loss": 0.4202,
+      "step": 5243
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4254825022840133e-05,
+      "loss": 0.4454,
+      "step": 5244
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4252704819161498e-05,
+      "loss": 0.4055,
+      "step": 5245
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.425058438207933e-05,
+      "loss": 0.4172,
+      "step": 5246
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4248463711710006e-05,
+      "loss": 0.4318,
+      "step": 5247
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4246342808169915e-05,
+      "loss": 0.4315,
+      "step": 5248
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.424422167157546e-05,
+      "loss": 0.4103,
+      "step": 5249
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4242100302043056e-05,
+      "loss": 0.4342,
+      "step": 5250
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4239978699689132e-05,
+      "loss": 0.423,
+      "step": 5251
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4237856864630129e-05,
+      "loss": 0.4279,
+      "step": 5252
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4235734796982502e-05,
+      "loss": 0.4168,
+      "step": 5253
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4233612496862715e-05,
+      "loss": 0.4348,
+      "step": 5254
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4231489964387249e-05,
+      "loss": 0.4206,
+      "step": 5255
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4229367199672594e-05,
+      "loss": 0.4103,
+      "step": 5256
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4227244202835258e-05,
+      "loss": 0.437,
+      "step": 5257
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4225120973991758e-05,
+      "loss": 0.4647,
+      "step": 5258
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4222997513258623e-05,
+      "loss": 0.4393,
+      "step": 5259
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4220873820752397e-05,
+      "loss": 0.4199,
+      "step": 5260
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4218749896589637e-05,
+      "loss": 0.4423,
+      "step": 5261
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4216625740886904e-05,
+      "loss": 0.4993,
+      "step": 5262
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.421450135376079e-05,
+      "loss": 0.4021,
+      "step": 5263
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4212376735327883e-05,
+      "loss": 0.4292,
+      "step": 5264
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4210251885704792e-05,
+      "loss": 0.455,
+      "step": 5265
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.420812680500813e-05,
+      "loss": 0.419,
+      "step": 5266
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.420600149335454e-05,
+      "loss": 0.434,
+      "step": 5267
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4203875950860656e-05,
+      "loss": 0.411,
+      "step": 5268
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4201750177643143e-05,
+      "loss": 0.4389,
+      "step": 5269
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4199624173818667e-05,
+      "loss": 0.4209,
+      "step": 5270
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.419749793950391e-05,
+      "loss": 0.4334,
+      "step": 5271
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.419537147481557e-05,
+      "loss": 0.4417,
+      "step": 5272
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4193244779870355e-05,
+      "loss": 0.4201,
+      "step": 5273
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4191117854784981e-05,
+      "loss": 0.42,
+      "step": 5274
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4188990699676186e-05,
+      "loss": 0.4363,
+      "step": 5275
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4186863314660712e-05,
+      "loss": 0.4051,
+      "step": 5276
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4184735699855325e-05,
+      "loss": 0.4364,
+      "step": 5277
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4182607855376787e-05,
+      "loss": 0.4431,
+      "step": 5278
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4180479781341886e-05,
+      "loss": 0.4312,
+      "step": 5279
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4178351477867417e-05,
+      "loss": 0.4093,
+      "step": 5280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4176222945070192e-05,
+      "loss": 0.4407,
+      "step": 5281
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.417409418306703e-05,
+      "loss": 0.3861,
+      "step": 5282
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4171965191974761e-05,
+      "loss": 0.4291,
+      "step": 5283
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4169835971910237e-05,
+      "loss": 0.4627,
+      "step": 5284
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4167706522990318e-05,
+      "loss": 0.4212,
+      "step": 5285
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4165576845331873e-05,
+      "loss": 0.4163,
+      "step": 5286
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4163446939051784e-05,
+      "loss": 0.4031,
+      "step": 5287
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4161316804266952e-05,
+      "loss": 0.4518,
+      "step": 5288
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4159186441094286e-05,
+      "loss": 0.4159,
+      "step": 5289
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4157055849650705e-05,
+      "loss": 0.4231,
+      "step": 5290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4154925030053143e-05,
+      "loss": 0.4444,
+      "step": 5291
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4152793982418554e-05,
+      "loss": 0.4132,
+      "step": 5292
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.415066270686389e-05,
+      "loss": 0.4107,
+      "step": 5293
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4148531203506127e-05,
+      "loss": 0.4348,
+      "step": 5294
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4146399472462247e-05,
+      "loss": 0.4294,
+      "step": 5295
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4144267513849245e-05,
+      "loss": 0.4464,
+      "step": 5296
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.414213532778414e-05,
+      "loss": 0.4365,
+      "step": 5297
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4140002914383944e-05,
+      "loss": 0.4354,
+      "step": 5298
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4137870273765692e-05,
+      "loss": 0.4024,
+      "step": 5299
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4135737406046435e-05,
+      "loss": 0.4488,
+      "step": 5300
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4133604311343234e-05,
+      "loss": 0.4334,
+      "step": 5301
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4131470989773157e-05,
+      "loss": 0.4218,
+      "step": 5302
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4129337441453287e-05,
+      "loss": 0.4101,
+      "step": 5303
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4127203666500725e-05,
+      "loss": 0.387,
+      "step": 5304
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4125069665032575e-05,
+      "loss": 0.4374,
+      "step": 5305
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4122935437165965e-05,
+      "loss": 0.4054,
+      "step": 5306
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4120800983018024e-05,
+      "loss": 0.4199,
+      "step": 5307
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4118666302705899e-05,
+      "loss": 0.3908,
+      "step": 5308
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.411653139634675e-05,
+      "loss": 0.3995,
+      "step": 5309
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.411439626405775e-05,
+      "loss": 0.4316,
+      "step": 5310
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4112260905956079e-05,
+      "loss": 0.4468,
+      "step": 5311
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4110125322158935e-05,
+      "loss": 0.4051,
+      "step": 5312
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4107989512783524e-05,
+      "loss": 0.4461,
+      "step": 5313
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4105853477947071e-05,
+      "loss": 0.4603,
+      "step": 5314
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4103717217766807e-05,
+      "loss": 0.4357,
+      "step": 5315
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4101580732359978e-05,
+      "loss": 0.444,
+      "step": 5316
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.409944402184384e-05,
+      "loss": 0.4444,
+      "step": 5317
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4097307086335665e-05,
+      "loss": 0.4003,
+      "step": 5318
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4095169925952737e-05,
+      "loss": 0.4516,
+      "step": 5319
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4093032540812347e-05,
+      "loss": 0.4098,
+      "step": 5320
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4090894931031805e-05,
+      "loss": 0.4335,
+      "step": 5321
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4088757096728432e-05,
+      "loss": 0.4639,
+      "step": 5322
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4086619038019559e-05,
+      "loss": 0.4308,
+      "step": 5323
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4084480755022527e-05,
+      "loss": 0.4204,
+      "step": 5324
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4082342247854694e-05,
+      "loss": 0.4344,
+      "step": 5325
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.408020351663343e-05,
+      "loss": 0.4466,
+      "step": 5326
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.407806456147612e-05,
+      "loss": 0.4346,
+      "step": 5327
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.407592538250015e-05,
+      "loss": 0.4011,
+      "step": 5328
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4073785979822929e-05,
+      "loss": 0.417,
+      "step": 5329
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4071646353561874e-05,
+      "loss": 0.4424,
+      "step": 5330
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.406950650383442e-05,
+      "loss": 0.4353,
+      "step": 5331
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4067366430758004e-05,
+      "loss": 0.4056,
+      "step": 5332
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4065226134450083e-05,
+      "loss": 0.4138,
+      "step": 5333
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4063085615028122e-05,
+      "loss": 0.3963,
+      "step": 5334
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4060944872609605e-05,
+      "loss": 0.4184,
+      "step": 5335
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4058803907312021e-05,
+      "loss": 0.4334,
+      "step": 5336
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4056662719252872e-05,
+      "loss": 0.4008,
+      "step": 5337
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4054521308549674e-05,
+      "loss": 0.4052,
+      "step": 5338
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.405237967531996e-05,
+      "loss": 0.4395,
+      "step": 5339
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4050237819681265e-05,
+      "loss": 0.4415,
+      "step": 5340
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4048095741751145e-05,
+      "loss": 0.4467,
+      "step": 5341
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.404595344164716e-05,
+      "loss": 0.398,
+      "step": 5342
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4043810919486895e-05,
+      "loss": 0.4255,
+      "step": 5343
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4041668175387935e-05,
+      "loss": 0.425,
+      "step": 5344
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4039525209467878e-05,
+      "loss": 0.441,
+      "step": 5345
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4037382021844344e-05,
+      "loss": 0.4267,
+      "step": 5346
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4035238612634951e-05,
+      "loss": 0.456,
+      "step": 5347
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4033094981957347e-05,
+      "loss": 0.4172,
+      "step": 5348
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4030951129929173e-05,
+      "loss": 0.3914,
+      "step": 5349
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4028807056668096e-05,
+      "loss": 0.4633,
+      "step": 5350
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4026662762291787e-05,
+      "loss": 0.4472,
+      "step": 5351
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4024518246917938e-05,
+      "loss": 0.4625,
+      "step": 5352
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.402237351066424e-05,
+      "loss": 0.4342,
+      "step": 5353
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4020228553648412e-05,
+      "loss": 0.4266,
+      "step": 5354
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.401808337598817e-05,
+      "loss": 0.4329,
+      "step": 5355
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4015937977801255e-05,
+      "loss": 0.4595,
+      "step": 5356
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4013792359205407e-05,
+      "loss": 0.4212,
+      "step": 5357
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4011646520318391e-05,
+      "loss": 0.4106,
+      "step": 5358
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4009500461257977e-05,
+      "loss": 0.4396,
+      "step": 5359
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4007354182141945e-05,
+      "loss": 0.4003,
+      "step": 5360
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4005207683088096e-05,
+      "loss": 0.3787,
+      "step": 5361
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4003060964214232e-05,
+      "loss": 0.4106,
+      "step": 5362
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4000914025638176e-05,
+      "loss": 0.4254,
+      "step": 5363
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.399876686747776e-05,
+      "loss": 0.4289,
+      "step": 5364
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3996619489850822e-05,
+      "loss": 0.4373,
+      "step": 5365
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3994471892875228e-05,
+      "loss": 0.4431,
+      "step": 5366
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3992324076668837e-05,
+      "loss": 0.4331,
+      "step": 5367
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.399017604134953e-05,
+      "loss": 0.4763,
+      "step": 5368
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.39880277870352e-05,
+      "loss": 0.44,
+      "step": 5369
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3985879313843752e-05,
+      "loss": 0.4583,
+      "step": 5370
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3983730621893106e-05,
+      "loss": 0.4166,
+      "step": 5371
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3981581711301177e-05,
+      "loss": 0.4286,
+      "step": 5372
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3979432582185917e-05,
+      "loss": 0.4039,
+      "step": 5373
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3977283234665273e-05,
+      "loss": 0.4094,
+      "step": 5374
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.397513366885721e-05,
+      "loss": 0.4269,
+      "step": 5375
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.39729838848797e-05,
+      "loss": 0.4237,
+      "step": 5376
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3970833882850734e-05,
+      "loss": 0.4423,
+      "step": 5377
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3968683662888316e-05,
+      "loss": 0.4492,
+      "step": 5378
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3966533225110452e-05,
+      "loss": 0.3866,
+      "step": 5379
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3964382569635165e-05,
+      "loss": 0.4342,
+      "step": 5380
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.396223169658049e-05,
+      "loss": 0.4448,
+      "step": 5381
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3960080606064483e-05,
+      "loss": 0.3912,
+      "step": 5382
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3957929298205195e-05,
+      "loss": 0.413,
+      "step": 5383
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3955777773120703e-05,
+      "loss": 0.4326,
+      "step": 5384
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.395362603092908e-05,
+      "loss": 0.4426,
+      "step": 5385
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3951474071748435e-05,
+      "loss": 0.4393,
+      "step": 5386
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3949321895696868e-05,
+      "loss": 0.4272,
+      "step": 5387
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.39471695028925e-05,
+      "loss": 0.4439,
+      "step": 5388
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3945016893453453e-05,
+      "loss": 0.382,
+      "step": 5389
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3942864067497884e-05,
+      "loss": 0.4202,
+      "step": 5390
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3940711025143943e-05,
+      "loss": 0.4447,
+      "step": 5391
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3938557766509792e-05,
+      "loss": 0.4061,
+      "step": 5392
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3936404291713612e-05,
+      "loss": 0.3999,
+      "step": 5393
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3934250600873593e-05,
+      "loss": 0.427,
+      "step": 5394
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3932096694107941e-05,
+      "loss": 0.4464,
+      "step": 5395
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3929942571534866e-05,
+      "loss": 0.3959,
+      "step": 5396
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3927788233272594e-05,
+      "loss": 0.4582,
+      "step": 5397
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3925633679439364e-05,
+      "loss": 0.4358,
+      "step": 5398
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3923478910153425e-05,
+      "loss": 0.4249,
+      "step": 5399
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3921323925533042e-05,
+      "loss": 0.4073,
+      "step": 5400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3919168725696478e-05,
+      "loss": 0.4379,
+      "step": 5401
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3917013310762029e-05,
+      "loss": 0.4237,
+      "step": 5402
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3914857680847989e-05,
+      "loss": 0.4213,
+      "step": 5403
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3912701836072664e-05,
+      "loss": 0.4289,
+      "step": 5404
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3910545776554376e-05,
+      "loss": 0.422,
+      "step": 5405
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3908389502411452e-05,
+      "loss": 0.4222,
+      "step": 5406
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3906233013762247e-05,
+      "loss": 0.416,
+      "step": 5407
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.390407631072511e-05,
+      "loss": 0.4323,
+      "step": 5408
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3901919393418407e-05,
+      "loss": 0.4447,
+      "step": 5409
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3899762261960519e-05,
+      "loss": 0.4071,
+      "step": 5410
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3897604916469838e-05,
+      "loss": 0.4466,
+      "step": 5411
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3895447357064766e-05,
+      "loss": 0.418,
+      "step": 5412
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.389328958386372e-05,
+      "loss": 0.4175,
+      "step": 5413
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3891131596985121e-05,
+      "loss": 0.3938,
+      "step": 5414
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3888973396547409e-05,
+      "loss": 0.434,
+      "step": 5415
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3886814982669038e-05,
+      "loss": 0.4005,
+      "step": 5416
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3884656355468466e-05,
+      "loss": 0.4518,
+      "step": 5417
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3882497515064164e-05,
+      "loss": 0.3932,
+      "step": 5418
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.388033846157462e-05,
+      "loss": 0.4189,
+      "step": 5419
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.387817919511833e-05,
+      "loss": 0.3989,
+      "step": 5420
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3876019715813802e-05,
+      "loss": 0.3859,
+      "step": 5421
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3873860023779556e-05,
+      "loss": 0.4597,
+      "step": 5422
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.387170011913412e-05,
+      "loss": 0.4647,
+      "step": 5423
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3869540001996043e-05,
+      "loss": 0.4373,
+      "step": 5424
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3867379672483878e-05,
+      "loss": 0.4264,
+      "step": 5425
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3865219130716187e-05,
+      "loss": 0.4272,
+      "step": 5426
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3863058376811554e-05,
+      "loss": 0.4494,
+      "step": 5427
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.386089741088857e-05,
+      "loss": 0.4578,
+      "step": 5428
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.385873623306583e-05,
+      "loss": 0.4156,
+      "step": 5429
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3856574843461952e-05,
+      "loss": 0.403,
+      "step": 5430
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3854413242195559e-05,
+      "loss": 0.4654,
+      "step": 5431
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3852251429385286e-05,
+      "loss": 0.3893,
+      "step": 5432
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3850089405149785e-05,
+      "loss": 0.4492,
+      "step": 5433
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3847927169607711e-05,
+      "loss": 0.4454,
+      "step": 5434
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.384576472287774e-05,
+      "loss": 0.4406,
+      "step": 5435
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3843602065078547e-05,
+      "loss": 0.4037,
+      "step": 5436
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3841439196328836e-05,
+      "loss": 0.4248,
+      "step": 5437
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3839276116747306e-05,
+      "loss": 0.4228,
+      "step": 5438
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3837112826452679e-05,
+      "loss": 0.4514,
+      "step": 5439
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.383494932556368e-05,
+      "loss": 0.4085,
+      "step": 5440
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3832785614199052e-05,
+      "loss": 0.4371,
+      "step": 5441
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3830621692477544e-05,
+      "loss": 0.4232,
+      "step": 5442
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3828457560517924e-05,
+      "loss": 0.4312,
+      "step": 5443
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3826293218438966e-05,
+      "loss": 0.4317,
+      "step": 5444
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3824128666359458e-05,
+      "loss": 0.4273,
+      "step": 5445
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3821963904398194e-05,
+      "loss": 0.4026,
+      "step": 5446
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3819798932673986e-05,
+      "loss": 0.4289,
+      "step": 5447
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3817633751305658e-05,
+      "loss": 0.4002,
+      "step": 5448
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.381546836041204e-05,
+      "loss": 0.4255,
+      "step": 5449
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3813302760111975e-05,
+      "loss": 0.4505,
+      "step": 5450
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3811136950524325e-05,
+      "loss": 0.4213,
+      "step": 5451
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3808970931767949e-05,
+      "loss": 0.4361,
+      "step": 5452
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3806804703961731e-05,
+      "loss": 0.4597,
+      "step": 5453
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3804638267224559e-05,
+      "loss": 0.4376,
+      "step": 5454
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3802471621675337e-05,
+      "loss": 0.4154,
+      "step": 5455
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3800304767432979e-05,
+      "loss": 0.4114,
+      "step": 5456
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3798137704616404e-05,
+      "loss": 0.4353,
+      "step": 5457
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3795970433344552e-05,
+      "loss": 0.4218,
+      "step": 5458
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3793802953736375e-05,
+      "loss": 0.4064,
+      "step": 5459
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3791635265910823e-05,
+      "loss": 0.4149,
+      "step": 5460
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3789467369986872e-05,
+      "loss": 0.4093,
+      "step": 5461
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3787299266083503e-05,
+      "loss": 0.4092,
+      "step": 5462
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3785130954319707e-05,
+      "loss": 0.4183,
+      "step": 5463
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3782962434814493e-05,
+      "loss": 0.3887,
+      "step": 5464
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3780793707686872e-05,
+      "loss": 0.4812,
+      "step": 5465
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3778624773055876e-05,
+      "loss": 0.4296,
+      "step": 5466
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.377645563104054e-05,
+      "loss": 0.4464,
+      "step": 5467
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3774286281759917e-05,
+      "loss": 0.4289,
+      "step": 5468
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.377211672533307e-05,
+      "loss": 0.4442,
+      "step": 5469
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3769946961879064e-05,
+      "loss": 0.4314,
+      "step": 5470
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.376777699151699e-05,
+      "loss": 0.4401,
+      "step": 5471
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3765606814365947e-05,
+      "loss": 0.3999,
+      "step": 5472
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3763436430545034e-05,
+      "loss": 0.4539,
+      "step": 5473
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3761265840173372e-05,
+      "loss": 0.4239,
+      "step": 5474
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3759095043370092e-05,
+      "loss": 0.412,
+      "step": 5475
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3756924040254336e-05,
+      "loss": 0.4453,
+      "step": 5476
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3754752830945255e-05,
+      "loss": 0.4375,
+      "step": 5477
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3752581415562012e-05,
+      "loss": 0.4284,
+      "step": 5478
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3750409794223783e-05,
+      "loss": 0.4325,
+      "step": 5479
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3748237967049753e-05,
+      "loss": 0.4468,
+      "step": 5480
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3746065934159123e-05,
+      "loss": 0.4145,
+      "step": 5481
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3743893695671095e-05,
+      "loss": 0.4379,
+      "step": 5482
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3741721251704895e-05,
+      "loss": 0.4254,
+      "step": 5483
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3739548602379755e-05,
+      "loss": 0.4502,
+      "step": 5484
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3737375747814916e-05,
+      "loss": 0.4499,
+      "step": 5485
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3735202688129631e-05,
+      "loss": 0.4407,
+      "step": 5486
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3733029423443163e-05,
+      "loss": 0.3873,
+      "step": 5487
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3730855953874796e-05,
+      "loss": 0.4049,
+      "step": 5488
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.372868227954381e-05,
+      "loss": 0.4231,
+      "step": 5489
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3726508400569509e-05,
+      "loss": 0.4374,
+      "step": 5490
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3724334317071198e-05,
+      "loss": 0.4122,
+      "step": 5491
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3722160029168204e-05,
+      "loss": 0.4012,
+      "step": 5492
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3719985536979859e-05,
+      "loss": 0.4616,
+      "step": 5493
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3717810840625502e-05,
+      "loss": 0.4285,
+      "step": 5494
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3715635940224492e-05,
+      "loss": 0.447,
+      "step": 5495
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3713460835896193e-05,
+      "loss": 0.4466,
+      "step": 5496
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3711285527759988e-05,
+      "loss": 0.3863,
+      "step": 5497
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.370911001593526e-05,
+      "loss": 0.4454,
+      "step": 5498
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3706934300541409e-05,
+      "loss": 0.4014,
+      "step": 5499
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3704758381697845e-05,
+      "loss": 0.4497,
+      "step": 5500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3702582259523996e-05,
+      "loss": 0.4434,
+      "step": 5501
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3700405934139292e-05,
+      "loss": 0.4154,
+      "step": 5502
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3698229405663175e-05,
+      "loss": 0.4273,
+      "step": 5503
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3696052674215103e-05,
+      "loss": 0.4054,
+      "step": 5504
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3693875739914545e-05,
+      "loss": 0.4387,
+      "step": 5505
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3691698602880976e-05,
+      "loss": 0.407,
+      "step": 5506
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3689521263233883e-05,
+      "loss": 0.4903,
+      "step": 5507
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3687343721092768e-05,
+      "loss": 0.4304,
+      "step": 5508
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3685165976577146e-05,
+      "loss": 0.4243,
+      "step": 5509
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3682988029806537e-05,
+      "loss": 0.4708,
+      "step": 5510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.368080988090047e-05,
+      "loss": 0.4254,
+      "step": 5511
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3678631529978494e-05,
+      "loss": 0.4041,
+      "step": 5512
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3676452977160163e-05,
+      "loss": 0.4358,
+      "step": 5513
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.367427422256505e-05,
+      "loss": 0.4544,
+      "step": 5514
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3672095266312722e-05,
+      "loss": 0.4462,
+      "step": 5515
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3669916108522774e-05,
+      "loss": 0.4254,
+      "step": 5516
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3667736749314806e-05,
+      "loss": 0.4074,
+      "step": 5517
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3665557188808429e-05,
+      "loss": 0.4618,
+      "step": 5518
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3663377427123261e-05,
+      "loss": 0.4089,
+      "step": 5519
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3661197464378941e-05,
+      "loss": 0.413,
+      "step": 5520
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3659017300695108e-05,
+      "loss": 0.417,
+      "step": 5521
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.365683693619142e-05,
+      "loss": 0.438,
+      "step": 5522
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3654656370987542e-05,
+      "loss": 0.4221,
+      "step": 5523
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3652475605203152e-05,
+      "loss": 0.4039,
+      "step": 5524
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3650294638957939e-05,
+      "loss": 0.4233,
+      "step": 5525
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.36481134723716e-05,
+      "loss": 0.4306,
+      "step": 5526
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3645932105563846e-05,
+      "loss": 0.4203,
+      "step": 5527
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.36437505386544e-05,
+      "loss": 0.4289,
+      "step": 5528
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.364156877176299e-05,
+      "loss": 0.4079,
+      "step": 5529
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3639386805009363e-05,
+      "loss": 0.4472,
+      "step": 5530
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3637204638513273e-05,
+      "loss": 0.4073,
+      "step": 5531
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.363502227239448e-05,
+      "loss": 0.402,
+      "step": 5532
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3632839706772764e-05,
+      "loss": 0.4121,
+      "step": 5533
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3630656941767915e-05,
+      "loss": 0.413,
+      "step": 5534
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3628473977499726e-05,
+      "loss": 0.3976,
+      "step": 5535
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3626290814088005e-05,
+      "loss": 0.4241,
+      "step": 5536
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3624107451652577e-05,
+      "loss": 0.4214,
+      "step": 5537
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3621923890313267e-05,
+      "loss": 0.4403,
+      "step": 5538
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3619740130189922e-05,
+      "loss": 0.4244,
+      "step": 5539
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.361755617140239e-05,
+      "loss": 0.4245,
+      "step": 5540
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3615372014070539e-05,
+      "loss": 0.4408,
+      "step": 5541
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3613187658314235e-05,
+      "loss": 0.4184,
+      "step": 5542
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3611003104253373e-05,
+      "loss": 0.402,
+      "step": 5543
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3608818352007845e-05,
+      "loss": 0.4345,
+      "step": 5544
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3606633401697557e-05,
+      "loss": 0.4436,
+      "step": 5545
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3604448253442427e-05,
+      "loss": 0.3981,
+      "step": 5546
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3602262907362384e-05,
+      "loss": 0.4424,
+      "step": 5547
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3600077363577369e-05,
+      "loss": 0.4324,
+      "step": 5548
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3597891622207331e-05,
+      "loss": 0.4152,
+      "step": 5549
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3595705683372229e-05,
+      "loss": 0.4049,
+      "step": 5550
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3593519547192042e-05,
+      "loss": 0.4374,
+      "step": 5551
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3591333213786744e-05,
+      "loss": 0.4015,
+      "step": 5552
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3589146683276338e-05,
+      "loss": 0.4236,
+      "step": 5553
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3586959955780822e-05,
+      "loss": 0.4092,
+      "step": 5554
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3584773031420211e-05,
+      "loss": 0.4421,
+      "step": 5555
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3582585910314535e-05,
+      "loss": 0.4315,
+      "step": 5556
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.358039859258383e-05,
+      "loss": 0.4529,
+      "step": 5557
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3578211078348142e-05,
+      "loss": 0.4144,
+      "step": 5558
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.357602336772753e-05,
+      "loss": 0.4124,
+      "step": 5559
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3573835460842064e-05,
+      "loss": 0.435,
+      "step": 5560
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3571647357811826e-05,
+      "loss": 0.4084,
+      "step": 5561
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3569459058756904e-05,
+      "loss": 0.3691,
+      "step": 5562
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3567270563797399e-05,
+      "loss": 0.4495,
+      "step": 5563
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3565081873053425e-05,
+      "loss": 0.4197,
+      "step": 5564
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3562892986645107e-05,
+      "loss": 0.4208,
+      "step": 5565
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3560703904692576e-05,
+      "loss": 0.4076,
+      "step": 5566
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3558514627315974e-05,
+      "loss": 0.4346,
+      "step": 5567
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3556325154635464e-05,
+      "loss": 0.4094,
+      "step": 5568
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3554135486771207e-05,
+      "loss": 0.4006,
+      "step": 5569
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.355194562384338e-05,
+      "loss": 0.4118,
+      "step": 5570
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3549755565972171e-05,
+      "loss": 0.4297,
+      "step": 5571
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3547565313277778e-05,
+      "loss": 0.4613,
+      "step": 5572
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3545374865880409e-05,
+      "loss": 0.4123,
+      "step": 5573
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3543184223900287e-05,
+      "loss": 0.4699,
+      "step": 5574
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.354099338745764e-05,
+      "loss": 0.4544,
+      "step": 5575
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3538802356672705e-05,
+      "loss": 0.4291,
+      "step": 5576
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3536611131665737e-05,
+      "loss": 0.4056,
+      "step": 5577
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3534419712557003e-05,
+      "loss": 0.4003,
+      "step": 5578
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.353222809946677e-05,
+      "loss": 0.4433,
+      "step": 5579
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3530036292515318e-05,
+      "loss": 0.4168,
+      "step": 5580
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3527844291822949e-05,
+      "loss": 0.4464,
+      "step": 5581
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3525652097509964e-05,
+      "loss": 0.4191,
+      "step": 5582
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3523459709696682e-05,
+      "loss": 0.408,
+      "step": 5583
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3521267128503421e-05,
+      "loss": 0.4088,
+      "step": 5584
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3519074354050525e-05,
+      "loss": 0.3914,
+      "step": 5585
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.351688138645834e-05,
+      "loss": 0.4197,
+      "step": 5586
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3514688225847224e-05,
+      "loss": 0.3973,
+      "step": 5587
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.351249487233754e-05,
+      "loss": 0.4198,
+      "step": 5588
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3510301326049673e-05,
+      "loss": 0.4116,
+      "step": 5589
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.350810758710401e-05,
+      "loss": 0.4532,
+      "step": 5590
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3505913655620955e-05,
+      "loss": 0.4079,
+      "step": 5591
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3503719531720912e-05,
+      "loss": 0.4262,
+      "step": 5592
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3501525215524307e-05,
+      "loss": 0.4424,
+      "step": 5593
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3499330707151572e-05,
+      "loss": 0.4154,
+      "step": 5594
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3497136006723151e-05,
+      "loss": 0.4275,
+      "step": 5595
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.349494111435949e-05,
+      "loss": 0.4307,
+      "step": 5596
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3492746030181056e-05,
+      "loss": 0.4618,
+      "step": 5597
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3490550754308328e-05,
+      "loss": 0.4564,
+      "step": 5598
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3488355286861783e-05,
+      "loss": 0.4126,
+      "step": 5599
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3486159627961922e-05,
+      "loss": 0.3968,
+      "step": 5600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3483963777729246e-05,
+      "loss": 0.4127,
+      "step": 5601
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.348176773628427e-05,
+      "loss": 0.4503,
+      "step": 5602
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3479571503747529e-05,
+      "loss": 0.4157,
+      "step": 5603
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3477375080239554e-05,
+      "loss": 0.4009,
+      "step": 5604
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.347517846588089e-05,
+      "loss": 0.4196,
+      "step": 5605
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3472981660792097e-05,
+      "loss": 0.409,
+      "step": 5606
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3470784665093748e-05,
+      "loss": 0.4061,
+      "step": 5607
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3468587478906419e-05,
+      "loss": 0.3907,
+      "step": 5608
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3466390102350696e-05,
+      "loss": 0.4383,
+      "step": 5609
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3464192535547181e-05,
+      "loss": 0.4443,
+      "step": 5610
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3461994778616488e-05,
+      "loss": 0.4201,
+      "step": 5611
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3459796831679234e-05,
+      "loss": 0.423,
+      "step": 5612
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3457598694856047e-05,
+      "loss": 0.3805,
+      "step": 5613
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3455400368267576e-05,
+      "loss": 0.4142,
+      "step": 5614
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3453201852034468e-05,
+      "loss": 0.4266,
+      "step": 5615
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3451003146277388e-05,
+      "loss": 0.4116,
+      "step": 5616
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3448804251117005e-05,
+      "loss": 0.4193,
+      "step": 5617
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3446605166674007e-05,
+      "loss": 0.4239,
+      "step": 5618
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3444405893069083e-05,
+      "loss": 0.3936,
+      "step": 5619
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3442206430422941e-05,
+      "loss": 0.3872,
+      "step": 5620
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3440006778856292e-05,
+      "loss": 0.4316,
+      "step": 5621
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3437806938489861e-05,
+      "loss": 0.427,
+      "step": 5622
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3435606909444385e-05,
+      "loss": 0.4372,
+      "step": 5623
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3433406691840611e-05,
+      "loss": 0.4211,
+      "step": 5624
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.343120628579929e-05,
+      "loss": 0.3965,
+      "step": 5625
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.342900569144119e-05,
+      "loss": 0.4176,
+      "step": 5626
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3426804908887087e-05,
+      "loss": 0.4614,
+      "step": 5627
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3424603938257772e-05,
+      "loss": 0.4158,
+      "step": 5628
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3422402779674035e-05,
+      "loss": 0.445,
+      "step": 5629
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.43,
+      "step": 5630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3417999899126547e-05,
+      "loss": 0.4161,
+      "step": 5631
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3415798177404445e-05,
+      "loss": 0.4293,
+      "step": 5632
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3413596268211211e-05,
+      "loss": 0.4333,
+      "step": 5633
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3411394171667702e-05,
+      "loss": 0.4285,
+      "step": 5634
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.340919188789477e-05,
+      "loss": 0.4284,
+      "step": 5635
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3406989417013292e-05,
+      "loss": 0.38,
+      "step": 5636
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3404786759144139e-05,
+      "loss": 0.393,
+      "step": 5637
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3402583914408207e-05,
+      "loss": 0.4649,
+      "step": 5638
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3400380882926394e-05,
+      "loss": 0.4372,
+      "step": 5639
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3398177664819608e-05,
+      "loss": 0.4403,
+      "step": 5640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3395974260208771e-05,
+      "loss": 0.396,
+      "step": 5641
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3393770669214816e-05,
+      "loss": 0.4178,
+      "step": 5642
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3391566891958681e-05,
+      "loss": 0.4331,
+      "step": 5643
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3389362928561319e-05,
+      "loss": 0.4158,
+      "step": 5644
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3387158779143688e-05,
+      "loss": 0.4253,
+      "step": 5645
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3384954443826762e-05,
+      "loss": 0.4101,
+      "step": 5646
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3382749922731527e-05,
+      "loss": 0.4261,
+      "step": 5647
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3380545215978965e-05,
+      "loss": 0.442,
+      "step": 5648
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3378340323690085e-05,
+      "loss": 0.4051,
+      "step": 5649
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.33761352459859e-05,
+      "loss": 0.4267,
+      "step": 5650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3373929982987427e-05,
+      "loss": 0.4125,
+      "step": 5651
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3371724534815705e-05,
+      "loss": 0.4607,
+      "step": 5652
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3369518901591771e-05,
+      "loss": 0.4295,
+      "step": 5653
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3367313083436686e-05,
+      "loss": 0.4014,
+      "step": 5654
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3365107080471507e-05,
+      "loss": 0.4187,
+      "step": 5655
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3362900892817306e-05,
+      "loss": 0.4106,
+      "step": 5656
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3360694520595168e-05,
+      "loss": 0.4138,
+      "step": 5657
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3358487963926188e-05,
+      "loss": 0.4307,
+      "step": 5658
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3356281222931471e-05,
+      "loss": 0.4362,
+      "step": 5659
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3354074297732127e-05,
+      "loss": 0.4275,
+      "step": 5660
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3351867188449279e-05,
+      "loss": 0.3776,
+      "step": 5661
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3349659895204068e-05,
+      "loss": 0.3992,
+      "step": 5662
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3347452418117632e-05,
+      "loss": 0.4232,
+      "step": 5663
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3345244757311129e-05,
+      "loss": 0.4125,
+      "step": 5664
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3343036912905718e-05,
+      "loss": 0.4121,
+      "step": 5665
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3340828885022576e-05,
+      "loss": 0.4314,
+      "step": 5666
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3338620673782891e-05,
+      "loss": 0.4361,
+      "step": 5667
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3336412279307853e-05,
+      "loss": 0.4672,
+      "step": 5668
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3334203701718667e-05,
+      "loss": 0.4509,
+      "step": 5669
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3331994941136548e-05,
+      "loss": 0.4137,
+      "step": 5670
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3329785997682722e-05,
+      "loss": 0.4498,
+      "step": 5671
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3327576871478425e-05,
+      "loss": 0.4258,
+      "step": 5672
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3325367562644896e-05,
+      "loss": 0.4276,
+      "step": 5673
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3323158071303393e-05,
+      "loss": 0.4158,
+      "step": 5674
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3320948397575181e-05,
+      "loss": 0.4405,
+      "step": 5675
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3318738541581537e-05,
+      "loss": 0.4223,
+      "step": 5676
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3316528503443745e-05,
+      "loss": 0.4178,
+      "step": 5677
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.331431828328309e-05,
+      "loss": 0.44,
+      "step": 5678
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3312107881220891e-05,
+      "loss": 0.4231,
+      "step": 5679
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3309897297378456e-05,
+      "loss": 0.4475,
+      "step": 5680
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.330768653187711e-05,
+      "loss": 0.4046,
+      "step": 5681
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3305475584838186e-05,
+      "loss": 0.4343,
+      "step": 5682
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.330326445638303e-05,
+      "loss": 0.4372,
+      "step": 5683
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3301053146633001e-05,
+      "loss": 0.4135,
+      "step": 5684
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3298841655709457e-05,
+      "loss": 0.401,
+      "step": 5685
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3296629983733775e-05,
+      "loss": 0.4232,
+      "step": 5686
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3294418130827337e-05,
+      "loss": 0.42,
+      "step": 5687
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3292206097111544e-05,
+      "loss": 0.4208,
+      "step": 5688
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3289993882707797e-05,
+      "loss": 0.4204,
+      "step": 5689
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3287781487737503e-05,
+      "loss": 0.4285,
+      "step": 5690
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3285568912322096e-05,
+      "loss": 0.4244,
+      "step": 5691
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3283356156583007e-05,
+      "loss": 0.4489,
+      "step": 5692
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.328114322064168e-05,
+      "loss": 0.405,
+      "step": 5693
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3278930104619563e-05,
+      "loss": 0.4006,
+      "step": 5694
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3276716808638128e-05,
+      "loss": 0.465,
+      "step": 5695
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3274503332818847e-05,
+      "loss": 0.442,
+      "step": 5696
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.32722896772832e-05,
+      "loss": 0.3801,
+      "step": 5697
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3270075842152679e-05,
+      "loss": 0.434,
+      "step": 5698
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3267861827548791e-05,
+      "loss": 0.4433,
+      "step": 5699
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3265647633593051e-05,
+      "loss": 0.4333,
+      "step": 5700
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3263433260406977e-05,
+      "loss": 0.4327,
+      "step": 5701
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3261218708112104e-05,
+      "loss": 0.4404,
+      "step": 5702
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3259003976829973e-05,
+      "loss": 0.4222,
+      "step": 5703
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3256789066682137e-05,
+      "loss": 0.4218,
+      "step": 5704
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.325457397779016e-05,
+      "loss": 0.4146,
+      "step": 5705
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3252358710275611e-05,
+      "loss": 0.4481,
+      "step": 5706
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3250143264260073e-05,
+      "loss": 0.4146,
+      "step": 5707
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3247927639865137e-05,
+      "loss": 0.4095,
+      "step": 5708
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3245711837212408e-05,
+      "loss": 0.4062,
+      "step": 5709
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3243495856423489e-05,
+      "loss": 0.4098,
+      "step": 5710
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3241279697620009e-05,
+      "loss": 0.4492,
+      "step": 5711
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3239063360923594e-05,
+      "loss": 0.4561,
+      "step": 5712
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3236846846455888e-05,
+      "loss": 0.4251,
+      "step": 5713
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.323463015433854e-05,
+      "loss": 0.4347,
+      "step": 5714
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3232413284693206e-05,
+      "loss": 0.4274,
+      "step": 5715
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.323019623764156e-05,
+      "loss": 0.4335,
+      "step": 5716
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.322797901330528e-05,
+      "loss": 0.3863,
+      "step": 5717
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3225761611806056e-05,
+      "loss": 0.3896,
+      "step": 5718
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3223544033265588e-05,
+      "loss": 0.4133,
+      "step": 5719
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3221326277805578e-05,
+      "loss": 0.4489,
+      "step": 5720
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3219108345547754e-05,
+      "loss": 0.4277,
+      "step": 5721
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3216890236613836e-05,
+      "loss": 0.4149,
+      "step": 5722
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3214671951125566e-05,
+      "loss": 0.4167,
+      "step": 5723
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3212453489204691e-05,
+      "loss": 0.4259,
+      "step": 5724
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3210234850972966e-05,
+      "loss": 0.4371,
+      "step": 5725
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3208016036552155e-05,
+      "loss": 0.4077,
+      "step": 5726
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3205797046064044e-05,
+      "loss": 0.4321,
+      "step": 5727
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3203577879630411e-05,
+      "loss": 0.3934,
+      "step": 5728
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3201358537373054e-05,
+      "loss": 0.4229,
+      "step": 5729
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.319913901941378e-05,
+      "loss": 0.4162,
+      "step": 5730
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3196919325874403e-05,
+      "loss": 0.4019,
+      "step": 5731
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3194699456876747e-05,
+      "loss": 0.4226,
+      "step": 5732
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3192479412542642e-05,
+      "loss": 0.4247,
+      "step": 5733
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3190259192993939e-05,
+      "loss": 0.4395,
+      "step": 5734
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3188038798352492e-05,
+      "loss": 0.4179,
+      "step": 5735
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.318581822874016e-05,
+      "loss": 0.4401,
+      "step": 5736
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3183597484278816e-05,
+      "loss": 0.4101,
+      "step": 5737
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3181376565090343e-05,
+      "loss": 0.4066,
+      "step": 5738
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3179155471296633e-05,
+      "loss": 0.4029,
+      "step": 5739
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3176934203019586e-05,
+      "loss": 0.464,
+      "step": 5740
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3174712760381117e-05,
+      "loss": 0.4256,
+      "step": 5741
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3172491143503146e-05,
+      "loss": 0.4241,
+      "step": 5742
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3170269352507596e-05,
+      "loss": 0.3968,
+      "step": 5743
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3168047387516417e-05,
+      "loss": 0.4106,
+      "step": 5744
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3165825248651553e-05,
+      "loss": 0.3983,
+      "step": 5745
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3163602936034962e-05,
+      "loss": 0.4222,
+      "step": 5746
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3161380449788613e-05,
+      "loss": 0.4387,
+      "step": 5747
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3159157790034488e-05,
+      "loss": 0.4282,
+      "step": 5748
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3156934956894571e-05,
+      "loss": 0.4088,
+      "step": 5749
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3154711950490857e-05,
+      "loss": 0.4037,
+      "step": 5750
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3152488770945354e-05,
+      "loss": 0.373,
+      "step": 5751
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3150265418380082e-05,
+      "loss": 0.4108,
+      "step": 5752
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3148041892917063e-05,
+      "loss": 0.4075,
+      "step": 5753
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3145818194678329e-05,
+      "loss": 0.4097,
+      "step": 5754
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.314359432378593e-05,
+      "loss": 0.404,
+      "step": 5755
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3141370280361918e-05,
+      "loss": 0.4224,
+      "step": 5756
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3139146064528356e-05,
+      "loss": 0.3798,
+      "step": 5757
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3136921676407315e-05,
+      "loss": 0.4354,
+      "step": 5758
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3134697116120878e-05,
+      "loss": 0.4249,
+      "step": 5759
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3132472383791142e-05,
+      "loss": 0.4023,
+      "step": 5760
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3130247479540204e-05,
+      "loss": 0.3995,
+      "step": 5761
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3128022403490172e-05,
+      "loss": 0.403,
+      "step": 5762
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3125797155763166e-05,
+      "loss": 0.4324,
+      "step": 5763
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3123571736481324e-05,
+      "loss": 0.3854,
+      "step": 5764
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3121346145766781e-05,
+      "loss": 0.3659,
+      "step": 5765
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.311912038374168e-05,
+      "loss": 0.4333,
+      "step": 5766
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3116894450528182e-05,
+      "loss": 0.43,
+      "step": 5767
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3114668346248452e-05,
+      "loss": 0.3992,
+      "step": 5768
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3112442071024671e-05,
+      "loss": 0.4286,
+      "step": 5769
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3110215624979026e-05,
+      "loss": 0.4256,
+      "step": 5770
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3107989008233703e-05,
+      "loss": 0.4236,
+      "step": 5771
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3105762220910916e-05,
+      "loss": 0.4243,
+      "step": 5772
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3103535263132879e-05,
+      "loss": 0.4002,
+      "step": 5773
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3101308135021808e-05,
+      "loss": 0.3973,
+      "step": 5774
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3099080836699938e-05,
+      "loss": 0.4276,
+      "step": 5775
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3096853368289515e-05,
+      "loss": 0.3932,
+      "step": 5776
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3094625729912789e-05,
+      "loss": 0.4033,
+      "step": 5777
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3092397921692019e-05,
+      "loss": 0.4418,
+      "step": 5778
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.4247,
+      "step": 5779
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3087941796207438e-05,
+      "loss": 0.4155,
+      "step": 5780
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3085713479188195e-05,
+      "loss": 0.4444,
+      "step": 5781
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3083484992814049e-05,
+      "loss": 0.3877,
+      "step": 5782
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3081256337207297e-05,
+      "loss": 0.4095,
+      "step": 5783
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3079027512490263e-05,
+      "loss": 0.4092,
+      "step": 5784
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3076798518785274e-05,
+      "loss": 0.3865,
+      "step": 5785
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3074569356214661e-05,
+      "loss": 0.4272,
+      "step": 5786
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.307234002490077e-05,
+      "loss": 0.4123,
+      "step": 5787
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3070110524965953e-05,
+      "loss": 0.396,
+      "step": 5788
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3067880856532575e-05,
+      "loss": 0.4116,
+      "step": 5789
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3065651019723008e-05,
+      "loss": 0.4395,
+      "step": 5790
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3063421014659632e-05,
+      "loss": 0.4281,
+      "step": 5791
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.306119084146484e-05,
+      "loss": 0.4378,
+      "step": 5792
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3058960500261027e-05,
+      "loss": 0.4298,
+      "step": 5793
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3056729991170607e-05,
+      "loss": 0.425,
+      "step": 5794
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3054499314315996e-05,
+      "loss": 0.4401,
+      "step": 5795
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3052268469819624e-05,
+      "loss": 0.3986,
+      "step": 5796
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3050037457803923e-05,
+      "loss": 0.4335,
+      "step": 5797
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3047806278391341e-05,
+      "loss": 0.3795,
+      "step": 5798
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3045574931704334e-05,
+      "loss": 0.4123,
+      "step": 5799
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3043343417865368e-05,
+      "loss": 0.4185,
+      "step": 5800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3041111736996913e-05,
+      "loss": 0.3996,
+      "step": 5801
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3038879889221452e-05,
+      "loss": 0.4325,
+      "step": 5802
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3036647874661478e-05,
+      "loss": 0.4481,
+      "step": 5803
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3034415693439493e-05,
+      "loss": 0.4213,
+      "step": 5804
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3032183345678e-05,
+      "loss": 0.4101,
+      "step": 5805
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3029950831499529e-05,
+      "loss": 0.3956,
+      "step": 5806
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3027718151026601e-05,
+      "loss": 0.4281,
+      "step": 5807
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3025485304381754e-05,
+      "loss": 0.4188,
+      "step": 5808
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3023252291687535e-05,
+      "loss": 0.4331,
+      "step": 5809
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3021019113066503e-05,
+      "loss": 0.4462,
+      "step": 5810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.301878576864122e-05,
+      "loss": 0.4515,
+      "step": 5811
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3016552258534256e-05,
+      "loss": 0.4201,
+      "step": 5812
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3014318582868204e-05,
+      "loss": 0.4258,
+      "step": 5813
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3012084741765644e-05,
+      "loss": 0.4298,
+      "step": 5814
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.300985073534919e-05,
+      "loss": 0.381,
+      "step": 5815
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3007616563741439e-05,
+      "loss": 0.45,
+      "step": 5816
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3005382227065022e-05,
+      "loss": 0.4354,
+      "step": 5817
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3003147725442558e-05,
+      "loss": 0.4271,
+      "step": 5818
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3000913058996688e-05,
+      "loss": 0.413,
+      "step": 5819
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2998678227850062e-05,
+      "loss": 0.4414,
+      "step": 5820
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2996443232125331e-05,
+      "loss": 0.3955,
+      "step": 5821
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2994208071945158e-05,
+      "loss": 0.4309,
+      "step": 5822
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2991972747432224e-05,
+      "loss": 0.4461,
+      "step": 5823
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2989737258709203e-05,
+      "loss": 0.422,
+      "step": 5824
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2987501605898793e-05,
+      "loss": 0.4314,
+      "step": 5825
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2985265789123692e-05,
+      "loss": 0.4286,
+      "step": 5826
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2983029808506606e-05,
+      "loss": 0.4236,
+      "step": 5827
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2980793664170259e-05,
+      "loss": 0.472,
+      "step": 5828
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2978557356237377e-05,
+      "loss": 0.4249,
+      "step": 5829
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2976320884830698e-05,
+      "loss": 0.4589,
+      "step": 5830
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2974084250072962e-05,
+      "loss": 0.4137,
+      "step": 5831
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2971847452086927e-05,
+      "loss": 0.3902,
+      "step": 5832
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.296961049099536e-05,
+      "loss": 0.4126,
+      "step": 5833
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2967373366921027e-05,
+      "loss": 0.4374,
+      "step": 5834
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2965136079986713e-05,
+      "loss": 0.3919,
+      "step": 5835
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2962898630315206e-05,
+      "loss": 0.4309,
+      "step": 5836
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.296066101802931e-05,
+      "loss": 0.4268,
+      "step": 5837
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2958423243251827e-05,
+      "loss": 0.4271,
+      "step": 5838
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2956185306105578e-05,
+      "loss": 0.4101,
+      "step": 5839
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2953947206713385e-05,
+      "loss": 0.4202,
+      "step": 5840
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2951708945198088e-05,
+      "loss": 0.4241,
+      "step": 5841
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2949470521682529e-05,
+      "loss": 0.3973,
+      "step": 5842
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2947231936289557e-05,
+      "loss": 0.4482,
+      "step": 5843
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2944993189142037e-05,
+      "loss": 0.4141,
+      "step": 5844
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2942754280362839e-05,
+      "loss": 0.434,
+      "step": 5845
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2940515210074844e-05,
+      "loss": 0.4049,
+      "step": 5846
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2938275978400936e-05,
+      "loss": 0.4218,
+      "step": 5847
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2936036585464011e-05,
+      "loss": 0.406,
+      "step": 5848
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2933797031386984e-05,
+      "loss": 0.4223,
+      "step": 5849
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.293155731629276e-05,
+      "loss": 0.3871,
+      "step": 5850
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.292931744030427e-05,
+      "loss": 0.4007,
+      "step": 5851
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2927077403544437e-05,
+      "loss": 0.4149,
+      "step": 5852
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.292483720613621e-05,
+      "loss": 0.4379,
+      "step": 5853
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2922596848202535e-05,
+      "loss": 0.4008,
+      "step": 5854
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2920356329866377e-05,
+      "loss": 0.4189,
+      "step": 5855
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2918115651250694e-05,
+      "loss": 0.4338,
+      "step": 5856
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2915874812478469e-05,
+      "loss": 0.4103,
+      "step": 5857
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2913633813672687e-05,
+      "loss": 0.3984,
+      "step": 5858
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2911392654956339e-05,
+      "loss": 0.4067,
+      "step": 5859
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.290915133645243e-05,
+      "loss": 0.4137,
+      "step": 5860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2906909858283969e-05,
+      "loss": 0.4135,
+      "step": 5861
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2904668220573979e-05,
+      "loss": 0.4315,
+      "step": 5862
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.290242642344549e-05,
+      "loss": 0.4055,
+      "step": 5863
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2900184467021534e-05,
+      "loss": 0.4054,
+      "step": 5864
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2897942351425163e-05,
+      "loss": 0.4717,
+      "step": 5865
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2895700076779435e-05,
+      "loss": 0.4403,
+      "step": 5866
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2893457643207406e-05,
+      "loss": 0.4259,
+      "step": 5867
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.289121505083215e-05,
+      "loss": 0.4595,
+      "step": 5868
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2888972299776754e-05,
+      "loss": 0.4265,
+      "step": 5869
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2886729390164305e-05,
+      "loss": 0.3649,
+      "step": 5870
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2884486322117902e-05,
+      "loss": 0.4218,
+      "step": 5871
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2882243095760649e-05,
+      "loss": 0.4202,
+      "step": 5872
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2879999711215669e-05,
+      "loss": 0.4629,
+      "step": 5873
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2877756168606084e-05,
+      "loss": 0.3949,
+      "step": 5874
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2875512468055025e-05,
+      "loss": 0.4662,
+      "step": 5875
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2873268609685638e-05,
+      "loss": 0.4666,
+      "step": 5876
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2871024593621069e-05,
+      "loss": 0.4384,
+      "step": 5877
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2868780419984482e-05,
+      "loss": 0.4042,
+      "step": 5878
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2866536088899047e-05,
+      "loss": 0.4314,
+      "step": 5879
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2864291600487935e-05,
+      "loss": 0.4071,
+      "step": 5880
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2862046954874332e-05,
+      "loss": 0.434,
+      "step": 5881
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2859802152181438e-05,
+      "loss": 0.4501,
+      "step": 5882
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.285755719253245e-05,
+      "loss": 0.4099,
+      "step": 5883
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2855312076050585e-05,
+      "loss": 0.3942,
+      "step": 5884
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2853066802859057e-05,
+      "loss": 0.4282,
+      "step": 5885
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2850821373081094e-05,
+      "loss": 0.4165,
+      "step": 5886
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2848575786839943e-05,
+      "loss": 0.4173,
+      "step": 5887
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2846330044258838e-05,
+      "loss": 0.4415,
+      "step": 5888
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.284408414546104e-05,
+      "loss": 0.44,
+      "step": 5889
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.284183809056981e-05,
+      "loss": 0.429,
+      "step": 5890
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2839591879708419e-05,
+      "loss": 0.4096,
+      "step": 5891
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2837345513000151e-05,
+      "loss": 0.3829,
+      "step": 5892
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2835098990568287e-05,
+      "loss": 0.4113,
+      "step": 5893
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.283285231253613e-05,
+      "loss": 0.4368,
+      "step": 5894
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2830605479026984e-05,
+      "loss": 0.4487,
+      "step": 5895
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.282835849016416e-05,
+      "loss": 0.4393,
+      "step": 5896
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2826111346070987e-05,
+      "loss": 0.4214,
+      "step": 5897
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2823864046870793e-05,
+      "loss": 0.418,
+      "step": 5898
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2821616592686919e-05,
+      "loss": 0.4322,
+      "step": 5899
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2819368983642711e-05,
+      "loss": 0.4488,
+      "step": 5900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2817121219861525e-05,
+      "loss": 0.4281,
+      "step": 5901
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2814873301466733e-05,
+      "loss": 0.4216,
+      "step": 5902
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.28126252285817e-05,
+      "loss": 0.4193,
+      "step": 5903
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2810377001329815e-05,
+      "loss": 0.4434,
+      "step": 5904
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.280812861983446e-05,
+      "loss": 0.4177,
+      "step": 5905
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2805880084219047e-05,
+      "loss": 0.4351,
+      "step": 5906
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2803631394606972e-05,
+      "loss": 0.4499,
+      "step": 5907
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.280138255112166e-05,
+      "loss": 0.4065,
+      "step": 5908
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2799133553886528e-05,
+      "loss": 0.424,
+      "step": 5909
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2796884403025014e-05,
+      "loss": 0.4275,
+      "step": 5910
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2794635098660556e-05,
+      "loss": 0.4718,
+      "step": 5911
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2792385640916604e-05,
+      "loss": 0.4201,
+      "step": 5912
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2790136029916619e-05,
+      "loss": 0.4092,
+      "step": 5913
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2787886265784068e-05,
+      "loss": 0.4311,
+      "step": 5914
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2785636348642426e-05,
+      "loss": 0.4211,
+      "step": 5915
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.278338627861517e-05,
+      "loss": 0.4396,
+      "step": 5916
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2781136055825796e-05,
+      "loss": 0.4339,
+      "step": 5917
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2778885680397808e-05,
+      "loss": 0.4394,
+      "step": 5918
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.277663515245471e-05,
+      "loss": 0.4375,
+      "step": 5919
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.277438447212002e-05,
+      "loss": 0.4123,
+      "step": 5920
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2772133639517264e-05,
+      "loss": 0.4125,
+      "step": 5921
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2769882654769977e-05,
+      "loss": 0.4137,
+      "step": 5922
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2767631518001697e-05,
+      "loss": 0.4232,
+      "step": 5923
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2765380229335975e-05,
+      "loss": 0.3977,
+      "step": 5924
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2763128788896373e-05,
+      "loss": 0.4384,
+      "step": 5925
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2760877196806457e-05,
+      "loss": 0.4036,
+      "step": 5926
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2758625453189803e-05,
+      "loss": 0.4211,
+      "step": 5927
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2756373558169992e-05,
+      "loss": 0.43,
+      "step": 5928
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2754121511870619e-05,
+      "loss": 0.4022,
+      "step": 5929
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2751869314415282e-05,
+      "loss": 0.4383,
+      "step": 5930
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2749616965927589e-05,
+      "loss": 0.3936,
+      "step": 5931
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2747364466531162e-05,
+      "loss": 0.4142,
+      "step": 5932
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2745111816349615e-05,
+      "loss": 0.4248,
+      "step": 5933
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2742859015506594e-05,
+      "loss": 0.4126,
+      "step": 5934
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2740606064125738e-05,
+      "loss": 0.3896,
+      "step": 5935
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2738352962330693e-05,
+      "loss": 0.4168,
+      "step": 5936
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2736099710245116e-05,
+      "loss": 0.436,
+      "step": 5937
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2733846307992676e-05,
+      "loss": 0.4352,
+      "step": 5938
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.273159275569705e-05,
+      "loss": 0.4199,
+      "step": 5939
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.272933905348192e-05,
+      "loss": 0.401,
+      "step": 5940
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2727085201470972e-05,
+      "loss": 0.4162,
+      "step": 5941
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2724831199787913e-05,
+      "loss": 0.4258,
+      "step": 5942
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2722577048556446e-05,
+      "loss": 0.4301,
+      "step": 5943
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2720322747900288e-05,
+      "loss": 0.4353,
+      "step": 5944
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2718068297943162e-05,
+      "loss": 0.4373,
+      "step": 5945
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.27158136988088e-05,
+      "loss": 0.428,
+      "step": 5946
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2713558950620946e-05,
+      "loss": 0.418,
+      "step": 5947
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2711304053503347e-05,
+      "loss": 0.4747,
+      "step": 5948
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2709049007579756e-05,
+      "loss": 0.4201,
+      "step": 5949
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2706793812973941e-05,
+      "loss": 0.4779,
+      "step": 5950
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2704538469809678e-05,
+      "loss": 0.4476,
+      "step": 5951
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2702282978210742e-05,
+      "loss": 0.3858,
+      "step": 5952
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2700027338300928e-05,
+      "loss": 0.4502,
+      "step": 5953
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2697771550204028e-05,
+      "loss": 0.4244,
+      "step": 5954
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2695515614043852e-05,
+      "loss": 0.4212,
+      "step": 5955
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2693259529944212e-05,
+      "loss": 0.411,
+      "step": 5956
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2691003298028935e-05,
+      "loss": 0.4462,
+      "step": 5957
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.268874691842184e-05,
+      "loss": 0.4321,
+      "step": 5958
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2686490391246774e-05,
+      "loss": 0.4072,
+      "step": 5959
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2684233716627578e-05,
+      "loss": 0.4253,
+      "step": 5960
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2681976894688114e-05,
+      "loss": 0.4237,
+      "step": 5961
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2679719925552236e-05,
+      "loss": 0.3942,
+      "step": 5962
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.267746280934382e-05,
+      "loss": 0.4157,
+      "step": 5963
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.267520554618674e-05,
+      "loss": 0.4102,
+      "step": 5964
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2672948136204888e-05,
+      "loss": 0.3857,
+      "step": 5965
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2670690579522153e-05,
+      "loss": 0.4038,
+      "step": 5966
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.266843287626244e-05,
+      "loss": 0.4256,
+      "step": 5967
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2666175026549662e-05,
+      "loss": 0.4341,
+      "step": 5968
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2663917030507737e-05,
+      "loss": 0.4049,
+      "step": 5969
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2661658888260588e-05,
+      "loss": 0.4404,
+      "step": 5970
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2659400599932155e-05,
+      "loss": 0.4193,
+      "step": 5971
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2657142165646377e-05,
+      "loss": 0.4494,
+      "step": 5972
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2654883585527208e-05,
+      "loss": 0.3979,
+      "step": 5973
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2652624859698604e-05,
+      "loss": 0.4154,
+      "step": 5974
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2650365988284534e-05,
+      "loss": 0.3829,
+      "step": 5975
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2648106971408973e-05,
+      "loss": 0.4155,
+      "step": 5976
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2645847809195901e-05,
+      "loss": 0.3629,
+      "step": 5977
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2643588501769314e-05,
+      "loss": 0.4336,
+      "step": 5978
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2641329049253205e-05,
+      "loss": 0.4252,
+      "step": 5979
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2639069451771584e-05,
+      "loss": 0.4151,
+      "step": 5980
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2636809709448466e-05,
+      "loss": 0.4137,
+      "step": 5981
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2634549822407871e-05,
+      "loss": 0.4152,
+      "step": 5982
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2632289790773833e-05,
+      "loss": 0.4365,
+      "step": 5983
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2630029614670387e-05,
+      "loss": 0.3894,
+      "step": 5984
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2627769294221585e-05,
+      "loss": 0.4214,
+      "step": 5985
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2625508829551475e-05,
+      "loss": 0.4317,
+      "step": 5986
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2623248220784123e-05,
+      "loss": 0.4448,
+      "step": 5987
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2620987468043597e-05,
+      "loss": 0.4321,
+      "step": 5988
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.261872657145398e-05,
+      "loss": 0.4103,
+      "step": 5989
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2616465531139353e-05,
+      "loss": 0.4414,
+      "step": 5990
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2614204347223812e-05,
+      "loss": 0.4676,
+      "step": 5991
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2611943019831457e-05,
+      "loss": 0.39,
+      "step": 5992
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2609681549086401e-05,
+      "loss": 0.4412,
+      "step": 5993
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2607419935112757e-05,
+      "loss": 0.4217,
+      "step": 5994
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2605158178034656e-05,
+      "loss": 0.4224,
+      "step": 5995
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2602896277976227e-05,
+      "loss": 0.4415,
+      "step": 5996
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2600634235061611e-05,
+      "loss": 0.4223,
+      "step": 5997
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.259837204941496e-05,
+      "loss": 0.4228,
+      "step": 5998
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2596109721160428e-05,
+      "loss": 0.4264,
+      "step": 5999
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2593847250422182e-05,
+      "loss": 0.4256,
+      "step": 6000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.259158463732439e-05,
+      "loss": 0.4088,
+      "step": 6001
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2589321881991236e-05,
+      "loss": 0.4216,
+      "step": 6002
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2587058984546912e-05,
+      "loss": 0.4354,
+      "step": 6003
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2584795945115605e-05,
+      "loss": 0.4459,
+      "step": 6004
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2582532763821523e-05,
+      "loss": 0.4028,
+      "step": 6005
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2580269440788877e-05,
+      "loss": 0.4246,
+      "step": 6006
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.257800597614189e-05,
+      "loss": 0.4324,
+      "step": 6007
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2575742370004781e-05,
+      "loss": 0.42,
+      "step": 6008
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2573478622501791e-05,
+      "loss": 0.4265,
+      "step": 6009
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.257121473375716e-05,
+      "loss": 0.4394,
+      "step": 6010
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.256895070389514e-05,
+      "loss": 0.408,
+      "step": 6011
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.256668653303999e-05,
+      "loss": 0.4185,
+      "step": 6012
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2564422221315968e-05,
+      "loss": 0.429,
+      "step": 6013
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2562157768847356e-05,
+      "loss": 0.3787,
+      "step": 6014
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2559893175758433e-05,
+      "loss": 0.4267,
+      "step": 6015
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2557628442173489e-05,
+      "loss": 0.4421,
+      "step": 6016
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2555363568216817e-05,
+      "loss": 0.4399,
+      "step": 6017
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2553098554012722e-05,
+      "loss": 0.3969,
+      "step": 6018
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2550833399685516e-05,
+      "loss": 0.4002,
+      "step": 6019
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2548568105359523e-05,
+      "loss": 0.4411,
+      "step": 6020
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2546302671159067e-05,
+      "loss": 0.4276,
+      "step": 6021
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2544037097208481e-05,
+      "loss": 0.4494,
+      "step": 6022
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2541771383632111e-05,
+      "loss": 0.4209,
+      "step": 6023
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2539505530554305e-05,
+      "loss": 0.4217,
+      "step": 6024
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2537239538099426e-05,
+      "loss": 0.4493,
+      "step": 6025
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2534973406391835e-05,
+      "loss": 0.4459,
+      "step": 6026
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2532707135555904e-05,
+      "loss": 0.4165,
+      "step": 6027
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.253044072571602e-05,
+      "loss": 0.4282,
+      "step": 6028
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2528174176996567e-05,
+      "loss": 0.4085,
+      "step": 6029
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2525907489521942e-05,
+      "loss": 0.4204,
+      "step": 6030
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.252364066341655e-05,
+      "loss": 0.4141,
+      "step": 6031
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2521373698804802e-05,
+      "loss": 0.4387,
+      "step": 6032
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2519106595811117e-05,
+      "loss": 0.4114,
+      "step": 6033
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2516839354559918e-05,
+      "loss": 0.4787,
+      "step": 6034
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2514571975175645e-05,
+      "loss": 0.4166,
+      "step": 6035
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.251230445778274e-05,
+      "loss": 0.4398,
+      "step": 6036
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2510036802505647e-05,
+      "loss": 0.4131,
+      "step": 6037
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2507769009468828e-05,
+      "loss": 0.4331,
+      "step": 6038
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2505501078796743e-05,
+      "loss": 0.4591,
+      "step": 6039
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2503233010613866e-05,
+      "loss": 0.4196,
+      "step": 6040
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.250096480504468e-05,
+      "loss": 0.4031,
+      "step": 6041
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.249869646221367e-05,
+      "loss": 0.4274,
+      "step": 6042
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2496427982245327e-05,
+      "loss": 0.4225,
+      "step": 6043
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2494159365264155e-05,
+      "loss": 0.3704,
+      "step": 6044
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.249189061139467e-05,
+      "loss": 0.4561,
+      "step": 6045
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2489621720761383e-05,
+      "loss": 0.4415,
+      "step": 6046
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2487352693488817e-05,
+      "loss": 0.4333,
+      "step": 6047
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2485083529701507e-05,
+      "loss": 0.3836,
+      "step": 6048
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2482814229523996e-05,
+      "loss": 0.4463,
+      "step": 6049
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2480544793080828e-05,
+      "loss": 0.42,
+      "step": 6050
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2478275220496558e-05,
+      "loss": 0.4429,
+      "step": 6051
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2476005511895746e-05,
+      "loss": 0.4249,
+      "step": 6052
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2473735667402966e-05,
+      "loss": 0.4259,
+      "step": 6053
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2471465687142793e-05,
+      "loss": 0.4321,
+      "step": 6054
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2469195571239811e-05,
+      "loss": 0.4336,
+      "step": 6055
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2466925319818612e-05,
+      "loss": 0.4078,
+      "step": 6056
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.24646549330038e-05,
+      "loss": 0.4133,
+      "step": 6057
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2462384410919975e-05,
+      "loss": 0.4252,
+      "step": 6058
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2460113753691755e-05,
+      "loss": 0.4141,
+      "step": 6059
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.245784296144376e-05,
+      "loss": 0.4032,
+      "step": 6060
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2455572034300618e-05,
+      "loss": 0.4077,
+      "step": 6061
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2453300972386972e-05,
+      "loss": 0.3931,
+      "step": 6062
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2451029775827459e-05,
+      "loss": 0.4176,
+      "step": 6063
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2448758444746733e-05,
+      "loss": 0.3984,
+      "step": 6064
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2446486979269454e-05,
+      "loss": 0.4231,
+      "step": 6065
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2444215379520286e-05,
+      "loss": 0.4167,
+      "step": 6066
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2441943645623904e-05,
+      "loss": 0.41,
+      "step": 6067
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2439671777704987e-05,
+      "loss": 0.3948,
+      "step": 6068
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2437399775888224e-05,
+      "loss": 0.45,
+      "step": 6069
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2435127640298312e-05,
+      "loss": 0.4462,
+      "step": 6070
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2432855371059951e-05,
+      "loss": 0.4209,
+      "step": 6071
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2430582968297855e-05,
+      "loss": 0.4328,
+      "step": 6072
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2428310432136736e-05,
+      "loss": 0.3896,
+      "step": 6073
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2426037762701326e-05,
+      "loss": 0.3914,
+      "step": 6074
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.242376496011635e-05,
+      "loss": 0.4151,
+      "step": 6075
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2421492024506555e-05,
+      "loss": 0.3952,
+      "step": 6076
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2419218955996677e-05,
+      "loss": 0.4408,
+      "step": 6077
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2416945754711482e-05,
+      "loss": 0.4459,
+      "step": 6078
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2414672420775724e-05,
+      "loss": 0.3952,
+      "step": 6079
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2412398954314176e-05,
+      "loss": 0.4223,
+      "step": 6080
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.241012535545161e-05,
+      "loss": 0.4076,
+      "step": 6081
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2407851624312808e-05,
+      "loss": 0.4222,
+      "step": 6082
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2405577761022565e-05,
+      "loss": 0.4457,
+      "step": 6083
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2403303765705679e-05,
+      "loss": 0.4252,
+      "step": 6084
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2401029638486952e-05,
+      "loss": 0.3841,
+      "step": 6085
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2398755379491193e-05,
+      "loss": 0.4021,
+      "step": 6086
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2396480988843231e-05,
+      "loss": 0.3915,
+      "step": 6087
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2394206466667885e-05,
+      "loss": 0.407,
+      "step": 6088
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.239193181308999e-05,
+      "loss": 0.45,
+      "step": 6089
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2389657028234388e-05,
+      "loss": 0.4436,
+      "step": 6090
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2387382112225928e-05,
+      "loss": 0.4084,
+      "step": 6091
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2385107065189462e-05,
+      "loss": 0.4504,
+      "step": 6092
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2382831887249857e-05,
+      "loss": 0.4302,
+      "step": 6093
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2380556578531981e-05,
+      "loss": 0.439,
+      "step": 6094
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2378281139160713e-05,
+      "loss": 0.4157,
+      "step": 6095
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2376005569260932e-05,
+      "loss": 0.408,
+      "step": 6096
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2373729868957535e-05,
+      "loss": 0.4372,
+      "step": 6097
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2371454038375416e-05,
+      "loss": 0.4271,
+      "step": 6098
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2369178077639487e-05,
+      "loss": 0.4328,
+      "step": 6099
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2366901986874652e-05,
+      "loss": 0.3742,
+      "step": 6100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.236462576620584e-05,
+      "loss": 0.4366,
+      "step": 6101
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2362349415757973e-05,
+      "loss": 0.4336,
+      "step": 6102
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2360072935655982e-05,
+      "loss": 0.4401,
+      "step": 6103
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2357796326024814e-05,
+      "loss": 0.4284,
+      "step": 6104
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2355519586989417e-05,
+      "loss": 0.4139,
+      "step": 6105
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2353242718674745e-05,
+      "loss": 0.4262,
+      "step": 6106
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.235096572120576e-05,
+      "loss": 0.4319,
+      "step": 6107
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.234868859470743e-05,
+      "loss": 0.4167,
+      "step": 6108
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2346411339304735e-05,
+      "loss": 0.4503,
+      "step": 6109
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2344133955122662e-05,
+      "loss": 0.4308,
+      "step": 6110
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2341856442286196e-05,
+      "loss": 0.4478,
+      "step": 6111
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2339578800920334e-05,
+      "loss": 0.4489,
+      "step": 6112
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2337301031150086e-05,
+      "loss": 0.4754,
+      "step": 6113
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2335023133100466e-05,
+      "loss": 0.4253,
+      "step": 6114
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2332745106896482e-05,
+      "loss": 0.3995,
+      "step": 6115
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2330466952663172e-05,
+      "loss": 0.4505,
+      "step": 6116
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2328188670525563e-05,
+      "loss": 0.4399,
+      "step": 6117
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2325910260608698e-05,
+      "loss": 0.4248,
+      "step": 6118
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2323631723037624e-05,
+      "loss": 0.4306,
+      "step": 6119
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2321353057937389e-05,
+      "loss": 0.4376,
+      "step": 6120
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2319074265433064e-05,
+      "loss": 0.4352,
+      "step": 6121
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2316795345649714e-05,
+      "loss": 0.4043,
+      "step": 6122
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.231451629871241e-05,
+      "loss": 0.4421,
+      "step": 6123
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2312237124746235e-05,
+      "loss": 0.4019,
+      "step": 6124
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2309957823876286e-05,
+      "loss": 0.3915,
+      "step": 6125
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2307678396227651e-05,
+      "loss": 0.4121,
+      "step": 6126
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2305398841925435e-05,
+      "loss": 0.4185,
+      "step": 6127
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2303119161094748e-05,
+      "loss": 0.4963,
+      "step": 6128
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2300839353860705e-05,
+      "loss": 0.4378,
+      "step": 6129
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2298559420348437e-05,
+      "loss": 0.3946,
+      "step": 6130
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.229627936068307e-05,
+      "loss": 0.4162,
+      "step": 6131
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.229399917498974e-05,
+      "loss": 0.4168,
+      "step": 6132
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.229171886339359e-05,
+      "loss": 0.4051,
+      "step": 6133
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2289438426019781e-05,
+      "loss": 0.413,
+      "step": 6134
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2287157862993464e-05,
+      "loss": 0.4074,
+      "step": 6135
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2284877174439807e-05,
+      "loss": 0.4058,
+      "step": 6136
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2282596360483979e-05,
+      "loss": 0.4223,
+      "step": 6137
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2280315421251163e-05,
+      "loss": 0.4504,
+      "step": 6138
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2278034356866544e-05,
+      "loss": 0.4365,
+      "step": 6139
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2275753167455314e-05,
+      "loss": 0.4254,
+      "step": 6140
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2273471853142671e-05,
+      "loss": 0.4297,
+      "step": 6141
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2271190414053828e-05,
+      "loss": 0.4261,
+      "step": 6142
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2268908850313994e-05,
+      "loss": 0.4435,
+      "step": 6143
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2266627162048388e-05,
+      "loss": 0.3758,
+      "step": 6144
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2264345349382238e-05,
+      "loss": 0.4292,
+      "step": 6145
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.226206341244078e-05,
+      "loss": 0.4019,
+      "step": 6146
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2259781351349255e-05,
+      "loss": 0.4024,
+      "step": 6147
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2257499166232907e-05,
+      "loss": 0.4547,
+      "step": 6148
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2255216857216993e-05,
+      "loss": 0.4173,
+      "step": 6149
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2252934424426774e-05,
+      "loss": 0.4659,
+      "step": 6150
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.225065186798752e-05,
+      "loss": 0.379,
+      "step": 6151
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.22483691880245e-05,
+      "loss": 0.3964,
+      "step": 6152
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2246086384663002e-05,
+      "loss": 0.3951,
+      "step": 6153
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2243803458028312e-05,
+      "loss": 0.403,
+      "step": 6154
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2241520408245724e-05,
+      "loss": 0.4455,
+      "step": 6155
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2239237235440542e-05,
+      "loss": 0.3954,
+      "step": 6156
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.223695393973807e-05,
+      "loss": 0.4045,
+      "step": 6157
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2234670521263628e-05,
+      "loss": 0.4645,
+      "step": 6158
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2232386980142538e-05,
+      "loss": 0.4443,
+      "step": 6159
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2230103316500129e-05,
+      "loss": 0.4177,
+      "step": 6160
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2227819530461729e-05,
+      "loss": 0.4447,
+      "step": 6161
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2225535622152689e-05,
+      "loss": 0.4435,
+      "step": 6162
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2223251591698358e-05,
+      "loss": 0.4247,
+      "step": 6163
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.222096743922409e-05,
+      "loss": 0.4394,
+      "step": 6164
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2218683164855241e-05,
+      "loss": 0.4115,
+      "step": 6165
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.221639876871719e-05,
+      "loss": 0.4369,
+      "step": 6166
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2214114250935305e-05,
+      "loss": 0.4199,
+      "step": 6167
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2211829611634976e-05,
+      "loss": 0.3939,
+      "step": 6168
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2209544850941583e-05,
+      "loss": 0.418,
+      "step": 6169
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2207259968980532e-05,
+      "loss": 0.3863,
+      "step": 6170
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2204974965877216e-05,
+      "loss": 0.4144,
+      "step": 6171
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2202689841757051e-05,
+      "loss": 0.4465,
+      "step": 6172
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2200404596745447e-05,
+      "loss": 0.4023,
+      "step": 6173
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2198119230967832e-05,
+      "loss": 0.393,
+      "step": 6174
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.219583374454963e-05,
+      "loss": 0.4487,
+      "step": 6175
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2193548137616282e-05,
+      "loss": 0.4113,
+      "step": 6176
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2191262410293223e-05,
+      "loss": 0.4128,
+      "step": 6177
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.218897656270591e-05,
+      "loss": 0.4101,
+      "step": 6178
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.218669059497979e-05,
+      "loss": 0.4413,
+      "step": 6179
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2184404507240333e-05,
+      "loss": 0.4393,
+      "step": 6180
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2182118299613001e-05,
+      "loss": 0.397,
+      "step": 6181
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2179831972223276e-05,
+      "loss": 0.4394,
+      "step": 6182
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2177545525196633e-05,
+      "loss": 0.4346,
+      "step": 6183
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2175258958658564e-05,
+      "loss": 0.4764,
+      "step": 6184
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2172972272734562e-05,
+      "loss": 0.4311,
+      "step": 6185
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2170685467550131e-05,
+      "loss": 0.4032,
+      "step": 6186
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2168398543230779e-05,
+      "loss": 0.4391,
+      "step": 6187
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2166111499902015e-05,
+      "loss": 0.403,
+      "step": 6188
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2163824337689367e-05,
+      "loss": 0.4476,
+      "step": 6189
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.216153705671836e-05,
+      "loss": 0.4385,
+      "step": 6190
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.215924965711453e-05,
+      "loss": 0.4428,
+      "step": 6191
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2156962139003409e-05,
+      "loss": 0.4112,
+      "step": 6192
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2154674502510556e-05,
+      "loss": 0.4073,
+      "step": 6193
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2152386747761515e-05,
+      "loss": 0.4094,
+      "step": 6194
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2150098874881854e-05,
+      "loss": 0.4296,
+      "step": 6195
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2147810883997133e-05,
+      "loss": 0.4063,
+      "step": 6196
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.214552277523293e-05,
+      "loss": 0.44,
+      "step": 6197
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2143234548714822e-05,
+      "loss": 0.3998,
+      "step": 6198
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2140946204568396e-05,
+      "loss": 0.4427,
+      "step": 6199
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2138657742919242e-05,
+      "loss": 0.409,
+      "step": 6200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.213636916389296e-05,
+      "loss": 0.4224,
+      "step": 6201
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2134080467615159e-05,
+      "loss": 0.4246,
+      "step": 6202
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2131791654211447e-05,
+      "loss": 0.4272,
+      "step": 6203
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2129502723807443e-05,
+      "loss": 0.4045,
+      "step": 6204
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2127213676528768e-05,
+      "loss": 0.4448,
+      "step": 6205
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2124924512501061e-05,
+      "loss": 0.4073,
+      "step": 6206
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2122635231849955e-05,
+      "loss": 0.4292,
+      "step": 6207
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2120345834701094e-05,
+      "loss": 0.4338,
+      "step": 6208
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2118056321180124e-05,
+      "loss": 0.421,
+      "step": 6209
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2115766691412712e-05,
+      "loss": 0.4341,
+      "step": 6210
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2113476945524513e-05,
+      "loss": 0.4317,
+      "step": 6211
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2111187083641198e-05,
+      "loss": 0.4101,
+      "step": 6212
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2108897105888441e-05,
+      "loss": 0.4485,
+      "step": 6213
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2106607012391929e-05,
+      "loss": 0.4207,
+      "step": 6214
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2104316803277347e-05,
+      "loss": 0.4383,
+      "step": 6215
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.210202647867039e-05,
+      "loss": 0.4228,
+      "step": 6216
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2099736038696758e-05,
+      "loss": 0.4341,
+      "step": 6217
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.209744548348216e-05,
+      "loss": 0.4452,
+      "step": 6218
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2095154813152312e-05,
+      "loss": 0.4099,
+      "step": 6219
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2092864027832934e-05,
+      "loss": 0.4118,
+      "step": 6220
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2090573127649746e-05,
+      "loss": 0.425,
+      "step": 6221
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2088282112728484e-05,
+      "loss": 0.4262,
+      "step": 6222
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2085990983194893e-05,
+      "loss": 0.4348,
+      "step": 6223
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2083699739174712e-05,
+      "loss": 0.4071,
+      "step": 6224
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2081408380793695e-05,
+      "loss": 0.438,
+      "step": 6225
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2079116908177592e-05,
+      "loss": 0.4098,
+      "step": 6226
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2076825321452181e-05,
+      "loss": 0.4263,
+      "step": 6227
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2074533620743225e-05,
+      "loss": 0.3999,
+      "step": 6228
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2072241806176501e-05,
+      "loss": 0.4479,
+      "step": 6229
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2069949877877791e-05,
+      "loss": 0.4063,
+      "step": 6230
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2067657835972883e-05,
+      "loss": 0.4263,
+      "step": 6231
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2065365680587577e-05,
+      "loss": 0.4106,
+      "step": 6232
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2063073411847674e-05,
+      "loss": 0.4249,
+      "step": 6233
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2060781029878978e-05,
+      "loss": 0.4409,
+      "step": 6234
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2058488534807305e-05,
+      "loss": 0.4226,
+      "step": 6235
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2056195926758476e-05,
+      "loss": 0.4501,
+      "step": 6236
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2053903205858317e-05,
+      "loss": 0.3837,
+      "step": 6237
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2051610372232661e-05,
+      "loss": 0.4173,
+      "step": 6238
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2049317426007343e-05,
+      "loss": 0.4264,
+      "step": 6239
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2047024367308215e-05,
+      "loss": 0.4387,
+      "step": 6240
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2044731196261125e-05,
+      "loss": 0.4363,
+      "step": 6241
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.204243791299193e-05,
+      "loss": 0.4106,
+      "step": 6242
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2040144517626488e-05,
+      "loss": 0.4065,
+      "step": 6243
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2037851010290683e-05,
+      "loss": 0.4073,
+      "step": 6244
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2035557391110375e-05,
+      "loss": 0.4033,
+      "step": 6245
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2033263660211456e-05,
+      "loss": 0.4433,
+      "step": 6246
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.203096981771981e-05,
+      "loss": 0.3897,
+      "step": 6247
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2028675863761332e-05,
+      "loss": 0.4303,
+      "step": 6248
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2026381798461925e-05,
+      "loss": 0.391,
+      "step": 6249
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.202408762194749e-05,
+      "loss": 0.4186,
+      "step": 6250
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2021793334343945e-05,
+      "loss": 0.4463,
+      "step": 6251
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2019498935777206e-05,
+      "loss": 0.4077,
+      "step": 6252
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2017204426373197e-05,
+      "loss": 0.4369,
+      "step": 6253
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.201490980625785e-05,
+      "loss": 0.4056,
+      "step": 6254
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2012615075557104e-05,
+      "loss": 0.4196,
+      "step": 6255
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2010320234396895e-05,
+      "loss": 0.411,
+      "step": 6256
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2008025282903181e-05,
+      "loss": 0.4229,
+      "step": 6257
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.200573022120191e-05,
+      "loss": 0.4093,
+      "step": 6258
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2003435049419048e-05,
+      "loss": 0.4284,
+      "step": 6259
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2001139767680559e-05,
+      "loss": 0.4297,
+      "step": 6260
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1998844376112417e-05,
+      "loss": 0.4494,
+      "step": 6261
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1996548874840601e-05,
+      "loss": 0.3916,
+      "step": 6262
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1994253263991097e-05,
+      "loss": 0.4166,
+      "step": 6263
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1991957543689899e-05,
+      "loss": 0.4219,
+      "step": 6264
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1989661714063e-05,
+      "loss": 0.4355,
+      "step": 6265
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.19873657752364e-05,
+      "loss": 0.4179,
+      "step": 6266
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1985069727336118e-05,
+      "loss": 0.4093,
+      "step": 6267
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1982773570488162e-05,
+      "loss": 0.4274,
+      "step": 6268
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1980477304818559e-05,
+      "loss": 0.4451,
+      "step": 6269
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1978180930453328e-05,
+      "loss": 0.3783,
+      "step": 6270
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1975884447518509e-05,
+      "loss": 0.4376,
+      "step": 6271
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1973587856140142e-05,
+      "loss": 0.4113,
+      "step": 6272
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1971291156444262e-05,
+      "loss": 0.4375,
+      "step": 6273
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.196899434855693e-05,
+      "loss": 0.383,
+      "step": 6274
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1966697432604202e-05,
+      "loss": 0.4505,
+      "step": 6275
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1964400408712139e-05,
+      "loss": 0.4067,
+      "step": 6276
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1962103277006806e-05,
+      "loss": 0.4166,
+      "step": 6277
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1959806037614284e-05,
+      "loss": 0.4103,
+      "step": 6278
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.195750869066065e-05,
+      "loss": 0.4678,
+      "step": 6279
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1955211236271992e-05,
+      "loss": 0.436,
+      "step": 6280
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1952913674574402e-05,
+      "loss": 0.3881,
+      "step": 6281
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1950616005693978e-05,
+      "loss": 0.419,
+      "step": 6282
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1948318229756828e-05,
+      "loss": 0.4626,
+      "step": 6283
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1946020346889057e-05,
+      "loss": 0.4209,
+      "step": 6284
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.194372235721678e-05,
+      "loss": 0.4222,
+      "step": 6285
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1941424260866123e-05,
+      "loss": 0.4504,
+      "step": 6286
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1939126057963215e-05,
+      "loss": 0.393,
+      "step": 6287
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1936827748634185e-05,
+      "loss": 0.3939,
+      "step": 6288
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1934529333005176e-05,
+      "loss": 0.4316,
+      "step": 6289
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1932230811202325e-05,
+      "loss": 0.4473,
+      "step": 6290
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1929932183351796e-05,
+      "loss": 0.4232,
+      "step": 6291
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1927633449579736e-05,
+      "loss": 0.4039,
+      "step": 6292
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1925334610012314e-05,
+      "loss": 0.443,
+      "step": 6293
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1923035664775692e-05,
+      "loss": 0.4532,
+      "step": 6294
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1920736613996046e-05,
+      "loss": 0.4138,
+      "step": 6295
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1918437457799563e-05,
+      "loss": 0.4299,
+      "step": 6296
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1916138196312423e-05,
+      "loss": 0.4123,
+      "step": 6297
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1913838829660815e-05,
+      "loss": 0.4296,
+      "step": 6298
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1911539357970941e-05,
+      "loss": 0.4157,
+      "step": 6299
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1909239781369003e-05,
+      "loss": 0.4299,
+      "step": 6300
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.190694009998121e-05,
+      "loss": 0.4559,
+      "step": 6301
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1904640313933775e-05,
+      "loss": 0.4125,
+      "step": 6302
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.190234042335292e-05,
+      "loss": 0.4056,
+      "step": 6303
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1900040428364872e-05,
+      "loss": 0.413,
+      "step": 6304
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1897740329095863e-05,
+      "loss": 0.416,
+      "step": 6305
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1895440125672126e-05,
+      "loss": 0.403,
+      "step": 6306
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.189313981821991e-05,
+      "loss": 0.4214,
+      "step": 6307
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1890839406865462e-05,
+      "loss": 0.4287,
+      "step": 6308
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1888538891735037e-05,
+      "loss": 0.4227,
+      "step": 6309
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1886238272954897e-05,
+      "loss": 0.4322,
+      "step": 6310
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1883937550651299e-05,
+      "loss": 0.3798,
+      "step": 6311
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1881636724950527e-05,
+      "loss": 0.4135,
+      "step": 6312
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1879335795978855e-05,
+      "loss": 0.4251,
+      "step": 6313
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1877034763862563e-05,
+      "loss": 0.4224,
+      "step": 6314
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1874733628727939e-05,
+      "loss": 0.4063,
+      "step": 6315
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1872432390701281e-05,
+      "loss": 0.4829,
+      "step": 6316
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1870131049908889e-05,
+      "loss": 0.4207,
+      "step": 6317
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1867829606477069e-05,
+      "loss": 0.4429,
+      "step": 6318
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1865528060532127e-05,
+      "loss": 0.4137,
+      "step": 6319
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1863226412200386e-05,
+      "loss": 0.4109,
+      "step": 6320
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1860924661608168e-05,
+      "loss": 0.4153,
+      "step": 6321
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1858622808881801e-05,
+      "loss": 0.4215,
+      "step": 6322
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1856320854147614e-05,
+      "loss": 0.4282,
+      "step": 6323
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1854018797531952e-05,
+      "loss": 0.3996,
+      "step": 6324
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.185171663916116e-05,
+      "loss": 0.4638,
+      "step": 6325
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1849414379161585e-05,
+      "loss": 0.4187,
+      "step": 6326
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1847112017659583e-05,
+      "loss": 0.4429,
+      "step": 6327
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.184480955478152e-05,
+      "loss": 0.4483,
+      "step": 6328
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1842506990653764e-05,
+      "loss": 0.4263,
+      "step": 6329
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1840204325402682e-05,
+      "loss": 0.4001,
+      "step": 6330
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1837901559154656e-05,
+      "loss": 0.4265,
+      "step": 6331
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1835598692036067e-05,
+      "loss": 0.3939,
+      "step": 6332
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1833295724173312e-05,
+      "loss": 0.4249,
+      "step": 6333
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.183099265569278e-05,
+      "loss": 0.4266,
+      "step": 6334
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.182868948672087e-05,
+      "loss": 0.4102,
+      "step": 6335
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1826386217383993e-05,
+      "loss": 0.4166,
+      "step": 6336
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1824082847808558e-05,
+      "loss": 0.418,
+      "step": 6337
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1821779378120983e-05,
+      "loss": 0.4219,
+      "step": 6338
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1819475808447689e-05,
+      "loss": 0.4114,
+      "step": 6339
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1817172138915105e-05,
+      "loss": 0.4432,
+      "step": 6340
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1814868369649664e-05,
+      "loss": 0.4239,
+      "step": 6341
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1812564500777807e-05,
+      "loss": 0.4462,
+      "step": 6342
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1810260532425976e-05,
+      "loss": 0.4035,
+      "step": 6343
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1807956464720624e-05,
+      "loss": 0.425,
+      "step": 6344
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1805652297788204e-05,
+      "loss": 0.4683,
+      "step": 6345
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.180334803175518e-05,
+      "loss": 0.4398,
+      "step": 6346
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.180104366674801e-05,
+      "loss": 0.3878,
+      "step": 6347
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1798739202893178e-05,
+      "loss": 0.4269,
+      "step": 6348
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1796434640317147e-05,
+      "loss": 0.4159,
+      "step": 6349
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1794129979146413e-05,
+      "loss": 0.4059,
+      "step": 6350
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1791825219507456e-05,
+      "loss": 0.4046,
+      "step": 6351
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1789520361526771e-05,
+      "loss": 0.4278,
+      "step": 6352
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1787215405330862e-05,
+      "loss": 0.4622,
+      "step": 6353
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1784910351046224e-05,
+      "loss": 0.4166,
+      "step": 6354
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1782605198799371e-05,
+      "loss": 0.437,
+      "step": 6355
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.178029994871682e-05,
+      "loss": 0.4256,
+      "step": 6356
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1777994600925088e-05,
+      "loss": 0.4038,
+      "step": 6357
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1775689155550702e-05,
+      "loss": 0.4353,
+      "step": 6358
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1773383612720193e-05,
+      "loss": 0.4591,
+      "step": 6359
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1771077972560098e-05,
+      "loss": 0.4127,
+      "step": 6360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.176877223519696e-05,
+      "loss": 0.3851,
+      "step": 6361
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1766466400757322e-05,
+      "loss": 0.4011,
+      "step": 6362
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.176416046936774e-05,
+      "loss": 0.4638,
+      "step": 6363
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1761854441154767e-05,
+      "loss": 0.4606,
+      "step": 6364
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1759548316244973e-05,
+      "loss": 0.4176,
+      "step": 6365
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1757242094764921e-05,
+      "loss": 0.4575,
+      "step": 6366
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1754935776841186e-05,
+      "loss": 0.3794,
+      "step": 6367
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1752629362600347e-05,
+      "loss": 0.4035,
+      "step": 6368
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.175032285216899e-05,
+      "loss": 0.426,
+      "step": 6369
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1748016245673699e-05,
+      "loss": 0.4257,
+      "step": 6370
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1745709543241077e-05,
+      "loss": 0.3823,
+      "step": 6371
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1743402744997716e-05,
+      "loss": 0.3906,
+      "step": 6372
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1741095851070228e-05,
+      "loss": 0.4174,
+      "step": 6373
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.173878886158522e-05,
+      "loss": 0.3859,
+      "step": 6374
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.4451,
+      "step": 6375
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1734174596449108e-05,
+      "loss": 0.4551,
+      "step": 6376
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1731867321051257e-05,
+      "loss": 0.4277,
+      "step": 6377
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1729559950602381e-05,
+      "loss": 0.4105,
+      "step": 6378
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1727252485229113e-05,
+      "loss": 0.3959,
+      "step": 6379
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.17249449250581e-05,
+      "loss": 0.4315,
+      "step": 6380
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1722637270215988e-05,
+      "loss": 0.4397,
+      "step": 6381
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.172032952082943e-05,
+      "loss": 0.4327,
+      "step": 6382
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.171802167702508e-05,
+      "loss": 0.423,
+      "step": 6383
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1715713738929601e-05,
+      "loss": 0.4252,
+      "step": 6384
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1713405706669666e-05,
+      "loss": 0.4252,
+      "step": 6385
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1711097580371946e-05,
+      "loss": 0.4169,
+      "step": 6386
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1708789360163116e-05,
+      "loss": 0.4332,
+      "step": 6387
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.170648104616986e-05,
+      "loss": 0.4234,
+      "step": 6388
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1704172638518871e-05,
+      "loss": 0.4289,
+      "step": 6389
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1701864137336838e-05,
+      "loss": 0.3821,
+      "step": 6390
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.169955554275046e-05,
+      "loss": 0.4379,
+      "step": 6391
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1697246854886442e-05,
+      "loss": 0.4372,
+      "step": 6392
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1694938073871493e-05,
+      "loss": 0.4171,
+      "step": 6393
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1692629199832331e-05,
+      "loss": 0.4447,
+      "step": 6394
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1690320232895666e-05,
+      "loss": 0.4151,
+      "step": 6395
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1688011173188228e-05,
+      "loss": 0.402,
+      "step": 6396
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1685702020836747e-05,
+      "loss": 0.3797,
+      "step": 6397
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1683392775967957e-05,
+      "loss": 0.4677,
+      "step": 6398
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1681083438708598e-05,
+      "loss": 0.4263,
+      "step": 6399
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.167877400918541e-05,
+      "loss": 0.4212,
+      "step": 6400
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1676464487525148e-05,
+      "loss": 0.4135,
+      "step": 6401
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1674154873854566e-05,
+      "loss": 0.3708,
+      "step": 6402
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1671845168300422e-05,
+      "loss": 0.4425,
+      "step": 6403
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.166953537098948e-05,
+      "loss": 0.4279,
+      "step": 6404
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1667225482048513e-05,
+      "loss": 0.4071,
+      "step": 6405
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1664915501604295e-05,
+      "loss": 0.4239,
+      "step": 6406
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1662605429783607e-05,
+      "loss": 0.4108,
+      "step": 6407
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1660295266713228e-05,
+      "loss": 0.4142,
+      "step": 6408
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1657985012519951e-05,
+      "loss": 0.4174,
+      "step": 6409
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1655674667330579e-05,
+      "loss": 0.4717,
+      "step": 6410
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1653364231271901e-05,
+      "loss": 0.4405,
+      "step": 6411
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1651053704470727e-05,
+      "loss": 0.4128,
+      "step": 6412
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1648743087053864e-05,
+      "loss": 0.4196,
+      "step": 6413
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1646432379148132e-05,
+      "loss": 0.4274,
+      "step": 6414
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1644121580880345e-05,
+      "loss": 0.4094,
+      "step": 6415
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1641810692377331e-05,
+      "loss": 0.4062,
+      "step": 6416
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1639499713765918e-05,
+      "loss": 0.4317,
+      "step": 6417
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1637188645172946e-05,
+      "loss": 0.4588,
+      "step": 6418
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1634877486725248e-05,
+      "loss": 0.4271,
+      "step": 6419
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.163256623854967e-05,
+      "loss": 0.437,
+      "step": 6420
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1630254900773066e-05,
+      "loss": 0.4664,
+      "step": 6421
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1627943473522284e-05,
+      "loss": 0.4626,
+      "step": 6422
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1625631956924189e-05,
+      "loss": 0.4331,
+      "step": 6423
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.162332035110564e-05,
+      "loss": 0.4287,
+      "step": 6424
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1621008656193512e-05,
+      "loss": 0.3871,
+      "step": 6425
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1618696872314672e-05,
+      "loss": 0.4656,
+      "step": 6426
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1616384999596006e-05,
+      "loss": 0.4016,
+      "step": 6427
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1614073038164393e-05,
+      "loss": 0.3874,
+      "step": 6428
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1611760988146724e-05,
+      "loss": 0.4356,
+      "step": 6429
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1609448849669889e-05,
+      "loss": 0.4215,
+      "step": 6430
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1607136622860793e-05,
+      "loss": 0.3985,
+      "step": 6431
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1604824307846334e-05,
+      "loss": 0.4162,
+      "step": 6432
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1602511904753419e-05,
+      "loss": 0.4068,
+      "step": 6433
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1600199413708964e-05,
+      "loss": 0.4364,
+      "step": 6434
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1597886834839887e-05,
+      "loss": 0.4215,
+      "step": 6435
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1595574168273112e-05,
+      "loss": 0.4047,
+      "step": 6436
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.159326141413556e-05,
+      "loss": 0.4625,
+      "step": 6437
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1590948572554168e-05,
+      "loss": 0.4655,
+      "step": 6438
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.158863564365587e-05,
+      "loss": 0.4639,
+      "step": 6439
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1586322627567612e-05,
+      "loss": 0.401,
+      "step": 6440
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1584009524416337e-05,
+      "loss": 0.3992,
+      "step": 6441
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1581696334329e-05,
+      "loss": 0.4064,
+      "step": 6442
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1579383057432552e-05,
+      "loss": 0.4384,
+      "step": 6443
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1577069693853957e-05,
+      "loss": 0.3896,
+      "step": 6444
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.157475624372018e-05,
+      "loss": 0.4409,
+      "step": 6445
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1572442707158191e-05,
+      "loss": 0.4072,
+      "step": 6446
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1570129084294966e-05,
+      "loss": 0.4153,
+      "step": 6447
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1567815375257485e-05,
+      "loss": 0.416,
+      "step": 6448
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.156550158017273e-05,
+      "loss": 0.4582,
+      "step": 6449
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1563187699167692e-05,
+      "loss": 0.3843,
+      "step": 6450
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1560873732369366e-05,
+      "loss": 0.444,
+      "step": 6451
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1558559679904752e-05,
+      "loss": 0.438,
+      "step": 6452
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.155624554190085e-05,
+      "loss": 0.4404,
+      "step": 6453
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.155393131848467e-05,
+      "loss": 0.4343,
+      "step": 6454
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1551617009783223e-05,
+      "loss": 0.4165,
+      "step": 6455
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1549302615923528e-05,
+      "loss": 0.3949,
+      "step": 6456
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1546988137032604e-05,
+      "loss": 0.4629,
+      "step": 6457
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1544673573237486e-05,
+      "loss": 0.3804,
+      "step": 6458
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1542358924665197e-05,
+      "loss": 0.4243,
+      "step": 6459
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1540044191442777e-05,
+      "loss": 0.3845,
+      "step": 6460
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1537729373697264e-05,
+      "loss": 0.4699,
+      "step": 6461
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1535414471555708e-05,
+      "loss": 0.4028,
+      "step": 6462
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1533099485145156e-05,
+      "loss": 0.4283,
+      "step": 6463
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1530784414592662e-05,
+      "loss": 0.4273,
+      "step": 6464
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1528469260025286e-05,
+      "loss": 0.3999,
+      "step": 6465
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1526154021570095e-05,
+      "loss": 0.3904,
+      "step": 6466
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1523838699354153e-05,
+      "loss": 0.4178,
+      "step": 6467
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1521523293504535e-05,
+      "loss": 0.422,
+      "step": 6468
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1519207804148315e-05,
+      "loss": 0.4332,
+      "step": 6469
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1516892231412586e-05,
+      "loss": 0.3903,
+      "step": 6470
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1514576575424425e-05,
+      "loss": 0.4407,
+      "step": 6471
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1512260836310925e-05,
+      "loss": 0.4468,
+      "step": 6472
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1509945014199182e-05,
+      "loss": 0.4424,
+      "step": 6473
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.15076291092163e-05,
+      "loss": 0.4355,
+      "step": 6474
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1505313121489384e-05,
+      "loss": 0.3879,
+      "step": 6475
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1502997051145535e-05,
+      "loss": 0.4029,
+      "step": 6476
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1500680898311875e-05,
+      "loss": 0.419,
+      "step": 6477
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1498364663115526e-05,
+      "loss": 0.4022,
+      "step": 6478
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1496048345683603e-05,
+      "loss": 0.4441,
+      "step": 6479
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.149373194614324e-05,
+      "loss": 0.4057,
+      "step": 6480
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1491415464621562e-05,
+      "loss": 0.4182,
+      "step": 6481
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1489098901245712e-05,
+      "loss": 0.4135,
+      "step": 6482
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1486782256142832e-05,
+      "loss": 0.3909,
+      "step": 6483
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1484465529440064e-05,
+      "loss": 0.4222,
+      "step": 6484
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1482148721264556e-05,
+      "loss": 0.4339,
+      "step": 6485
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1479831831743467e-05,
+      "loss": 0.4122,
+      "step": 6486
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1477514861003958e-05,
+      "loss": 0.4251,
+      "step": 6487
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1475197809173188e-05,
+      "loss": 0.4326,
+      "step": 6488
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1472880676378324e-05,
+      "loss": 0.411,
+      "step": 6489
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1470563462746542e-05,
+      "loss": 0.4269,
+      "step": 6490
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.146824616840502e-05,
+      "loss": 0.4506,
+      "step": 6491
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1465928793480939e-05,
+      "loss": 0.4222,
+      "step": 6492
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.146361133810148e-05,
+      "loss": 0.4067,
+      "step": 6493
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1461293802393834e-05,
+      "loss": 0.4001,
+      "step": 6494
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1458976186485202e-05,
+      "loss": 0.4096,
+      "step": 6495
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1456658490502778e-05,
+      "loss": 0.392,
+      "step": 6496
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1454340714573765e-05,
+      "loss": 0.4426,
+      "step": 6497
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.145202285882537e-05,
+      "loss": 0.4132,
+      "step": 6498
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1449704923384813e-05,
+      "loss": 0.4167,
+      "step": 6499
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1447386908379302e-05,
+      "loss": 0.4315,
+      "step": 6500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.144506881393606e-05,
+      "loss": 0.4018,
+      "step": 6501
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1442750640182312e-05,
+      "loss": 0.3962,
+      "step": 6502
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1440432387245289e-05,
+      "loss": 0.4105,
+      "step": 6503
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1438114055252225e-05,
+      "loss": 0.4447,
+      "step": 6504
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1435795644330358e-05,
+      "loss": 0.4263,
+      "step": 6505
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1433477154606929e-05,
+      "loss": 0.4206,
+      "step": 6506
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1431158586209185e-05,
+      "loss": 0.3986,
+      "step": 6507
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1428839939264383e-05,
+      "loss": 0.4036,
+      "step": 6508
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.142652121389977e-05,
+      "loss": 0.4121,
+      "step": 6509
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.142420241024261e-05,
+      "loss": 0.4236,
+      "step": 6510
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1421883528420168e-05,
+      "loss": 0.4349,
+      "step": 6511
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1419564568559714e-05,
+      "loss": 0.4144,
+      "step": 6512
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1417245530788515e-05,
+      "loss": 0.4163,
+      "step": 6513
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1414926415233851e-05,
+      "loss": 0.4532,
+      "step": 6514
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1412607222023005e-05,
+      "loss": 0.3957,
+      "step": 6515
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1410287951283261e-05,
+      "loss": 0.4415,
+      "step": 6516
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.140796860314191e-05,
+      "loss": 0.4421,
+      "step": 6517
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1405649177726244e-05,
+      "loss": 0.4611,
+      "step": 6518
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1403329675163561e-05,
+      "loss": 0.4179,
+      "step": 6519
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1401010095581167e-05,
+      "loss": 0.4427,
+      "step": 6520
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1398690439106368e-05,
+      "loss": 0.393,
+      "step": 6521
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1396370705866471e-05,
+      "loss": 0.381,
+      "step": 6522
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1394050895988795e-05,
+      "loss": 0.3768,
+      "step": 6523
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1391731009600655e-05,
+      "loss": 0.417,
+      "step": 6524
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1389411046829382e-05,
+      "loss": 0.4179,
+      "step": 6525
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1387091007802298e-05,
+      "loss": 0.4092,
+      "step": 6526
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1384770892646736e-05,
+      "loss": 0.4313,
+      "step": 6527
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1382450701490034e-05,
+      "loss": 0.4085,
+      "step": 6528
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1380130434459531e-05,
+      "loss": 0.4172,
+      "step": 6529
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1377810091682569e-05,
+      "loss": 0.4188,
+      "step": 6530
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1375489673286501e-05,
+      "loss": 0.4185,
+      "step": 6531
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1373169179398678e-05,
+      "loss": 0.4386,
+      "step": 6532
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1370848610146456e-05,
+      "loss": 0.4326,
+      "step": 6533
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.13685279656572e-05,
+      "loss": 0.4118,
+      "step": 6534
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1366207246058269e-05,
+      "loss": 0.4282,
+      "step": 6535
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1363886451477036e-05,
+      "loss": 0.4101,
+      "step": 6536
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1361565582040876e-05,
+      "loss": 0.413,
+      "step": 6537
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1359244637877163e-05,
+      "loss": 0.4144,
+      "step": 6538
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.135692361911328e-05,
+      "loss": 0.427,
+      "step": 6539
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1354602525876612e-05,
+      "loss": 0.4286,
+      "step": 6540
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1352281358294552e-05,
+      "loss": 0.4183,
+      "step": 6541
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.134996011649449e-05,
+      "loss": 0.4251,
+      "step": 6542
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.134763880060383e-05,
+      "loss": 0.4415,
+      "step": 6543
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1345317410749965e-05,
+      "loss": 0.4246,
+      "step": 6544
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1342995947060304e-05,
+      "loss": 0.3969,
+      "step": 6545
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1340674409662262e-05,
+      "loss": 0.4125,
+      "step": 6546
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1338352798683251e-05,
+      "loss": 0.4288,
+      "step": 6547
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.133603111425069e-05,
+      "loss": 0.4283,
+      "step": 6548
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1333709356491996e-05,
+      "loss": 0.4452,
+      "step": 6549
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1331387525534598e-05,
+      "loss": 0.4458,
+      "step": 6550
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.132906562150593e-05,
+      "loss": 0.4549,
+      "step": 6551
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1326743644533425e-05,
+      "loss": 0.3921,
+      "step": 6552
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1324421594744516e-05,
+      "loss": 0.4117,
+      "step": 6553
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1322099472266652e-05,
+      "loss": 0.4245,
+      "step": 6554
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1319777277227277e-05,
+      "loss": 0.434,
+      "step": 6555
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1317455009753843e-05,
+      "loss": 0.4115,
+      "step": 6556
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1315132669973798e-05,
+      "loss": 0.4014,
+      "step": 6557
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1312810258014606e-05,
+      "loss": 0.4186,
+      "step": 6558
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1310487774003729e-05,
+      "loss": 0.4486,
+      "step": 6559
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1308165218068631e-05,
+      "loss": 0.3939,
+      "step": 6560
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1305842590336784e-05,
+      "loss": 0.4215,
+      "step": 6561
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1303519890935657e-05,
+      "loss": 0.4052,
+      "step": 6562
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1301197119992737e-05,
+      "loss": 0.4298,
+      "step": 6563
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1298874277635501e-05,
+      "loss": 0.39,
+      "step": 6564
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1296551363991432e-05,
+      "loss": 0.3972,
+      "step": 6565
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1294228379188022e-05,
+      "loss": 0.4024,
+      "step": 6566
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1291905323352767e-05,
+      "loss": 0.4271,
+      "step": 6567
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1289582196613163e-05,
+      "loss": 0.428,
+      "step": 6568
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1287258999096713e-05,
+      "loss": 0.4298,
+      "step": 6569
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1284935730930917e-05,
+      "loss": 0.4401,
+      "step": 6570
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1282612392243286e-05,
+      "loss": 0.3926,
+      "step": 6571
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.128028898316134e-05,
+      "loss": 0.4413,
+      "step": 6572
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1277965503812589e-05,
+      "loss": 0.4455,
+      "step": 6573
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1275641954324554e-05,
+      "loss": 0.445,
+      "step": 6574
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1273318334824762e-05,
+      "loss": 0.4488,
+      "step": 6575
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.127099464544074e-05,
+      "loss": 0.4205,
+      "step": 6576
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1268670886300023e-05,
+      "loss": 0.4258,
+      "step": 6577
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1266347057530145e-05,
+      "loss": 0.4679,
+      "step": 6578
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1264023159258644e-05,
+      "loss": 0.4189,
+      "step": 6579
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1261699191613067e-05,
+      "loss": 0.4475,
+      "step": 6580
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1259375154720962e-05,
+      "loss": 0.4245,
+      "step": 6581
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1257051048709877e-05,
+      "loss": 0.4431,
+      "step": 6582
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1254726873707367e-05,
+      "loss": 0.4242,
+      "step": 6583
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1252402629840995e-05,
+      "loss": 0.4309,
+      "step": 6584
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1250078317238322e-05,
+      "loss": 0.4367,
+      "step": 6585
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1247753936026917e-05,
+      "loss": 0.4067,
+      "step": 6586
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1245429486334343e-05,
+      "loss": 0.4242,
+      "step": 6587
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1243104968288179e-05,
+      "loss": 0.403,
+      "step": 6588
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1240780382016006e-05,
+      "loss": 0.4246,
+      "step": 6589
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1238455727645401e-05,
+      "loss": 0.4514,
+      "step": 6590
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1236131005303948e-05,
+      "loss": 0.4219,
+      "step": 6591
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1233806215119239e-05,
+      "loss": 0.4134,
+      "step": 6592
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1231481357218868e-05,
+      "loss": 0.4386,
+      "step": 6593
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1229156431730429e-05,
+      "loss": 0.4331,
+      "step": 6594
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1226831438781517e-05,
+      "loss": 0.4245,
+      "step": 6595
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1224506378499746e-05,
+      "loss": 0.4566,
+      "step": 6596
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.122218125101272e-05,
+      "loss": 0.4085,
+      "step": 6597
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1219856056448051e-05,
+      "loss": 0.4349,
+      "step": 6598
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.121753079493335e-05,
+      "loss": 0.4321,
+      "step": 6599
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1215205466596235e-05,
+      "loss": 0.4241,
+      "step": 6600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1212880071564334e-05,
+      "loss": 0.4193,
+      "step": 6601
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.121055460996527e-05,
+      "loss": 0.3995,
+      "step": 6602
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1208229081926671e-05,
+      "loss": 0.4115,
+      "step": 6603
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1205903487576173e-05,
+      "loss": 0.4098,
+      "step": 6604
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1203577827041412e-05,
+      "loss": 0.397,
+      "step": 6605
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1201252100450031e-05,
+      "loss": 0.4286,
+      "step": 6606
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1198926307929665e-05,
+      "loss": 0.4219,
+      "step": 6607
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1196600449607972e-05,
+      "loss": 0.4117,
+      "step": 6608
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.11942745256126e-05,
+      "loss": 0.4127,
+      "step": 6609
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1191948536071203e-05,
+      "loss": 0.4331,
+      "step": 6610
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.118962248111144e-05,
+      "loss": 0.4149,
+      "step": 6611
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1187296360860973e-05,
+      "loss": 0.3968,
+      "step": 6612
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1184970175447464e-05,
+      "loss": 0.4246,
+      "step": 6613
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1182643924998593e-05,
+      "loss": 0.4022,
+      "step": 6614
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1180317609642021e-05,
+      "loss": 0.4021,
+      "step": 6615
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1177991229505431e-05,
+      "loss": 0.4219,
+      "step": 6616
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1175664784716502e-05,
+      "loss": 0.4206,
+      "step": 6617
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1173338275402918e-05,
+      "loss": 0.4089,
+      "step": 6618
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1171011701692363e-05,
+      "loss": 0.416,
+      "step": 6619
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1168685063712532e-05,
+      "loss": 0.4202,
+      "step": 6620
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1166358361591113e-05,
+      "loss": 0.4052,
+      "step": 6621
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1164031595455814e-05,
+      "loss": 0.4265,
+      "step": 6622
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1161704765434323e-05,
+      "loss": 0.4201,
+      "step": 6623
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1159377871654356e-05,
+      "loss": 0.3852,
+      "step": 6624
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1157050914243614e-05,
+      "loss": 0.4052,
+      "step": 6625
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1154723893329813e-05,
+      "loss": 0.4272,
+      "step": 6626
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1152396809040664e-05,
+      "loss": 0.4156,
+      "step": 6627
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1150069661503892e-05,
+      "loss": 0.3863,
+      "step": 6628
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1147742450847215e-05,
+      "loss": 0.4375,
+      "step": 6629
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1145415177198355e-05,
+      "loss": 0.3823,
+      "step": 6630
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1143087840685048e-05,
+      "loss": 0.4174,
+      "step": 6631
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1140760441435023e-05,
+      "loss": 0.4258,
+      "step": 6632
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1138432979576017e-05,
+      "loss": 0.4317,
+      "step": 6633
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1136105455235766e-05,
+      "loss": 0.4495,
+      "step": 6634
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1133777868542014e-05,
+      "loss": 0.4471,
+      "step": 6635
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1131450219622514e-05,
+      "loss": 0.388,
+      "step": 6636
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1129122508605011e-05,
+      "loss": 0.4387,
+      "step": 6637
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1126794735617252e-05,
+      "loss": 0.4314,
+      "step": 6638
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1124466900787003e-05,
+      "loss": 0.4266,
+      "step": 6639
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1122139004242017e-05,
+      "loss": 0.3888,
+      "step": 6640
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1119811046110061e-05,
+      "loss": 0.395,
+      "step": 6641
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1117483026518897e-05,
+      "loss": 0.4506,
+      "step": 6642
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1115154945596306e-05,
+      "loss": 0.4261,
+      "step": 6643
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1112826803470047e-05,
+      "loss": 0.4042,
+      "step": 6644
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1110498600267907e-05,
+      "loss": 0.4283,
+      "step": 6645
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.110817033611766e-05,
+      "loss": 0.4307,
+      "step": 6646
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1105842011147094e-05,
+      "loss": 0.4312,
+      "step": 6647
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.110351362548399e-05,
+      "loss": 0.4409,
+      "step": 6648
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1101185179256146e-05,
+      "loss": 0.4091,
+      "step": 6649
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1098856672591351e-05,
+      "loss": 0.417,
+      "step": 6650
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1096528105617402e-05,
+      "loss": 0.4369,
+      "step": 6651
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1094199478462095e-05,
+      "loss": 0.4229,
+      "step": 6652
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.109187079125324e-05,
+      "loss": 0.4386,
+      "step": 6653
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1089542044118642e-05,
+      "loss": 0.4357,
+      "step": 6654
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1087213237186107e-05,
+      "loss": 0.444,
+      "step": 6655
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1084884370583453e-05,
+      "loss": 0.4352,
+      "step": 6656
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1082555444438494e-05,
+      "loss": 0.4146,
+      "step": 6657
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1080226458879051e-05,
+      "loss": 0.4258,
+      "step": 6658
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1077897414032943e-05,
+      "loss": 0.3888,
+      "step": 6659
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1075568310028e-05,
+      "loss": 0.4277,
+      "step": 6660
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1073239146992054e-05,
+      "loss": 0.4007,
+      "step": 6661
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1070909925052932e-05,
+      "loss": 0.4228,
+      "step": 6662
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1068580644338474e-05,
+      "loss": 0.425,
+      "step": 6663
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1066251304976517e-05,
+      "loss": 0.4157,
+      "step": 6664
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1063921907094903e-05,
+      "loss": 0.4202,
+      "step": 6665
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1061592450821481e-05,
+      "loss": 0.417,
+      "step": 6666
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.10592629362841e-05,
+      "loss": 0.4172,
+      "step": 6667
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.10569333636106e-05,
+      "loss": 0.4048,
+      "step": 6668
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1054603732928855e-05,
+      "loss": 0.4261,
+      "step": 6669
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1052274044366711e-05,
+      "loss": 0.3962,
+      "step": 6670
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1049944298052033e-05,
+      "loss": 0.4104,
+      "step": 6671
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1047614494112686e-05,
+      "loss": 0.4046,
+      "step": 6672
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 0.4321,
+      "step": 6673
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1042954713871457e-05,
+      "loss": 0.4274,
+      "step": 6674
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1040624737825323e-05,
+      "loss": 0.4334,
+      "step": 6675
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1038294704666007e-05,
+      "loss": 0.4328,
+      "step": 6676
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1035964614521392e-05,
+      "loss": 0.4306,
+      "step": 6677
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1033634467519364e-05,
+      "loss": 0.4328,
+      "step": 6678
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1031304263787811e-05,
+      "loss": 0.4264,
+      "step": 6679
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1028974003454615e-05,
+      "loss": 0.4506,
+      "step": 6680
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1026643686647675e-05,
+      "loss": 0.4297,
+      "step": 6681
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1024313313494887e-05,
+      "loss": 0.3868,
+      "step": 6682
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1021982884124152e-05,
+      "loss": 0.4032,
+      "step": 6683
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1019652398663362e-05,
+      "loss": 0.4123,
+      "step": 6684
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1017321857240432e-05,
+      "loss": 0.4429,
+      "step": 6685
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1014991259983272e-05,
+      "loss": 0.4095,
+      "step": 6686
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1012660607019788e-05,
+      "loss": 0.4281,
+      "step": 6687
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1010329898477891e-05,
+      "loss": 0.3968,
+      "step": 6688
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1007999134485507e-05,
+      "loss": 0.3996,
+      "step": 6689
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1005668315170554e-05,
+      "loss": 0.3872,
+      "step": 6690
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1003337440660956e-05,
+      "loss": 0.3982,
+      "step": 6691
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1001006511084634e-05,
+      "loss": 0.4613,
+      "step": 6692
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0998675526569524e-05,
+      "loss": 0.3976,
+      "step": 6693
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0996344487243555e-05,
+      "loss": 0.4197,
+      "step": 6694
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0994013393234665e-05,
+      "loss": 0.4142,
+      "step": 6695
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0991682244670794e-05,
+      "loss": 0.4178,
+      "step": 6696
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.098935104167988e-05,
+      "loss": 0.4481,
+      "step": 6697
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0987019784389872e-05,
+      "loss": 0.4198,
+      "step": 6698
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0984688472928712e-05,
+      "loss": 0.435,
+      "step": 6699
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0982357107424355e-05,
+      "loss": 0.3994,
+      "step": 6700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0980025688004756e-05,
+      "loss": 0.4244,
+      "step": 6701
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0977694214797866e-05,
+      "loss": 0.4178,
+      "step": 6702
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.097536268793165e-05,
+      "loss": 0.4383,
+      "step": 6703
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0973031107534065e-05,
+      "loss": 0.4269,
+      "step": 6704
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0970699473733082e-05,
+      "loss": 0.4362,
+      "step": 6705
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0968367786656664e-05,
+      "loss": 0.4133,
+      "step": 6706
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0966036046432786e-05,
+      "loss": 0.4227,
+      "step": 6707
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0963704253189424e-05,
+      "loss": 0.4004,
+      "step": 6708
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0961372407054549e-05,
+      "loss": 0.4248,
+      "step": 6709
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0959040508156145e-05,
+      "loss": 0.4368,
+      "step": 6710
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0956708556622195e-05,
+      "loss": 0.426,
+      "step": 6711
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0954376552580685e-05,
+      "loss": 0.42,
+      "step": 6712
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.09520444961596e-05,
+      "loss": 0.424,
+      "step": 6713
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0949712387486937e-05,
+      "loss": 0.4139,
+      "step": 6714
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0947380226690686e-05,
+      "loss": 0.4461,
+      "step": 6715
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0945048013898844e-05,
+      "loss": 0.3839,
+      "step": 6716
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0942715749239414e-05,
+      "loss": 0.429,
+      "step": 6717
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.09403834328404e-05,
+      "loss": 0.4081,
+      "step": 6718
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0938051064829805e-05,
+      "loss": 0.4275,
+      "step": 6719
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0935718645335638e-05,
+      "loss": 0.4185,
+      "step": 6720
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.093338617448591e-05,
+      "loss": 0.4132,
+      "step": 6721
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0931053652408636e-05,
+      "loss": 0.4493,
+      "step": 6722
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0928721079231833e-05,
+      "loss": 0.4273,
+      "step": 6723
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0926388455083523e-05,
+      "loss": 0.4188,
+      "step": 6724
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0924055780091725e-05,
+      "loss": 0.4197,
+      "step": 6725
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0921723054384466e-05,
+      "loss": 0.432,
+      "step": 6726
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0919390278089775e-05,
+      "loss": 0.4233,
+      "step": 6727
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0917057451335682e-05,
+      "loss": 0.4422,
+      "step": 6728
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0914724574250221e-05,
+      "loss": 0.4077,
+      "step": 6729
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0912391646961432e-05,
+      "loss": 0.4013,
+      "step": 6730
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.091005866959735e-05,
+      "loss": 0.4411,
+      "step": 6731
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0907725642286016e-05,
+      "loss": 0.4378,
+      "step": 6732
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0905392565155477e-05,
+      "loss": 0.4173,
+      "step": 6733
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0903059438333782e-05,
+      "loss": 0.4158,
+      "step": 6734
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0900726261948984e-05,
+      "loss": 0.4083,
+      "step": 6735
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0898393036129126e-05,
+      "loss": 0.3978,
+      "step": 6736
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.089605976100227e-05,
+      "loss": 0.379,
+      "step": 6737
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0893726436696479e-05,
+      "loss": 0.4099,
+      "step": 6738
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0891393063339805e-05,
+      "loss": 0.4109,
+      "step": 6739
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0889059641060318e-05,
+      "loss": 0.4437,
+      "step": 6740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.088672616998608e-05,
+      "loss": 0.3983,
+      "step": 6741
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0884392650245166e-05,
+      "loss": 0.4239,
+      "step": 6742
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0882059081965643e-05,
+      "loss": 0.4222,
+      "step": 6743
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0879725465275588e-05,
+      "loss": 0.4116,
+      "step": 6744
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0877391800303074e-05,
+      "loss": 0.4414,
+      "step": 6745
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.087505808717619e-05,
+      "loss": 0.4299,
+      "step": 6746
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0872724326023008e-05,
+      "loss": 0.4179,
+      "step": 6747
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0870390516971619e-05,
+      "loss": 0.415,
+      "step": 6748
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0868056660150106e-05,
+      "loss": 0.4206,
+      "step": 6749
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0865722755686568e-05,
+      "loss": 0.446,
+      "step": 6750
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.086338880370909e-05,
+      "loss": 0.4459,
+      "step": 6751
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0861054804345772e-05,
+      "loss": 0.4188,
+      "step": 6752
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0858720757724705e-05,
+      "loss": 0.4235,
+      "step": 6753
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0856386663974e-05,
+      "loss": 0.396,
+      "step": 6754
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0854052523221754e-05,
+      "loss": 0.428,
+      "step": 6755
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0851718335596077e-05,
+      "loss": 0.4167,
+      "step": 6756
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0849384101225072e-05,
+      "loss": 0.3887,
+      "step": 6757
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0847049820236852e-05,
+      "loss": 0.3979,
+      "step": 6758
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0844715492759534e-05,
+      "loss": 0.4502,
+      "step": 6759
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0842381118921233e-05,
+      "loss": 0.3703,
+      "step": 6760
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0840046698850065e-05,
+      "loss": 0.4081,
+      "step": 6761
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.083771223267415e-05,
+      "loss": 0.4347,
+      "step": 6762
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0835377720521616e-05,
+      "loss": 0.413,
+      "step": 6763
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0833043162520591e-05,
+      "loss": 0.3429,
+      "step": 6764
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0830708558799197e-05,
+      "loss": 0.4049,
+      "step": 6765
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.082837390948557e-05,
+      "loss": 0.4563,
+      "step": 6766
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0826039214707844e-05,
+      "loss": 0.3872,
+      "step": 6767
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0823704474594155e-05,
+      "loss": 0.4248,
+      "step": 6768
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0821369689272639e-05,
+      "loss": 0.4038,
+      "step": 6769
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0819034858871438e-05,
+      "loss": 0.4225,
+      "step": 6770
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.08166999835187e-05,
+      "loss": 0.4037,
+      "step": 6771
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0814365063342567e-05,
+      "loss": 0.4125,
+      "step": 6772
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.081203009847119e-05,
+      "loss": 0.4221,
+      "step": 6773
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0809695089032716e-05,
+      "loss": 0.4244,
+      "step": 6774
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0807360035155306e-05,
+      "loss": 0.4225,
+      "step": 6775
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0805024936967112e-05,
+      "loss": 0.4315,
+      "step": 6776
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.080268979459629e-05,
+      "loss": 0.4422,
+      "step": 6777
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0800354608171005e-05,
+      "loss": 0.4376,
+      "step": 6778
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0798019377819415e-05,
+      "loss": 0.4228,
+      "step": 6779
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0795684103669692e-05,
+      "loss": 0.417,
+      "step": 6780
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0793348785850003e-05,
+      "loss": 0.4006,
+      "step": 6781
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0791013424488514e-05,
+      "loss": 0.437,
+      "step": 6782
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0788678019713401e-05,
+      "loss": 0.4201,
+      "step": 6783
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0786342571652841e-05,
+      "loss": 0.4308,
+      "step": 6784
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.078400708043501e-05,
+      "loss": 0.3998,
+      "step": 6785
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0781671546188086e-05,
+      "loss": 0.4148,
+      "step": 6786
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0779335969040252e-05,
+      "loss": 0.4358,
+      "step": 6787
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0777000349119697e-05,
+      "loss": 0.425,
+      "step": 6788
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0774664686554605e-05,
+      "loss": 0.4296,
+      "step": 6789
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0772328981473166e-05,
+      "loss": 0.4198,
+      "step": 6790
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0769993234003568e-05,
+      "loss": 0.4402,
+      "step": 6791
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0767657444274013e-05,
+      "loss": 0.4359,
+      "step": 6792
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.076532161241269e-05,
+      "loss": 0.4541,
+      "step": 6793
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0762985738547802e-05,
+      "loss": 0.4187,
+      "step": 6794
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0760649822807552e-05,
+      "loss": 0.4186,
+      "step": 6795
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0758313865320133e-05,
+      "loss": 0.4568,
+      "step": 6796
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0755977866213763e-05,
+      "loss": 0.3922,
+      "step": 6797
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0753641825616643e-05,
+      "loss": 0.4187,
+      "step": 6798
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0751305743656986e-05,
+      "loss": 0.407,
+      "step": 6799
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0748969620463001e-05,
+      "loss": 0.4211,
+      "step": 6800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0746633456162907e-05,
+      "loss": 0.4257,
+      "step": 6801
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0744297250884916e-05,
+      "loss": 0.4184,
+      "step": 6802
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0741961004757255e-05,
+      "loss": 0.3828,
+      "step": 6803
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0739624717908136e-05,
+      "loss": 0.406,
+      "step": 6804
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0737288390465792e-05,
+      "loss": 0.4071,
+      "step": 6805
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.073495202255844e-05,
+      "loss": 0.4055,
+      "step": 6806
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0732615614314317e-05,
+      "loss": 0.4127,
+      "step": 6807
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0730279165861645e-05,
+      "loss": 0.4067,
+      "step": 6808
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0727942677328664e-05,
+      "loss": 0.4014,
+      "step": 6809
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0725606148843603e-05,
+      "loss": 0.4421,
+      "step": 6810
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0723269580534703e-05,
+      "loss": 0.4558,
+      "step": 6811
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0720932972530198e-05,
+      "loss": 0.4103,
+      "step": 6812
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0718596324958337e-05,
+      "loss": 0.4587,
+      "step": 6813
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0716259637947357e-05,
+      "loss": 0.3873,
+      "step": 6814
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0713922911625507e-05,
+      "loss": 0.4761,
+      "step": 6815
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0711586146121036e-05,
+      "loss": 0.3909,
+      "step": 6816
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0709249341562187e-05,
+      "loss": 0.4156,
+      "step": 6817
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0706912498077218e-05,
+      "loss": 0.4192,
+      "step": 6818
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0704575615794383e-05,
+      "loss": 0.4269,
+      "step": 6819
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.070223869484194e-05,
+      "loss": 0.4199,
+      "step": 6820
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0699901735348141e-05,
+      "loss": 0.4062,
+      "step": 6821
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0697564737441254e-05,
+      "loss": 0.4111,
+      "step": 6822
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0695227701249537e-05,
+      "loss": 0.4428,
+      "step": 6823
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0692890626901258e-05,
+      "loss": 0.3897,
+      "step": 6824
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.069055351452468e-05,
+      "loss": 0.4256,
+      "step": 6825
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0688216364248075e-05,
+      "loss": 0.4432,
+      "step": 6826
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0685879176199714e-05,
+      "loss": 0.4119,
+      "step": 6827
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.068354195050787e-05,
+      "loss": 0.3847,
+      "step": 6828
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0681204687300815e-05,
+      "loss": 0.442,
+      "step": 6829
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.067886738670683e-05,
+      "loss": 0.4097,
+      "step": 6830
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0676530048854196e-05,
+      "loss": 0.3973,
+      "step": 6831
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0674192673871191e-05,
+      "loss": 0.3982,
+      "step": 6832
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0671855261886098e-05,
+      "loss": 0.4332,
+      "step": 6833
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0669517813027204e-05,
+      "loss": 0.4487,
+      "step": 6834
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0667180327422797e-05,
+      "loss": 0.4434,
+      "step": 6835
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0664842805201166e-05,
+      "loss": 0.4372,
+      "step": 6836
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0662505246490601e-05,
+      "loss": 0.4065,
+      "step": 6837
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0660167651419395e-05,
+      "loss": 0.4105,
+      "step": 6838
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0657830020115849e-05,
+      "loss": 0.4475,
+      "step": 6839
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0655492352708256e-05,
+      "loss": 0.4411,
+      "step": 6840
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0653154649324918e-05,
+      "loss": 0.425,
+      "step": 6841
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.065081691009413e-05,
+      "loss": 0.4354,
+      "step": 6842
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.06484791351442e-05,
+      "loss": 0.3815,
+      "step": 6843
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0646141324603439e-05,
+      "loss": 0.4123,
+      "step": 6844
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0643803478600147e-05,
+      "loss": 0.4625,
+      "step": 6845
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0641465597262632e-05,
+      "loss": 0.4035,
+      "step": 6846
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.063912768071921e-05,
+      "loss": 0.4701,
+      "step": 6847
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0636789729098193e-05,
+      "loss": 0.4277,
+      "step": 6848
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0634451742527898e-05,
+      "loss": 0.397,
+      "step": 6849
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0632113721136636e-05,
+      "loss": 0.4216,
+      "step": 6850
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.062977566505273e-05,
+      "loss": 0.4154,
+      "step": 6851
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0627437574404503e-05,
+      "loss": 0.4096,
+      "step": 6852
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0625099449320275e-05,
+      "loss": 0.4218,
+      "step": 6853
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0622761289928369e-05,
+      "loss": 0.4328,
+      "step": 6854
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0620423096357112e-05,
+      "loss": 0.4039,
+      "step": 6855
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0618084868734835e-05,
+      "loss": 0.4504,
+      "step": 6856
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0615746607189868e-05,
+      "loss": 0.4249,
+      "step": 6857
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0613408311850539e-05,
+      "loss": 0.4189,
+      "step": 6858
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0611069982845183e-05,
+      "loss": 0.4124,
+      "step": 6859
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0608731620302141e-05,
+      "loss": 0.4184,
+      "step": 6860
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0606393224349749e-05,
+      "loss": 0.4107,
+      "step": 6861
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0604054795116344e-05,
+      "loss": 0.4365,
+      "step": 6862
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0601716332730265e-05,
+      "loss": 0.4542,
+      "step": 6863
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.059937783731986e-05,
+      "loss": 0.3874,
+      "step": 6864
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0597039309013473e-05,
+      "loss": 0.4446,
+      "step": 6865
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.059470074793945e-05,
+      "loss": 0.4306,
+      "step": 6866
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0592362154226138e-05,
+      "loss": 0.4427,
+      "step": 6867
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0590023528001886e-05,
+      "loss": 0.4342,
+      "step": 6868
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0587684869395052e-05,
+      "loss": 0.4119,
+      "step": 6869
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0585346178533985e-05,
+      "loss": 0.412,
+      "step": 6870
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0583007455547042e-05,
+      "loss": 0.4278,
+      "step": 6871
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0580668700562581e-05,
+      "loss": 0.4329,
+      "step": 6872
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0578329913708964e-05,
+      "loss": 0.4021,
+      "step": 6873
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0575991095114548e-05,
+      "loss": 0.4005,
+      "step": 6874
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0573652244907693e-05,
+      "loss": 0.4083,
+      "step": 6875
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0571313363216768e-05,
+      "loss": 0.4584,
+      "step": 6876
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.056897445017014e-05,
+      "loss": 0.4513,
+      "step": 6877
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0566635505896175e-05,
+      "loss": 0.3923,
+      "step": 6878
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0564296530523239e-05,
+      "loss": 0.399,
+      "step": 6879
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.056195752417971e-05,
+      "loss": 0.4184,
+      "step": 6880
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0559618486993955e-05,
+      "loss": 0.3918,
+      "step": 6881
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0557279419094352e-05,
+      "loss": 0.4141,
+      "step": 6882
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0554940320609278e-05,
+      "loss": 0.4428,
+      "step": 6883
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0552601191667109e-05,
+      "loss": 0.44,
+      "step": 6884
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0550262032396224e-05,
+      "loss": 0.4127,
+      "step": 6885
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.054792284292501e-05,
+      "loss": 0.4182,
+      "step": 6886
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.054558362338184e-05,
+      "loss": 0.4291,
+      "step": 6887
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0543244373895108e-05,
+      "loss": 0.4252,
+      "step": 6888
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0540905094593195e-05,
+      "loss": 0.4153,
+      "step": 6889
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0538565785604494e-05,
+      "loss": 0.451,
+      "step": 6890
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0536226447057385e-05,
+      "loss": 0.3845,
+      "step": 6891
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0533887079080272e-05,
+      "loss": 0.4541,
+      "step": 6892
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0531547681801537e-05,
+      "loss": 0.4213,
+      "step": 6893
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.052920825534958e-05,
+      "loss": 0.3928,
+      "step": 6894
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0526868799852797e-05,
+      "loss": 0.4209,
+      "step": 6895
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0524529315439582e-05,
+      "loss": 0.4423,
+      "step": 6896
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0522189802238339e-05,
+      "loss": 0.4303,
+      "step": 6897
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0519850260377464e-05,
+      "loss": 0.4372,
+      "step": 6898
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0517510689985364e-05,
+      "loss": 0.428,
+      "step": 6899
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0515171091190438e-05,
+      "loss": 0.3998,
+      "step": 6900
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0512831464121099e-05,
+      "loss": 0.4167,
+      "step": 6901
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0510491808905745e-05,
+      "loss": 0.4065,
+      "step": 6902
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0508152125672788e-05,
+      "loss": 0.4579,
+      "step": 6903
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.050581241455064e-05,
+      "loss": 0.4066,
+      "step": 6904
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0503472675667715e-05,
+      "loss": 0.4471,
+      "step": 6905
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0501132909152418e-05,
+      "loss": 0.4556,
+      "step": 6906
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0498793115133168e-05,
+      "loss": 0.4175,
+      "step": 6907
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0496453293738385e-05,
+      "loss": 0.4284,
+      "step": 6908
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0494113445096483e-05,
+      "loss": 0.419,
+      "step": 6909
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0491773569335878e-05,
+      "loss": 0.4073,
+      "step": 6910
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0489433666584996e-05,
+      "loss": 0.4286,
+      "step": 6911
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.048709373697226e-05,
+      "loss": 0.4588,
+      "step": 6912
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0484753780626089e-05,
+      "loss": 0.4181,
+      "step": 6913
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0482413797674908e-05,
+      "loss": 0.4258,
+      "step": 6914
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0480073788247149e-05,
+      "loss": 0.4481,
+      "step": 6915
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0477733752471236e-05,
+      "loss": 0.4153,
+      "step": 6916
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0475393690475599e-05,
+      "loss": 0.4361,
+      "step": 6917
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0473053602388668e-05,
+      "loss": 0.3985,
+      "step": 6918
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0470713488338877e-05,
+      "loss": 0.4237,
+      "step": 6919
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0468373348454658e-05,
+      "loss": 0.3797,
+      "step": 6920
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0466033182864448e-05,
+      "loss": 0.3978,
+      "step": 6921
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0463692991696685e-05,
+      "loss": 0.4169,
+      "step": 6922
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0461352775079802e-05,
+      "loss": 0.4007,
+      "step": 6923
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0459012533142241e-05,
+      "loss": 0.429,
+      "step": 6924
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0456672266012446e-05,
+      "loss": 0.4052,
+      "step": 6925
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0454331973818857e-05,
+      "loss": 0.4327,
+      "step": 6926
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0451991656689912e-05,
+      "loss": 0.4037,
+      "step": 6927
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0449651314754062e-05,
+      "loss": 0.4509,
+      "step": 6928
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0447310948139755e-05,
+      "loss": 0.4127,
+      "step": 6929
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0444970556975434e-05,
+      "loss": 0.4305,
+      "step": 6930
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.044263014138955e-05,
+      "loss": 0.3914,
+      "step": 6931
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0440289701510551e-05,
+      "loss": 0.4019,
+      "step": 6932
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.043794923746689e-05,
+      "loss": 0.415,
+      "step": 6933
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0435608749387025e-05,
+      "loss": 0.3825,
+      "step": 6934
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0433268237399403e-05,
+      "loss": 0.3872,
+      "step": 6935
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.043092770163248e-05,
+      "loss": 0.4368,
+      "step": 6936
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0428587142214721e-05,
+      "loss": 0.4083,
+      "step": 6937
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0426246559274575e-05,
+      "loss": 0.4382,
+      "step": 6938
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0423905952940508e-05,
+      "loss": 0.3969,
+      "step": 6939
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0421565323340971e-05,
+      "loss": 0.4031,
+      "step": 6940
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.041922467060444e-05,
+      "loss": 0.4061,
+      "step": 6941
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0416883994859369e-05,
+      "loss": 0.4404,
+      "step": 6942
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0414543296234224e-05,
+      "loss": 0.4061,
+      "step": 6943
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.041220257485747e-05,
+      "loss": 0.4405,
+      "step": 6944
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.040986183085758e-05,
+      "loss": 0.4169,
+      "step": 6945
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0407521064363016e-05,
+      "loss": 0.4327,
+      "step": 6946
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0405180275502249e-05,
+      "loss": 0.4172,
+      "step": 6947
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0402839464403752e-05,
+      "loss": 0.4654,
+      "step": 6948
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0400498631195992e-05,
+      "loss": 0.3888,
+      "step": 6949
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.039815777600745e-05,
+      "loss": 0.4439,
+      "step": 6950
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0395816898966596e-05,
+      "loss": 0.4556,
+      "step": 6951
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0393476000201901e-05,
+      "loss": 0.3837,
+      "step": 6952
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.039113507984185e-05,
+      "loss": 0.4075,
+      "step": 6953
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0388794138014916e-05,
+      "loss": 0.419,
+      "step": 6954
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0386453174849585e-05,
+      "loss": 0.4447,
+      "step": 6955
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0384112190474326e-05,
+      "loss": 0.427,
+      "step": 6956
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0381771185017629e-05,
+      "loss": 0.4221,
+      "step": 6957
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0379430158607974e-05,
+      "loss": 0.4429,
+      "step": 6958
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0377089111373846e-05,
+      "loss": 0.4133,
+      "step": 6959
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0374748043443729e-05,
+      "loss": 0.4158,
+      "step": 6960
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0372406954946107e-05,
+      "loss": 0.4265,
+      "step": 6961
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0370065846009474e-05,
+      "loss": 0.4456,
+      "step": 6962
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0367724716762314e-05,
+      "loss": 0.3885,
+      "step": 6963
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0365383567333111e-05,
+      "loss": 0.414,
+      "step": 6964
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0363042397850366e-05,
+      "loss": 0.405,
+      "step": 6965
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0360701208442564e-05,
+      "loss": 0.4352,
+      "step": 6966
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0358359999238202e-05,
+      "loss": 0.416,
+      "step": 6967
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0356018770365765e-05,
+      "loss": 0.4276,
+      "step": 6968
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.035367752195376e-05,
+      "loss": 0.4442,
+      "step": 6969
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0351336254130674e-05,
+      "loss": 0.4155,
+      "step": 6970
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0348994967025012e-05,
+      "loss": 0.3941,
+      "step": 6971
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0346653660765263e-05,
+      "loss": 0.4119,
+      "step": 6972
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0344312335479933e-05,
+      "loss": 0.4203,
+      "step": 6973
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0341970991297518e-05,
+      "loss": 0.4093,
+      "step": 6974
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0339629628346526e-05,
+      "loss": 0.4133,
+      "step": 6975
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.033728824675545e-05,
+      "loss": 0.4143,
+      "step": 6976
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.03349468466528e-05,
+      "loss": 0.3934,
+      "step": 6977
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0332605428167078e-05,
+      "loss": 0.3933,
+      "step": 6978
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.033026399142679e-05,
+      "loss": 0.4041,
+      "step": 6979
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.032792253656044e-05,
+      "loss": 0.4153,
+      "step": 6980
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.032558106369654e-05,
+      "loss": 0.4216,
+      "step": 6981
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0323239572963592e-05,
+      "loss": 0.4308,
+      "step": 6982
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0320898064490113e-05,
+      "loss": 0.4396,
+      "step": 6983
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.031855653840461e-05,
+      "loss": 0.4058,
+      "step": 6984
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.031621499483559e-05,
+      "loss": 0.3926,
+      "step": 6985
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0313873433911573e-05,
+      "loss": 0.4289,
+      "step": 6986
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.031153185576106e-05,
+      "loss": 0.4306,
+      "step": 6987
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.030919026051258e-05,
+      "loss": 0.399,
+      "step": 6988
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0306848648294641e-05,
+      "loss": 0.4078,
+      "step": 6989
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0304507019235756e-05,
+      "loss": 0.3934,
+      "step": 6990
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0302165373464448e-05,
+      "loss": 0.4077,
+      "step": 6991
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0299823711109232e-05,
+      "loss": 0.4112,
+      "step": 6992
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0297482032298622e-05,
+      "loss": 0.4168,
+      "step": 6993
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0295140337161146e-05,
+      "loss": 0.4092,
+      "step": 6994
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.029279862582532e-05,
+      "loss": 0.387,
+      "step": 6995
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0290456898419667e-05,
+      "loss": 0.408,
+      "step": 6996
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0288115155072708e-05,
+      "loss": 0.4556,
+      "step": 6997
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0285773395912968e-05,
+      "loss": 0.4196,
+      "step": 6998
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0283431621068967e-05,
+      "loss": 0.4019,
+      "step": 6999
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0281089830669235e-05,
+      "loss": 0.4065,
+      "step": 7000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0278748024842294e-05,
+      "loss": 0.4122,
+      "step": 7001
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0276406203716674e-05,
+      "loss": 0.4473,
+      "step": 7002
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0274064367420897e-05,
+      "loss": 0.4446,
+      "step": 7003
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.02717225160835e-05,
+      "loss": 0.4617,
+      "step": 7004
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0269380649833003e-05,
+      "loss": 0.4353,
+      "step": 7005
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.026703876879794e-05,
+      "loss": 0.4396,
+      "step": 7006
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0264696873106845e-05,
+      "loss": 0.4412,
+      "step": 7007
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0262354962888243e-05,
+      "loss": 0.4289,
+      "step": 7008
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0260013038270669e-05,
+      "loss": 0.4257,
+      "step": 7009
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.025767109938266e-05,
+      "loss": 0.4404,
+      "step": 7010
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0255329146352746e-05,
+      "loss": 0.413,
+      "step": 7011
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.025298717930946e-05,
+      "loss": 0.4021,
+      "step": 7012
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.025064519838134e-05,
+      "loss": 0.4365,
+      "step": 7013
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0248303203696924e-05,
+      "loss": 0.3959,
+      "step": 7014
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0245961195384744e-05,
+      "loss": 0.4227,
+      "step": 7015
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0243619173573343e-05,
+      "loss": 0.384,
+      "step": 7016
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0241277138391255e-05,
+      "loss": 0.4202,
+      "step": 7017
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0238935089967022e-05,
+      "loss": 0.4295,
+      "step": 7018
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0236593028429184e-05,
+      "loss": 0.4297,
+      "step": 7019
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0234250953906282e-05,
+      "loss": 0.435,
+      "step": 7020
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0231908866526851e-05,
+      "loss": 0.4138,
+      "step": 7021
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0229566766419446e-05,
+      "loss": 0.4591,
+      "step": 7022
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.02272246537126e-05,
+      "loss": 0.437,
+      "step": 7023
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0224882528534856e-05,
+      "loss": 0.4323,
+      "step": 7024
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.022254039101476e-05,
+      "loss": 0.4565,
+      "step": 7025
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0220198241280859e-05,
+      "loss": 0.4093,
+      "step": 7026
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.02178560794617e-05,
+      "loss": 0.4122,
+      "step": 7027
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0215513905685823e-05,
+      "loss": 0.4182,
+      "step": 7028
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0213171720081778e-05,
+      "loss": 0.4065,
+      "step": 7029
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0210829522778111e-05,
+      "loss": 0.4399,
+      "step": 7030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0208487313903375e-05,
+      "loss": 0.417,
+      "step": 7031
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0206145093586116e-05,
+      "loss": 0.4157,
+      "step": 7032
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0203802861954877e-05,
+      "loss": 0.4166,
+      "step": 7033
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0201460619138217e-05,
+      "loss": 0.4224,
+      "step": 7034
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0199118365264684e-05,
+      "loss": 0.4325,
+      "step": 7035
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0196776100462829e-05,
+      "loss": 0.4502,
+      "step": 7036
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0194433824861201e-05,
+      "loss": 0.4222,
+      "step": 7037
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0192091538588353e-05,
+      "loss": 0.4738,
+      "step": 7038
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0189749241772843e-05,
+      "loss": 0.4529,
+      "step": 7039
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0187406934543221e-05,
+      "loss": 0.4379,
+      "step": 7040
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.018506461702804e-05,
+      "loss": 0.4164,
+      "step": 7041
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0182722289355855e-05,
+      "loss": 0.4169,
+      "step": 7042
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0180379951655224e-05,
+      "loss": 0.4092,
+      "step": 7043
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0178037604054702e-05,
+      "loss": 0.4332,
+      "step": 7044
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0175695246682841e-05,
+      "loss": 0.4457,
+      "step": 7045
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0173352879668204e-05,
+      "loss": 0.3982,
+      "step": 7046
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0171010503139346e-05,
+      "loss": 0.4353,
+      "step": 7047
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0168668117224825e-05,
+      "loss": 0.4139,
+      "step": 7048
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0166325722053197e-05,
+      "loss": 0.382,
+      "step": 7049
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0163983317753026e-05,
+      "loss": 0.3958,
+      "step": 7050
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0161640904452866e-05,
+      "loss": 0.4089,
+      "step": 7051
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0159298482281279e-05,
+      "loss": 0.4222,
+      "step": 7052
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0156956051366828e-05,
+      "loss": 0.4204,
+      "step": 7053
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0154613611838071e-05,
+      "loss": 0.4049,
+      "step": 7054
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0152271163823568e-05,
+      "loss": 0.4113,
+      "step": 7055
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0149928707451886e-05,
+      "loss": 0.4128,
+      "step": 7056
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0147586242851586e-05,
+      "loss": 0.4232,
+      "step": 7057
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0145243770151226e-05,
+      "loss": 0.4391,
+      "step": 7058
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0142901289479371e-05,
+      "loss": 0.4163,
+      "step": 7059
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0140558800964591e-05,
+      "loss": 0.4242,
+      "step": 7060
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0138216304735442e-05,
+      "loss": 0.4265,
+      "step": 7061
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.013587380092049e-05,
+      "loss": 0.4017,
+      "step": 7062
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0133531289648304e-05,
+      "loss": 0.3921,
+      "step": 7063
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0131188771047447e-05,
+      "loss": 0.4284,
+      "step": 7064
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0128846245246487e-05,
+      "loss": 0.4248,
+      "step": 7065
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0126503712373982e-05,
+      "loss": 0.3869,
+      "step": 7066
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0124161172558506e-05,
+      "loss": 0.4299,
+      "step": 7067
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0121818625928628e-05,
+      "loss": 0.4293,
+      "step": 7068
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0119476072612909e-05,
+      "loss": 0.4085,
+      "step": 7069
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0117133512739918e-05,
+      "loss": 0.411,
+      "step": 7070
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0114790946438226e-05,
+      "loss": 0.3949,
+      "step": 7071
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0112448373836397e-05,
+      "loss": 0.4196,
+      "step": 7072
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0110105795063006e-05,
+      "loss": 0.4233,
+      "step": 7073
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0107763210246616e-05,
+      "loss": 0.4077,
+      "step": 7074
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0105420619515798e-05,
+      "loss": 0.4357,
+      "step": 7075
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0103078022999123e-05,
+      "loss": 0.4234,
+      "step": 7076
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0100735420825159e-05,
+      "loss": 0.4131,
+      "step": 7077
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0098392813122477e-05,
+      "loss": 0.4102,
+      "step": 7078
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.009605020001965e-05,
+      "loss": 0.4296,
+      "step": 7079
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0093707581645247e-05,
+      "loss": 0.3879,
+      "step": 7080
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.009136495812784e-05,
+      "loss": 0.4195,
+      "step": 7081
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0089022329595998e-05,
+      "loss": 0.4179,
+      "step": 7082
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0086679696178296e-05,
+      "loss": 0.4093,
+      "step": 7083
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0084337058003304e-05,
+      "loss": 0.4032,
+      "step": 7084
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0081994415199596e-05,
+      "loss": 0.4338,
+      "step": 7085
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0079651767895742e-05,
+      "loss": 0.3888,
+      "step": 7086
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0077309116220319e-05,
+      "loss": 0.438,
+      "step": 7087
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0074966460301896e-05,
+      "loss": 0.4194,
+      "step": 7088
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0072623800269046e-05,
+      "loss": 0.4215,
+      "step": 7089
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0070281136250344e-05,
+      "loss": 0.4427,
+      "step": 7090
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0067938468374366e-05,
+      "loss": 0.3842,
+      "step": 7091
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0065595796769682e-05,
+      "loss": 0.4186,
+      "step": 7092
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0063253121564868e-05,
+      "loss": 0.4352,
+      "step": 7093
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0060910442888498e-05,
+      "loss": 0.3667,
+      "step": 7094
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.005856776086915e-05,
+      "loss": 0.4142,
+      "step": 7095
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0056225075635392e-05,
+      "loss": 0.4246,
+      "step": 7096
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0053882387315803e-05,
+      "loss": 0.4063,
+      "step": 7097
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0051539696038957e-05,
+      "loss": 0.4262,
+      "step": 7098
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0049197001933431e-05,
+      "loss": 0.4132,
+      "step": 7099
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0046854305127798e-05,
+      "loss": 0.4232,
+      "step": 7100
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0044511605750634e-05,
+      "loss": 0.4139,
+      "step": 7101
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0042168903930515e-05,
+      "loss": 0.4322,
+      "step": 7102
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0039826199796017e-05,
+      "loss": 0.4095,
+      "step": 7103
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0037483493475717e-05,
+      "loss": 0.4091,
+      "step": 7104
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0035140785098187e-05,
+      "loss": 0.4324,
+      "step": 7105
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0032798074792007e-05,
+      "loss": 0.4304,
+      "step": 7106
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0030455362685756e-05,
+      "loss": 0.3976,
+      "step": 7107
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0028112648908003e-05,
+      "loss": 0.4096,
+      "step": 7108
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0025769933587333e-05,
+      "loss": 0.3799,
+      "step": 7109
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0023427216852308e-05,
+      "loss": 0.42,
+      "step": 7110
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0021084498831522e-05,
+      "loss": 0.4036,
+      "step": 7111
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0018741779653543e-05,
+      "loss": 0.4356,
+      "step": 7112
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.001639905944695e-05,
+      "loss": 0.4436,
+      "step": 7113
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0014056338340312e-05,
+      "loss": 0.4612,
+      "step": 7114
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0011713616462215e-05,
+      "loss": 0.466,
+      "step": 7115
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0009370893941234e-05,
+      "loss": 0.3932,
+      "step": 7116
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0007028170905949e-05,
+      "loss": 0.4051,
+      "step": 7117
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0004685447484926e-05,
+      "loss": 0.4449,
+      "step": 7118
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.000234272380675e-05,
+      "loss": 0.3992,
+      "step": 7119
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1e-05,
+      "loss": 0.3797,
+      "step": 7120
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.997657276193253e-06,
+      "loss": 0.4375,
+      "step": 7121
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.995314552515076e-06,
+      "loss": 0.3996,
+      "step": 7122
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.992971829094058e-06,
+      "loss": 0.4486,
+      "step": 7123
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.99062910605877e-06,
+      "loss": 0.3863,
+      "step": 7124
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.988286383537786e-06,
+      "loss": 0.405,
+      "step": 7125
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.98594366165969e-06,
+      "loss": 0.4157,
+      "step": 7126
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.983600940553056e-06,
+      "loss": 0.3967,
+      "step": 7127
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.98125822034646e-06,
+      "loss": 0.4328,
+      "step": 7128
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.978915501168481e-06,
+      "loss": 0.4198,
+      "step": 7129
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.976572783147695e-06,
+      "loss": 0.3997,
+      "step": 7130
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.974230066412674e-06,
+      "loss": 0.4195,
+      "step": 7131
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.971887351092e-06,
+      "loss": 0.4064,
+      "step": 7132
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.969544637314244e-06,
+      "loss": 0.4102,
+      "step": 7133
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.967201925207992e-06,
+      "loss": 0.4346,
+      "step": 7134
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.964859214901815e-06,
+      "loss": 0.4399,
+      "step": 7135
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.962516506524285e-06,
+      "loss": 0.4364,
+      "step": 7136
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.960173800203986e-06,
+      "loss": 0.3983,
+      "step": 7137
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.957831096069489e-06,
+      "loss": 0.458,
+      "step": 7138
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.95548839424937e-06,
+      "loss": 0.388,
+      "step": 7139
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.953145694872207e-06,
+      "loss": 0.4242,
+      "step": 7140
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.95080299806657e-06,
+      "loss": 0.4212,
+      "step": 7141
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.948460303961045e-06,
+      "loss": 0.4107,
+      "step": 7142
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.946117612684199e-06,
+      "loss": 0.4382,
+      "step": 7143
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.94377492436461e-06,
+      "loss": 0.4211,
+      "step": 7144
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.941432239130854e-06,
+      "loss": 0.4032,
+      "step": 7145
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.939089557111504e-06,
+      "loss": 0.3931,
+      "step": 7146
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.936746878435137e-06,
+      "loss": 0.4263,
+      "step": 7147
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.934404203230321e-06,
+      "loss": 0.4271,
+      "step": 7148
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.932061531625636e-06,
+      "loss": 0.417,
+      "step": 7149
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.929718863749656e-06,
+      "loss": 0.4305,
+      "step": 7150
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.927376199730957e-06,
+      "loss": 0.4046,
+      "step": 7151
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.925033539698107e-06,
+      "loss": 0.4135,
+      "step": 7152
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.922690883779684e-06,
+      "loss": 0.4226,
+      "step": 7153
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.92034823210426e-06,
+      "loss": 0.4188,
+      "step": 7154
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.91800558480041e-06,
+      "loss": 0.4002,
+      "step": 7155
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.9156629419967e-06,
+      "loss": 0.3996,
+      "step": 7156
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.913320303821703e-06,
+      "loss": 0.4222,
+      "step": 7157
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.910977670404002e-06,
+      "loss": 0.4264,
+      "step": 7158
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.908635041872162e-06,
+      "loss": 0.4505,
+      "step": 7159
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.906292418354757e-06,
+      "loss": 0.4213,
+      "step": 7160
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.903949799980351e-06,
+      "loss": 0.4259,
+      "step": 7161
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.901607186877526e-06,
+      "loss": 0.4129,
+      "step": 7162
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.899264579174846e-06,
+      "loss": 0.4246,
+      "step": 7163
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.896921977000882e-06,
+      "loss": 0.4152,
+      "step": 7164
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.894579380484206e-06,
+      "loss": 0.4625,
+      "step": 7165
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.892236789753386e-06,
+      "loss": 0.4364,
+      "step": 7166
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.889894204936998e-06,
+      "loss": 0.3958,
+      "step": 7167
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.887551626163605e-06,
+      "loss": 0.4048,
+      "step": 7168
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.885209053561777e-06,
+      "loss": 0.3939,
+      "step": 7169
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.882866487260086e-06,
+      "loss": 0.4145,
+      "step": 7170
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.880523927387096e-06,
+      "loss": 0.4231,
+      "step": 7171
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.878181374071379e-06,
+      "loss": 0.4186,
+      "step": 7172
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.875838827441494e-06,
+      "loss": 0.4193,
+      "step": 7173
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.87349628762602e-06,
+      "loss": 0.4491,
+      "step": 7174
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.871153754753518e-06,
+      "loss": 0.4146,
+      "step": 7175
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.868811228952556e-06,
+      "loss": 0.4037,
+      "step": 7176
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.8664687103517e-06,
+      "loss": 0.4134,
+      "step": 7177
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.864126199079513e-06,
+      "loss": 0.4367,
+      "step": 7178
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.861783695264563e-06,
+      "loss": 0.4008,
+      "step": 7179
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.859441199035415e-06,
+      "loss": 0.3931,
+      "step": 7180
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.85709871052063e-06,
+      "loss": 0.4433,
+      "step": 7181
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.854756229848777e-06,
+      "loss": 0.4489,
+      "step": 7182
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.852413757148418e-06,
+      "loss": 0.393,
+      "step": 7183
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.850071292548115e-06,
+      "loss": 0.422,
+      "step": 7184
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.847728836176435e-06,
+      "loss": 0.4243,
+      "step": 7185
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.845386388161932e-06,
+      "loss": 0.429,
+      "step": 7186
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.843043948633177e-06,
+      "loss": 0.4237,
+      "step": 7187
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.840701517718724e-06,
+      "loss": 0.4194,
+      "step": 7188
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.838359095547136e-06,
+      "loss": 0.4222,
+      "step": 7189
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.836016682246978e-06,
+      "loss": 0.4095,
+      "step": 7190
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.833674277946803e-06,
+      "loss": 0.42,
+      "step": 7191
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.831331882775178e-06,
+      "loss": 0.4128,
+      "step": 7192
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.828989496860657e-06,
+      "loss": 0.436,
+      "step": 7193
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.8266471203318e-06,
+      "loss": 0.4098,
+      "step": 7194
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.82430475331716e-06,
+      "loss": 0.4301,
+      "step": 7195
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.821962395945303e-06,
+      "loss": 0.4298,
+      "step": 7196
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.81962004834478e-06,
+      "loss": 0.4253,
+      "step": 7197
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.817277710644146e-06,
+      "loss": 0.4295,
+      "step": 7198
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.814935382971963e-06,
+      "loss": 0.4026,
+      "step": 7199
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.81259306545678e-06,
+      "loss": 0.4171,
+      "step": 7200
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.81025075822716e-06,
+      "loss": 0.4276,
+      "step": 7201
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.80790846141165e-06,
+      "loss": 0.4213,
+      "step": 7202
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.805566175138804e-06,
+      "loss": 0.408,
+      "step": 7203
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.803223899537178e-06,
+      "loss": 0.4165,
+      "step": 7204
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.800881634735323e-06,
+      "loss": 0.3966,
+      "step": 7205
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.798539380861784e-06,
+      "loss": 0.4421,
+      "step": 7206
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.796197138045125e-06,
+      "loss": 0.4451,
+      "step": 7207
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.79385490641389e-06,
+      "loss": 0.403,
+      "step": 7208
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.791512686096629e-06,
+      "loss": 0.4296,
+      "step": 7209
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.789170477221892e-06,
+      "loss": 0.4075,
+      "step": 7210
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.786828279918225e-06,
+      "loss": 0.4235,
+      "step": 7211
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.784486094314182e-06,
+      "loss": 0.3951,
+      "step": 7212
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.782143920538306e-06,
+      "loss": 0.3953,
+      "step": 7213
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.779801758719141e-06,
+      "loss": 0.4117,
+      "step": 7214
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.77745960898524e-06,
+      "loss": 0.4352,
+      "step": 7215
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.775117471465146e-06,
+      "loss": 0.4288,
+      "step": 7216
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.772775346287405e-06,
+      "loss": 0.4123,
+      "step": 7217
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.770433233580557e-06,
+      "loss": 0.413,
+      "step": 7218
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.76809113347315e-06,
+      "loss": 0.4278,
+      "step": 7219
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.765749046093721e-06,
+      "loss": 0.4368,
+      "step": 7220
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.763406971570821e-06,
+      "loss": 0.437,
+      "step": 7221
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.761064910032978e-06,
+      "loss": 0.4048,
+      "step": 7222
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.758722861608747e-06,
+      "loss": 0.4348,
+      "step": 7223
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.75638082642666e-06,
+      "loss": 0.4131,
+      "step": 7224
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.754038804615257e-06,
+      "loss": 0.4744,
+      "step": 7225
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.751696796303081e-06,
+      "loss": 0.3701,
+      "step": 7226
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.749354801618665e-06,
+      "loss": 0.4047,
+      "step": 7227
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.747012820690544e-06,
+      "loss": 0.3982,
+      "step": 7228
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.74467085364726e-06,
+      "loss": 0.4371,
+      "step": 7229
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.74232890061734e-06,
+      "loss": 0.4602,
+      "step": 7230
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.739986961729331e-06,
+      "loss": 0.4074,
+      "step": 7231
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.737645037111759e-06,
+      "loss": 0.4078,
+      "step": 7232
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.735303126893157e-06,
+      "loss": 0.4622,
+      "step": 7233
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.732961231202061e-06,
+      "loss": 0.398,
+      "step": 7234
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.730619350167e-06,
+      "loss": 0.4361,
+      "step": 7235
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.728277483916504e-06,
+      "loss": 0.3901,
+      "step": 7236
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.725935632579105e-06,
+      "loss": 0.4089,
+      "step": 7237
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.723593796283326e-06,
+      "loss": 0.3874,
+      "step": 7238
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.721251975157706e-06,
+      "loss": 0.4334,
+      "step": 7239
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.718910169330768e-06,
+      "loss": 0.4205,
+      "step": 7240
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.716568378931036e-06,
+      "loss": 0.4189,
+      "step": 7241
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.714226604087034e-06,
+      "loss": 0.456,
+      "step": 7242
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.711884844927295e-06,
+      "loss": 0.4538,
+      "step": 7243
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.709543101580338e-06,
+      "loss": 0.413,
+      "step": 7244
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.707201374174684e-06,
+      "loss": 0.3708,
+      "step": 7245
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.704859662838855e-06,
+      "loss": 0.4071,
+      "step": 7246
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.702517967701378e-06,
+      "loss": 0.442,
+      "step": 7247
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.700176288890773e-06,
+      "loss": 0.4214,
+      "step": 7248
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.697834626535557e-06,
+      "loss": 0.424,
+      "step": 7249
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.695492980764246e-06,
+      "loss": 0.4315,
+      "step": 7250
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.693151351705362e-06,
+      "loss": 0.4218,
+      "step": 7251
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.690809739487424e-06,
+      "loss": 0.4216,
+      "step": 7252
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.688468144238943e-06,
+      "loss": 0.4254,
+      "step": 7253
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.686126566088432e-06,
+      "loss": 0.45,
+      "step": 7254
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.683785005164412e-06,
+      "loss": 0.4396,
+      "step": 7255
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.681443461595393e-06,
+      "loss": 0.4395,
+      "step": 7256
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.679101935509889e-06,
+      "loss": 0.4239,
+      "step": 7257
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.67676042703641e-06,
+      "loss": 0.4123,
+      "step": 7258
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.674418936303463e-06,
+      "loss": 0.4552,
+      "step": 7259
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.672077463439563e-06,
+      "loss": 0.3981,
+      "step": 7260
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.669736008573217e-06,
+      "loss": 0.4497,
+      "step": 7261
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.667394571832927e-06,
+      "loss": 0.4074,
+      "step": 7262
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.665053153347202e-06,
+      "loss": 0.4177,
+      "step": 7263
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.662711753244552e-06,
+      "loss": 0.4332,
+      "step": 7264
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.660370371653478e-06,
+      "loss": 0.41,
+      "step": 7265
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.658029008702483e-06,
+      "loss": 0.4391,
+      "step": 7266
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.655687664520069e-06,
+      "loss": 0.4233,
+      "step": 7267
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.653346339234739e-06,
+      "loss": 0.4188,
+      "step": 7268
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.651005032974994e-06,
+      "loss": 0.4091,
+      "step": 7269
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.648663745869327e-06,
+      "loss": 0.4092,
+      "step": 7270
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.646322478046242e-06,
+      "loss": 0.4145,
+      "step": 7271
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.643981229634234e-06,
+      "loss": 0.4151,
+      "step": 7272
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.641640000761803e-06,
+      "loss": 0.3987,
+      "step": 7273
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.639298791557438e-06,
+      "loss": 0.4381,
+      "step": 7274
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.636957602149636e-06,
+      "loss": 0.4426,
+      "step": 7275
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63461643266689e-06,
+      "loss": 0.4264,
+      "step": 7276
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.632275283237691e-06,
+      "loss": 0.4253,
+      "step": 7277
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.629934153990531e-06,
+      "loss": 0.3982,
+      "step": 7278
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.627593045053892e-06,
+      "loss": 0.3997,
+      "step": 7279
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.625251956556273e-06,
+      "loss": 0.425,
+      "step": 7280
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.622910888626156e-06,
+      "loss": 0.4038,
+      "step": 7281
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.620569841392029e-06,
+      "loss": 0.4366,
+      "step": 7282
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.618228814982376e-06,
+      "loss": 0.4111,
+      "step": 7283
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.615887809525676e-06,
+      "loss": 0.4464,
+      "step": 7284
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.613546825150421e-06,
+      "loss": 0.4773,
+      "step": 7285
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.611205861985089e-06,
+      "loss": 0.4062,
+      "step": 7286
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.608864920158152e-06,
+      "loss": 0.3965,
+      "step": 7287
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.606523999798102e-06,
+      "loss": 0.4328,
+      "step": 7288
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.60418310103341e-06,
+      "loss": 0.3848,
+      "step": 7289
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.601842223992554e-06,
+      "loss": 0.4225,
+      "step": 7290
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.599501368804011e-06,
+      "loss": 0.4077,
+      "step": 7291
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.597160535596252e-06,
+      "loss": 0.4539,
+      "step": 7292
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.594819724497755e-06,
+      "loss": 0.3982,
+      "step": 7293
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.59247893563699e-06,
+      "loss": 0.4099,
+      "step": 7294
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.590138169142422e-06,
+      "loss": 0.4184,
+      "step": 7295
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.587797425142531e-06,
+      "loss": 0.4078,
+      "step": 7296
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.585456703765777e-06,
+      "loss": 0.3915,
+      "step": 7297
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.583116005140633e-06,
+      "loss": 0.4354,
+      "step": 7298
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.580775329395563e-06,
+      "loss": 0.4258,
+      "step": 7299
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.57843467665903e-06,
+      "loss": 0.4292,
+      "step": 7300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.576094047059497e-06,
+      "loss": 0.4346,
+      "step": 7301
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.573753440725428e-06,
+      "loss": 0.4215,
+      "step": 7302
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.57141285778528e-06,
+      "loss": 0.4348,
+      "step": 7303
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.56907229836752e-06,
+      "loss": 0.4341,
+      "step": 7304
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.566731762600599e-06,
+      "loss": 0.4257,
+      "step": 7305
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.564391250612976e-06,
+      "loss": 0.4346,
+      "step": 7306
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.562050762533111e-06,
+      "loss": 0.4189,
+      "step": 7307
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.559710298489454e-06,
+      "loss": 0.4179,
+      "step": 7308
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.557369858610454e-06,
+      "loss": 0.4308,
+      "step": 7309
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.555029443024571e-06,
+      "loss": 0.3993,
+      "step": 7310
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.552689051860247e-06,
+      "loss": 0.4327,
+      "step": 7311
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.550348685245938e-06,
+      "loss": 0.4436,
+      "step": 7312
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.548008343310091e-06,
+      "loss": 0.4665,
+      "step": 7313
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.545668026181148e-06,
+      "loss": 0.398,
+      "step": 7314
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.543327733987557e-06,
+      "loss": 0.4293,
+      "step": 7315
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.540987466857762e-06,
+      "loss": 0.4301,
+      "step": 7316
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.538647224920203e-06,
+      "loss": 0.4062,
+      "step": 7317
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.53630700830332e-06,
+      "loss": 0.4446,
+      "step": 7318
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.533966817135553e-06,
+      "loss": 0.4257,
+      "step": 7319
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.531626651545344e-06,
+      "loss": 0.4129,
+      "step": 7320
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.529286511661126e-06,
+      "loss": 0.393,
+      "step": 7321
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.526946397611336e-06,
+      "loss": 0.4285,
+      "step": 7322
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.524606309524404e-06,
+      "loss": 0.4407,
+      "step": 7323
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.522266247528767e-06,
+      "loss": 0.4286,
+      "step": 7324
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.519926211752856e-06,
+      "loss": 0.4023,
+      "step": 7325
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.517586202325094e-06,
+      "loss": 0.4164,
+      "step": 7326
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.515246219373911e-06,
+      "loss": 0.4158,
+      "step": 7327
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.512906263027742e-06,
+      "loss": 0.4296,
+      "step": 7328
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.510566333415003e-06,
+      "loss": 0.4279,
+      "step": 7329
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.508226430664123e-06,
+      "loss": 0.4033,
+      "step": 7330
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.50588655490352e-06,
+      "loss": 0.4351,
+      "step": 7331
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.503546706261617e-06,
+      "loss": 0.435,
+      "step": 7332
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.501206884866834e-06,
+      "loss": 0.4167,
+      "step": 7333
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.498867090847585e-06,
+      "loss": 0.4292,
+      "step": 7334
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.49652732433229e-06,
+      "loss": 0.4165,
+      "step": 7335
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.494187585449358e-06,
+      "loss": 0.4256,
+      "step": 7336
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.491847874327213e-06,
+      "loss": 0.4109,
+      "step": 7337
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.48950819109426e-06,
+      "loss": 0.4127,
+      "step": 7338
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.487168535878905e-06,
+      "loss": 0.4335,
+      "step": 7339
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.484828908809563e-06,
+      "loss": 0.3683,
+      "step": 7340
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.482489310014639e-06,
+      "loss": 0.3839,
+      "step": 7341
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.480149739622539e-06,
+      "loss": 0.4395,
+      "step": 7342
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.477810197761666e-06,
+      "loss": 0.4072,
+      "step": 7343
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.475470684560417e-06,
+      "loss": 0.3688,
+      "step": 7344
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.473131200147205e-06,
+      "loss": 0.4317,
+      "step": 7345
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.470791744650421e-06,
+      "loss": 0.4096,
+      "step": 7346
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468452318198467e-06,
+      "loss": 0.3729,
+      "step": 7347
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.466112920919731e-06,
+      "loss": 0.4095,
+      "step": 7348
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.463773552942617e-06,
+      "loss": 0.4346,
+      "step": 7349
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.461434214395513e-06,
+      "loss": 0.4221,
+      "step": 7350
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.459094905406808e-06,
+      "loss": 0.4471,
+      "step": 7351
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.456755626104894e-06,
+      "loss": 0.3997,
+      "step": 7352
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.454416376618161e-06,
+      "loss": 0.4095,
+      "step": 7353
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.452077157074995e-06,
+      "loss": 0.3923,
+      "step": 7354
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.449737967603777e-06,
+      "loss": 0.3924,
+      "step": 7355
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.447398808332894e-06,
+      "loss": 0.4307,
+      "step": 7356
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.445059679390726e-06,
+      "loss": 0.3977,
+      "step": 7357
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.442720580905651e-06,
+      "loss": 0.4395,
+      "step": 7358
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.44038151300605e-06,
+      "loss": 0.4352,
+      "step": 7359
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.438042475820292e-06,
+      "loss": 0.4034,
+      "step": 7360
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.435703469476761e-06,
+      "loss": 0.4297,
+      "step": 7361
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.433364494103829e-06,
+      "loss": 0.4088,
+      "step": 7362
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.431025549829864e-06,
+      "loss": 0.4142,
+      "step": 7363
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.428686636783236e-06,
+      "loss": 0.4122,
+      "step": 7364
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.42634775509231e-06,
+      "loss": 0.4325,
+      "step": 7365
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.424008904885457e-06,
+      "loss": 0.4229,
+      "step": 7366
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.421670086291042e-06,
+      "loss": 0.3708,
+      "step": 7367
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.419331299437419e-06,
+      "loss": 0.416,
+      "step": 7368
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.41699254445296e-06,
+      "loss": 0.4401,
+      "step": 7369
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.414653821466016e-06,
+      "loss": 0.4078,
+      "step": 7370
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.412315130604951e-06,
+      "loss": 0.445,
+      "step": 7371
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.409976471998118e-06,
+      "loss": 0.4094,
+      "step": 7372
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.407637845773867e-06,
+      "loss": 0.4323,
+      "step": 7373
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.405299252060556e-06,
+      "loss": 0.4173,
+      "step": 7374
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.402960690986532e-06,
+      "loss": 0.3878,
+      "step": 7375
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.400622162680141e-06,
+      "loss": 0.4452,
+      "step": 7376
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.398283667269736e-06,
+      "loss": 0.381,
+      "step": 7377
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.395945204883659e-06,
+      "loss": 0.4235,
+      "step": 7378
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.393606775650254e-06,
+      "loss": 0.4166,
+      "step": 7379
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.39126837969786e-06,
+      "loss": 0.4342,
+      "step": 7380
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.388930017154819e-06,
+      "loss": 0.4314,
+      "step": 7381
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.386591688149464e-06,
+      "loss": 0.4399,
+      "step": 7382
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.38425339281014e-06,
+      "loss": 0.4164,
+      "step": 7383
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.381915131265167e-06,
+      "loss": 0.4106,
+      "step": 7384
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.37957690364289e-06,
+      "loss": 0.4252,
+      "step": 7385
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.377238710071633e-06,
+      "loss": 0.4072,
+      "step": 7386
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.374900550679728e-06,
+      "loss": 0.4012,
+      "step": 7387
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.372562425595499e-06,
+      "loss": 0.3846,
+      "step": 7388
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.370224334947273e-06,
+      "loss": 0.4428,
+      "step": 7389
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.367886278863366e-06,
+      "loss": 0.4022,
+      "step": 7390
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.365548257472107e-06,
+      "loss": 0.4154,
+      "step": 7391
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.363210270901805e-06,
+      "loss": 0.3939,
+      "step": 7392
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.360872319280791e-06,
+      "loss": 0.4235,
+      "step": 7393
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.35853440273737e-06,
+      "loss": 0.4067,
+      "step": 7394
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.356196521399856e-06,
+      "loss": 0.4198,
+      "step": 7395
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.353858675396564e-06,
+      "loss": 0.4568,
+      "step": 7396
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.351520864855801e-06,
+      "loss": 0.4291,
+      "step": 7397
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.349183089905871e-06,
+      "loss": 0.4299,
+      "step": 7398
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.346845350675087e-06,
+      "loss": 0.4081,
+      "step": 7399
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.344507647291749e-06,
+      "loss": 0.4492,
+      "step": 7400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.342169979884151e-06,
+      "loss": 0.4046,
+      "step": 7401
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.339832348580605e-06,
+      "loss": 0.4424,
+      "step": 7402
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.3374947535094e-06,
+      "loss": 0.3847,
+      "step": 7403
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.335157194798837e-06,
+      "loss": 0.4515,
+      "step": 7404
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.332819672577206e-06,
+      "loss": 0.4318,
+      "step": 7405
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.3304821869728e-06,
+      "loss": 0.4063,
+      "step": 7406
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.328144738113906e-06,
+      "loss": 0.3898,
+      "step": 7407
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.325807326128814e-06,
+      "loss": 0.4007,
+      "step": 7408
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.323469951145805e-06,
+      "loss": 0.44,
+      "step": 7409
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.32113261329317e-06,
+      "loss": 0.392,
+      "step": 7410
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.318795312699186e-06,
+      "loss": 0.4533,
+      "step": 7411
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.316458049492133e-06,
+      "loss": 0.4554,
+      "step": 7412
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.31412082380029e-06,
+      "loss": 0.444,
+      "step": 7413
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.311783635751928e-06,
+      "loss": 0.436,
+      "step": 7414
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.309446485475323e-06,
+      "loss": 0.4386,
+      "step": 7415
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.307109373098747e-06,
+      "loss": 0.4155,
+      "step": 7416
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.304772298750463e-06,
+      "loss": 0.4354,
+      "step": 7417
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.302435262558748e-06,
+      "loss": 0.4345,
+      "step": 7418
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.30009826465186e-06,
+      "loss": 0.3999,
+      "step": 7419
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.297761305158062e-06,
+      "loss": 0.409,
+      "step": 7420
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.295424384205618e-06,
+      "loss": 0.4385,
+      "step": 7421
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.293087501922785e-06,
+      "loss": 0.4329,
+      "step": 7422
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.290750658437818e-06,
+      "loss": 0.4332,
+      "step": 7423
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.28841385387897e-06,
+      "loss": 0.4276,
+      "step": 7424
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.286077088374493e-06,
+      "loss": 0.4327,
+      "step": 7425
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.283740362052644e-06,
+      "loss": 0.4119,
+      "step": 7426
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.281403675041664e-06,
+      "loss": 0.4178,
+      "step": 7427
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.279067027469804e-06,
+      "loss": 0.4165,
+      "step": 7428
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.276730419465299e-06,
+      "loss": 0.4183,
+      "step": 7429
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.2743938511564e-06,
+      "loss": 0.395,
+      "step": 7430
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.272057322671341e-06,
+      "loss": 0.4496,
+      "step": 7431
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.269720834138356e-06,
+      "loss": 0.4331,
+      "step": 7432
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.267384385685687e-06,
+      "loss": 0.4345,
+      "step": 7433
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.26504797744156e-06,
+      "loss": 0.4198,
+      "step": 7434
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.262711609534211e-06,
+      "loss": 0.4072,
+      "step": 7435
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.260375282091865e-06,
+      "loss": 0.3865,
+      "step": 7436
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.258038995242748e-06,
+      "loss": 0.4273,
+      "step": 7437
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.255702749115085e-06,
+      "loss": 0.4376,
+      "step": 7438
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.253366543837098e-06,
+      "loss": 0.3911,
+      "step": 7439
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.251030379537006e-06,
+      "loss": 0.42,
+      "step": 7440
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.248694256343017e-06,
+      "loss": 0.4219,
+      "step": 7441
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.246358174383359e-06,
+      "loss": 0.3997,
+      "step": 7442
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.244022133786238e-06,
+      "loss": 0.4114,
+      "step": 7443
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.241686134679868e-06,
+      "loss": 0.4545,
+      "step": 7444
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.239350177192452e-06,
+      "loss": 0.4053,
+      "step": 7445
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.2370142614522e-06,
+      "loss": 0.3972,
+      "step": 7446
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.234678387587313e-06,
+      "loss": 0.403,
+      "step": 7447
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.232342555725992e-06,
+      "loss": 0.4132,
+      "step": 7448
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.23000676599643e-06,
+      "loss": 0.4336,
+      "step": 7449
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.227671018526835e-06,
+      "loss": 0.401,
+      "step": 7450
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.225335313445396e-06,
+      "loss": 0.4345,
+      "step": 7451
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.222999650880305e-06,
+      "loss": 0.4376,
+      "step": 7452
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.22066403095975e-06,
+      "loss": 0.4149,
+      "step": 7453
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.218328453811918e-06,
+      "loss": 0.3958,
+      "step": 7454
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.215992919564996e-06,
+      "loss": 0.4486,
+      "step": 7455
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.213657428347165e-06,
+      "loss": 0.4062,
+      "step": 7456
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.2113219802866e-06,
+      "loss": 0.4023,
+      "step": 7457
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.208986575511489e-06,
+      "loss": 0.4365,
+      "step": 7458
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.206651214149999e-06,
+      "loss": 0.4148,
+      "step": 7459
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.20431589633031e-06,
+      "loss": 0.4452,
+      "step": 7460
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.201980622180589e-06,
+      "loss": 0.4006,
+      "step": 7461
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.199645391829e-06,
+      "loss": 0.4247,
+      "step": 7462
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.197310205403714e-06,
+      "loss": 0.4052,
+      "step": 7463
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.194975063032894e-06,
+      "loss": 0.4002,
+      "step": 7464
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.192639964844696e-06,
+      "loss": 0.4093,
+      "step": 7465
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.190304910967286e-06,
+      "loss": 0.4392,
+      "step": 7466
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.187969901528812e-06,
+      "loss": 0.403,
+      "step": 7467
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.185634936657434e-06,
+      "loss": 0.4502,
+      "step": 7468
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.183300016481303e-06,
+      "loss": 0.4217,
+      "step": 7469
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.180965141128566e-06,
+      "loss": 0.4188,
+      "step": 7470
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.178630310727366e-06,
+      "loss": 0.4125,
+      "step": 7471
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.17629552540585e-06,
+      "loss": 0.4423,
+      "step": 7472
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.173960785292161e-06,
+      "loss": 0.4213,
+      "step": 7473
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.171626090514432e-06,
+      "loss": 0.4184,
+      "step": 7474
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.169291441200806e-06,
+      "loss": 0.4201,
+      "step": 7475
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.166956837479412e-06,
+      "loss": 0.4368,
+      "step": 7476
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.164622279478386e-06,
+      "loss": 0.4469,
+      "step": 7477
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.162287767325854e-06,
+      "loss": 0.4169,
+      "step": 7478
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.15995330114994e-06,
+      "loss": 0.4151,
+      "step": 7479
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.157618881078772e-06,
+      "loss": 0.4079,
+      "step": 7480
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.155284507240471e-06,
+      "loss": 0.4362,
+      "step": 7481
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.152950179763148e-06,
+      "loss": 0.4366,
+      "step": 7482
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.150615898774932e-06,
+      "loss": 0.4324,
+      "step": 7483
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.148281664403926e-06,
+      "loss": 0.4159,
+      "step": 7484
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.145947476778248e-06,
+      "loss": 0.4424,
+      "step": 7485
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.143613336026002e-06,
+      "loss": 0.4216,
+      "step": 7486
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.141279242275298e-06,
+      "loss": 0.4303,
+      "step": 7487
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.138945195654234e-06,
+      "loss": 0.4371,
+      "step": 7488
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.136611196290915e-06,
+      "loss": 0.4105,
+      "step": 7489
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.134277244313434e-06,
+      "loss": 0.4417,
+      "step": 7490
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.131943339849895e-06,
+      "loss": 0.4251,
+      "step": 7491
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.129609483028386e-06,
+      "loss": 0.4025,
+      "step": 7492
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.127275673976996e-06,
+      "loss": 0.3713,
+      "step": 7493
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.124941912823814e-06,
+      "loss": 0.4285,
+      "step": 7494
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.122608199696928e-06,
+      "loss": 0.4163,
+      "step": 7495
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.120274534724416e-06,
+      "loss": 0.4035,
+      "step": 7496
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.11794091803436e-06,
+      "loss": 0.4527,
+      "step": 7497
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.115607349754834e-06,
+      "loss": 0.4154,
+      "step": 7498
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.11327383001392e-06,
+      "loss": 0.3981,
+      "step": 7499
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.110940358939685e-06,
+      "loss": 0.3928,
+      "step": 7500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.108606936660196e-06,
+      "loss": 0.4334,
+      "step": 7501
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.106273563303524e-06,
+      "loss": 0.4527,
+      "step": 7502
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.103940238997731e-06,
+      "loss": 0.419,
+      "step": 7503
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.101606963870877e-06,
+      "loss": 0.4252,
+      "step": 7504
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.099273738051022e-06,
+      "loss": 0.3904,
+      "step": 7505
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.096940561666216e-06,
+      "loss": 0.4044,
+      "step": 7506
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.094607434844523e-06,
+      "loss": 0.42,
+      "step": 7507
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.092274357713987e-06,
+      "loss": 0.4509,
+      "step": 7508
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.089941330402653e-06,
+      "loss": 0.3754,
+      "step": 7509
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.087608353038571e-06,
+      "loss": 0.4323,
+      "step": 7510
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.085275425749782e-06,
+      "loss": 0.4266,
+      "step": 7511
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.082942548664322e-06,
+      "loss": 0.4064,
+      "step": 7512
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.080609721910228e-06,
+      "loss": 0.396,
+      "step": 7513
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.078276945615534e-06,
+      "loss": 0.4366,
+      "step": 7514
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.075944219908277e-06,
+      "loss": 0.4559,
+      "step": 7515
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.07361154491648e-06,
+      "loss": 0.4128,
+      "step": 7516
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.07127892076817e-06,
+      "loss": 0.4046,
+      "step": 7517
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.068946347591366e-06,
+      "loss": 0.4238,
+      "step": 7518
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.066613825514095e-06,
+      "loss": 0.4089,
+      "step": 7519
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.064281354664367e-06,
+      "loss": 0.4302,
+      "step": 7520
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.0619489351702e-06,
+      "loss": 0.3992,
+      "step": 7521
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.059616567159603e-06,
+      "loss": 0.4008,
+      "step": 7522
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.057284250760586e-06,
+      "loss": 0.417,
+      "step": 7523
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.054951986101158e-06,
+      "loss": 0.4378,
+      "step": 7524
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.052619773309318e-06,
+      "loss": 0.4161,
+      "step": 7525
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.050287612513066e-06,
+      "loss": 0.412,
+      "step": 7526
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.047955503840402e-06,
+      "loss": 0.4021,
+      "step": 7527
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.04562344741932e-06,
+      "loss": 0.4084,
+      "step": 7528
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.04329144337781e-06,
+      "loss": 0.3992,
+      "step": 7529
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.040959491843857e-06,
+      "loss": 0.4221,
+      "step": 7530
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.038627592945451e-06,
+      "loss": 0.4381,
+      "step": 7531
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.036295746810579e-06,
+      "loss": 0.4274,
+      "step": 7532
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.033963953567216e-06,
+      "loss": 0.4099,
+      "step": 7533
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.03163221334334e-06,
+      "loss": 0.4291,
+      "step": 7534
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.029300526266922e-06,
+      "loss": 0.4535,
+      "step": 7535
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.026968892465939e-06,
+      "loss": 0.4081,
+      "step": 7536
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.024637312068355e-06,
+      "loss": 0.4068,
+      "step": 7537
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.022305785202138e-06,
+      "loss": 0.4023,
+      "step": 7538
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.019974311995247e-06,
+      "loss": 0.4116,
+      "step": 7539
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.017642892575645e-06,
+      "loss": 0.4229,
+      "step": 7540
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.01531152707129e-06,
+      "loss": 0.4232,
+      "step": 7541
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.012980215610131e-06,
+      "loss": 0.4383,
+      "step": 7542
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.010648958320121e-06,
+      "loss": 0.4085,
+      "step": 7543
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.00831775532921e-06,
+      "loss": 0.3999,
+      "step": 7544
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.005986606765338e-06,
+      "loss": 0.4225,
+      "step": 7545
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.00365551275645e-06,
+      "loss": 0.4102,
+      "step": 7546
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.001324473430479e-06,
+      "loss": 0.4301,
+      "step": 7547
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.998993488915368e-06,
+      "loss": 0.4165,
+      "step": 7548
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.99666255933905e-06,
+      "loss": 0.4328,
+      "step": 7549
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.994331684829449e-06,
+      "loss": 0.4306,
+      "step": 7550
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.992000865514496e-06,
+      "loss": 0.4087,
+      "step": 7551
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.98967010152211e-06,
+      "loss": 0.3791,
+      "step": 7552
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.987339392980217e-06,
+      "loss": 0.4288,
+      "step": 7553
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.985008740016733e-06,
+      "loss": 0.3984,
+      "step": 7554
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.982678142759566e-06,
+      "loss": 0.4248,
+      "step": 7555
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.98034760133664e-06,
+      "loss": 0.3808,
+      "step": 7556
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.978017115875853e-06,
+      "loss": 0.4079,
+      "step": 7557
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.975686686505114e-06,
+      "loss": 0.4181,
+      "step": 7558
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.973356313352326e-06,
+      "loss": 0.398,
+      "step": 7559
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.971025996545387e-06,
+      "loss": 0.4406,
+      "step": 7560
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.968695736212194e-06,
+      "loss": 0.435,
+      "step": 7561
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.966365532480639e-06,
+      "loss": 0.4457,
+      "step": 7562
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.964035385478607e-06,
+      "loss": 0.4052,
+      "step": 7563
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.961705295333995e-06,
+      "loss": 0.4144,
+      "step": 7564
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.95937526217468e-06,
+      "loss": 0.4244,
+      "step": 7565
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.957045286128547e-06,
+      "loss": 0.4272,
+      "step": 7566
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 0.416,
+      "step": 7567
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.952385505887317e-06,
+      "loss": 0.3755,
+      "step": 7568
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.95005570194797e-06,
+      "loss": 0.3777,
+      "step": 7569
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.947725955633294e-06,
+      "loss": 0.4581,
+      "step": 7570
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.945396267071147e-06,
+      "loss": 0.4531,
+      "step": 7571
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.943066636389398e-06,
+      "loss": 0.4416,
+      "step": 7572
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.940737063715906e-06,
+      "loss": 0.408,
+      "step": 7573
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.938407549178522e-06,
+      "loss": 0.3914,
+      "step": 7574
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.9360780929051e-06,
+      "loss": 0.4186,
+      "step": 7575
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.933748695023488e-06,
+      "loss": 0.3894,
+      "step": 7576
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.931419355661529e-06,
+      "loss": 0.4194,
+      "step": 7577
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.929090074947071e-06,
+      "loss": 0.4216,
+      "step": 7578
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.926760853007946e-06,
+      "loss": 0.3759,
+      "step": 7579
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.924431689971999e-06,
+      "loss": 0.3952,
+      "step": 7580
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.92210258596706e-06,
+      "loss": 0.3825,
+      "step": 7581
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.919773541120952e-06,
+      "loss": 0.425,
+      "step": 7582
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.917444555561508e-06,
+      "loss": 0.4005,
+      "step": 7583
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.91511562941655e-06,
+      "loss": 0.4481,
+      "step": 7584
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.912786762813895e-06,
+      "loss": 0.4221,
+      "step": 7585
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.910457955881361e-06,
+      "loss": 0.4113,
+      "step": 7586
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.90812920874676e-06,
+      "loss": 0.4233,
+      "step": 7587
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.905800521537905e-06,
+      "loss": 0.4133,
+      "step": 7588
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.903471894382603e-06,
+      "loss": 0.4406,
+      "step": 7589
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.90114332740865e-06,
+      "loss": 0.4081,
+      "step": 7590
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.898814820743855e-06,
+      "loss": 0.3739,
+      "step": 7591
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.896486374516011e-06,
+      "loss": 0.4172,
+      "step": 7592
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.894157988852911e-06,
+      "loss": 0.4201,
+      "step": 7593
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.891829663882343e-06,
+      "loss": 0.4275,
+      "step": 7594
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.889501399732095e-06,
+      "loss": 0.4241,
+      "step": 7595
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.887173196529955e-06,
+      "loss": 0.41,
+      "step": 7596
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.8848450544037e-06,
+      "loss": 0.4606,
+      "step": 7597
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.882516973481104e-06,
+      "loss": 0.3639,
+      "step": 7598
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.880188953889942e-06,
+      "loss": 0.4103,
+      "step": 7599
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.877860995757988e-06,
+      "loss": 0.4522,
+      "step": 7600
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.875533099213003e-06,
+      "loss": 0.3694,
+      "step": 7601
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.873205264382751e-06,
+      "loss": 0.4152,
+      "step": 7602
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.870877491394994e-06,
+      "loss": 0.3802,
+      "step": 7603
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.868549780377486e-06,
+      "loss": 0.3969,
+      "step": 7604
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.866222131457984e-06,
+      "loss": 0.4416,
+      "step": 7605
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.863894544764236e-06,
+      "loss": 0.4069,
+      "step": 7606
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.861567020423985e-06,
+      "loss": 0.4454,
+      "step": 7607
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.85923955856498e-06,
+      "loss": 0.4394,
+      "step": 7608
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.856912159314955e-06,
+      "loss": 0.3788,
+      "step": 7609
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.854584822801648e-06,
+      "loss": 0.4159,
+      "step": 7610
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.85225754915279e-06,
+      "loss": 0.446,
+      "step": 7611
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.849930338496108e-06,
+      "loss": 0.3917,
+      "step": 7612
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.847603190959336e-06,
+      "loss": 0.4147,
+      "step": 7613
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.84527610667019e-06,
+      "loss": 0.4132,
+      "step": 7614
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.842949085756389e-06,
+      "loss": 0.4129,
+      "step": 7615
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.840622128345647e-06,
+      "loss": 0.4301,
+      "step": 7616
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.838295234565679e-06,
+      "loss": 0.4172,
+      "step": 7617
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.835968404544193e-06,
+      "loss": 0.413,
+      "step": 7618
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.833641638408888e-06,
+      "loss": 0.4314,
+      "step": 7619
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.831314936287472e-06,
+      "loss": 0.4387,
+      "step": 7620
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.828988298307638e-06,
+      "loss": 0.4241,
+      "step": 7621
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.826661724597086e-06,
+      "loss": 0.3951,
+      "step": 7622
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.824335215283502e-06,
+      "loss": 0.3954,
+      "step": 7623
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.82200877049457e-06,
+      "loss": 0.3789,
+      "step": 7624
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.819682390357982e-06,
+      "loss": 0.4084,
+      "step": 7625
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.817356075001412e-06,
+      "loss": 0.3958,
+      "step": 7626
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.815029824552537e-06,
+      "loss": 0.4065,
+      "step": 7627
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.812703639139032e-06,
+      "loss": 0.4072,
+      "step": 7628
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.810377518888562e-06,
+      "loss": 0.3911,
+      "step": 7629
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.808051463928799e-06,
+      "loss": 0.3866,
+      "step": 7630
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.805725474387403e-06,
+      "loss": 0.4243,
+      "step": 7631
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.80339955039203e-06,
+      "loss": 0.4397,
+      "step": 7632
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.801073692070337e-06,
+      "loss": 0.467,
+      "step": 7633
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.798747899549975e-06,
+      "loss": 0.4032,
+      "step": 7634
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.796422172958591e-06,
+      "loss": 0.4274,
+      "step": 7635
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.794096512423827e-06,
+      "loss": 0.4236,
+      "step": 7636
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.791770918073328e-06,
+      "loss": 0.4083,
+      "step": 7637
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.789445390034732e-06,
+      "loss": 0.398,
+      "step": 7638
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.78711992843567e-06,
+      "loss": 0.4161,
+      "step": 7639
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.784794533403768e-06,
+      "loss": 0.4015,
+      "step": 7640
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.782469205066654e-06,
+      "loss": 0.4077,
+      "step": 7641
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.780143943551954e-06,
+      "loss": 0.442,
+      "step": 7642
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.777818748987283e-06,
+      "loss": 0.4134,
+      "step": 7643
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.775493621500254e-06,
+      "loss": 0.4081,
+      "step": 7644
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.773168561218484e-06,
+      "loss": 0.4132,
+      "step": 7645
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.770843568269576e-06,
+      "loss": 0.398,
+      "step": 7646
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.768518642781137e-06,
+      "loss": 0.4329,
+      "step": 7647
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.766193784880765e-06,
+      "loss": 0.411,
+      "step": 7648
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.763868994696055e-06,
+      "loss": 0.4609,
+      "step": 7649
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.761544272354604e-06,
+      "loss": 0.428,
+      "step": 7650
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.759219617984e-06,
+      "loss": 0.424,
+      "step": 7651
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.756895031711821e-06,
+      "loss": 0.4272,
+      "step": 7652
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.75457051366566e-06,
+      "loss": 0.4077,
+      "step": 7653
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.752246063973086e-06,
+      "loss": 0.4095,
+      "step": 7654
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.74992168276168e-06,
+      "loss": 0.4475,
+      "step": 7655
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.747597370159008e-06,
+      "loss": 0.4129,
+      "step": 7656
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.745273126292636e-06,
+      "loss": 0.4173,
+      "step": 7657
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.742948951290128e-06,
+      "loss": 0.4161,
+      "step": 7658
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.740624845279045e-06,
+      "loss": 0.4105,
+      "step": 7659
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.738300808386934e-06,
+      "loss": 0.406,
+      "step": 7660
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.735976840741358e-06,
+      "loss": 0.43,
+      "step": 7661
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.73365294246986e-06,
+      "loss": 0.4557,
+      "step": 7662
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.731329113699978e-06,
+      "loss": 0.4112,
+      "step": 7663
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.729005354559261e-06,
+      "loss": 0.4335,
+      "step": 7664
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.726681665175241e-06,
+      "loss": 0.4134,
+      "step": 7665
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.724358045675448e-06,
+      "loss": 0.4109,
+      "step": 7666
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.722034496187416e-06,
+      "loss": 0.4167,
+      "step": 7667
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.719711016838662e-06,
+      "loss": 0.4399,
+      "step": 7668
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.717387607756714e-06,
+      "loss": 0.4345,
+      "step": 7669
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.715064269069086e-06,
+      "loss": 0.4463,
+      "step": 7670
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.71274100090329e-06,
+      "loss": 0.4035,
+      "step": 7671
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.710417803386839e-06,
+      "loss": 0.3948,
+      "step": 7672
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.708094676647236e-06,
+      "loss": 0.4397,
+      "step": 7673
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.705771620811982e-06,
+      "loss": 0.4286,
+      "step": 7674
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.703448636008571e-06,
+      "loss": 0.4208,
+      "step": 7675
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.701125722364506e-06,
+      "loss": 0.3951,
+      "step": 7676
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.698802880007265e-06,
+      "loss": 0.4413,
+      "step": 7677
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.696480109064343e-06,
+      "loss": 0.4075,
+      "step": 7678
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.694157409663217e-06,
+      "loss": 0.465,
+      "step": 7679
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.69183478193137e-06,
+      "loss": 0.4133,
+      "step": 7680
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.689512225996274e-06,
+      "loss": 0.4167,
+      "step": 7681
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.687189741985399e-06,
+      "loss": 0.417,
+      "step": 7682
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.684867330026205e-06,
+      "loss": 0.4266,
+      "step": 7683
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.682544990246162e-06,
+      "loss": 0.4174,
+      "step": 7684
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.680222722772721e-06,
+      "loss": 0.4,
+      "step": 7685
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.677900527733348e-06,
+      "loss": 0.4582,
+      "step": 7686
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.675578405255485e-06,
+      "loss": 0.402,
+      "step": 7687
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.673256355466577e-06,
+      "loss": 0.4283,
+      "step": 7688
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.670934378494072e-06,
+      "loss": 0.4093,
+      "step": 7689
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.668612474465405e-06,
+      "loss": 0.4453,
+      "step": 7690
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.66629064350801e-06,
+      "loss": 0.4369,
+      "step": 7691
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.663968885749317e-06,
+      "loss": 0.4037,
+      "step": 7692
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.66164720131675e-06,
+      "loss": 0.4047,
+      "step": 7693
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.65932559033774e-06,
+      "loss": 0.4203,
+      "step": 7694
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.657004052939698e-06,
+      "loss": 0.4076,
+      "step": 7695
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.654682589250038e-06,
+      "loss": 0.3951,
+      "step": 7696
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.652361199396176e-06,
+      "loss": 0.4225,
+      "step": 7697
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.650039883505514e-06,
+      "loss": 0.4286,
+      "step": 7698
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.647718641705452e-06,
+      "loss": 0.3817,
+      "step": 7699
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.645397474123391e-06,
+      "loss": 0.3906,
+      "step": 7700
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.64307638088672e-06,
+      "loss": 0.408,
+      "step": 7701
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.640755362122838e-06,
+      "loss": 0.4673,
+      "step": 7702
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.638434417959128e-06,
+      "loss": 0.4623,
+      "step": 7703
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.636113548522966e-06,
+      "loss": 0.4308,
+      "step": 7704
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.633792753941733e-06,
+      "loss": 0.428,
+      "step": 7705
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.631472034342806e-06,
+      "loss": 0.405,
+      "step": 7706
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.629151389853548e-06,
+      "loss": 0.4142,
+      "step": 7707
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.626830820601325e-06,
+      "loss": 0.4344,
+      "step": 7708
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.6245103267135e-06,
+      "loss": 0.4259,
+      "step": 7709
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.622189908317433e-06,
+      "loss": 0.4464,
+      "step": 7710
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.619869565540472e-06,
+      "loss": 0.3722,
+      "step": 7711
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.617549298509971e-06,
+      "loss": 0.4295,
+      "step": 7712
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.615229107353267e-06,
+      "loss": 0.4029,
+      "step": 7713
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.612908992197707e-06,
+      "loss": 0.4302,
+      "step": 7714
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.610588953170623e-06,
+      "loss": 0.4296,
+      "step": 7715
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.60826899039935e-06,
+      "loss": 0.4191,
+      "step": 7716
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.605949104011208e-06,
+      "loss": 0.3827,
+      "step": 7717
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.60362929413353e-06,
+      "loss": 0.3953,
+      "step": 7718
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.601309560893635e-06,
+      "loss": 0.3851,
+      "step": 7719
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.598989904418835e-06,
+      "loss": 0.4124,
+      "step": 7720
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.596670324836442e-06,
+      "loss": 0.4371,
+      "step": 7721
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.59435082227376e-06,
+      "loss": 0.4215,
+      "step": 7722
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.592031396858094e-06,
+      "loss": 0.4246,
+      "step": 7723
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.589712048716742e-06,
+      "loss": 0.4461,
+      "step": 7724
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.587392777976995e-06,
+      "loss": 0.4384,
+      "step": 7725
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.58507358476615e-06,
+      "loss": 0.4166,
+      "step": 7726
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.582754469211486e-06,
+      "loss": 0.433,
+      "step": 7727
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.58043543144029e-06,
+      "loss": 0.4538,
+      "step": 7728
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.578116471579836e-06,
+      "loss": 0.4097,
+      "step": 7729
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.575797589757391e-06,
+      "loss": 0.373,
+      "step": 7730
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.573478786100235e-06,
+      "loss": 0.4149,
+      "step": 7731
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.571160060735624e-06,
+      "loss": 0.4192,
+      "step": 7732
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.568841413790816e-06,
+      "loss": 0.4392,
+      "step": 7733
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.566522845393073e-06,
+      "loss": 0.4261,
+      "step": 7734
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.564204355669644e-06,
+      "loss": 0.4278,
+      "step": 7735
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.561885944747777e-06,
+      "loss": 0.4035,
+      "step": 7736
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.559567612754715e-06,
+      "loss": 0.4295,
+      "step": 7737
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.55724935981769e-06,
+      "loss": 0.4269,
+      "step": 7738
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.554931186063945e-06,
+      "loss": 0.4336,
+      "step": 7739
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.552613091620703e-06,
+      "loss": 0.3932,
+      "step": 7740
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.550295076615189e-06,
+      "loss": 0.3886,
+      "step": 7741
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.54797714117463e-06,
+      "loss": 0.4052,
+      "step": 7742
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.545659285426235e-06,
+      "loss": 0.4358,
+      "step": 7743
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.543341509497225e-06,
+      "loss": 0.4346,
+      "step": 7744
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.541023813514801e-06,
+      "loss": 0.4118,
+      "step": 7745
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.538706197606168e-06,
+      "loss": 0.4286,
+      "step": 7746
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.536388661898525e-06,
+      "loss": 0.45,
+      "step": 7747
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.534071206519066e-06,
+      "loss": 0.404,
+      "step": 7748
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.531753831594984e-06,
+      "loss": 0.4015,
+      "step": 7749
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.529436537253458e-06,
+      "loss": 0.4033,
+      "step": 7750
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.527119323621679e-06,
+      "loss": 0.3971,
+      "step": 7751
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.524802190826815e-06,
+      "loss": 0.4459,
+      "step": 7752
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.522485138996046e-06,
+      "loss": 0.4009,
+      "step": 7753
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.520168168256535e-06,
+      "loss": 0.4417,
+      "step": 7754
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.517851278735446e-06,
+      "loss": 0.3823,
+      "step": 7755
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.515534470559941e-06,
+      "loss": 0.3941,
+      "step": 7756
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.513217743857173e-06,
+      "loss": 0.4359,
+      "step": 7757
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.510901098754287e-06,
+      "loss": 0.3962,
+      "step": 7758
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.50858453537844e-06,
+      "loss": 0.4136,
+      "step": 7759
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.506268053856761e-06,
+      "loss": 0.4219,
+      "step": 7760
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.503951654316399e-06,
+      "loss": 0.438,
+      "step": 7761
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.501635336884477e-06,
+      "loss": 0.3825,
+      "step": 7762
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.499319101688127e-06,
+      "loss": 0.4061,
+      "step": 7763
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.497002948854467e-06,
+      "loss": 0.4155,
+      "step": 7764
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.494686878510623e-06,
+      "loss": 0.4079,
+      "step": 7765
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.4923708907837e-06,
+      "loss": 0.4211,
+      "step": 7766
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.490054985800818e-06,
+      "loss": 0.4115,
+      "step": 7767
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.487739163689078e-06,
+      "loss": 0.4269,
+      "step": 7768
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.485423424575578e-06,
+      "loss": 0.4422,
+      "step": 7769
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.483107768587418e-06,
+      "loss": 0.3961,
+      "step": 7770
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.480792195851686e-06,
+      "loss": 0.4139,
+      "step": 7771
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.478476706495468e-06,
+      "loss": 0.4093,
+      "step": 7772
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.47616130064585e-06,
+      "loss": 0.3929,
+      "step": 7773
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.473845978429907e-06,
+      "loss": 0.4155,
+      "step": 7774
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.471530739974715e-06,
+      "loss": 0.3792,
+      "step": 7775
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.469215585407341e-06,
+      "loss": 0.4376,
+      "step": 7776
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.466900514854847e-06,
+      "loss": 0.4165,
+      "step": 7777
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.464585528444295e-06,
+      "loss": 0.4002,
+      "step": 7778
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.462270626302739e-06,
+      "loss": 0.4315,
+      "step": 7779
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.459955808557228e-06,
+      "loss": 0.453,
+      "step": 7780
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.457641075334808e-06,
+      "loss": 0.444,
+      "step": 7781
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.455326426762515e-06,
+      "loss": 0.4279,
+      "step": 7782
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.453011862967396e-06,
+      "loss": 0.4324,
+      "step": 7783
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.450697384076475e-06,
+      "loss": 0.4501,
+      "step": 7784
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.448382990216782e-06,
+      "loss": 0.4116,
+      "step": 7785
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.446068681515333e-06,
+      "loss": 0.3815,
+      "step": 7786
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.443754458099153e-06,
+      "loss": 0.4187,
+      "step": 7787
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.441440320095253e-06,
+      "loss": 0.419,
+      "step": 7788
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.439126267630635e-06,
+      "loss": 0.4126,
+      "step": 7789
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.43681230083231e-06,
+      "loss": 0.4107,
+      "step": 7790
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.43449841982727e-06,
+      "loss": 0.434,
+      "step": 7791
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.432184624742518e-06,
+      "loss": 0.4224,
+      "step": 7792
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.429870915705037e-06,
+      "loss": 0.406,
+      "step": 7793
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.427557292841812e-06,
+      "loss": 0.4331,
+      "step": 7794
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.425243756279824e-06,
+      "loss": 0.4443,
+      "step": 7795
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.42293030614605e-06,
+      "loss": 0.4258,
+      "step": 7796
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.420616942567453e-06,
+      "loss": 0.426,
+      "step": 7797
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.418303665671003e-06,
+      "loss": 0.415,
+      "step": 7798
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.415990475583665e-06,
+      "loss": 0.4285,
+      "step": 7799
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.41367737243239e-06,
+      "loss": 0.4288,
+      "step": 7800
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.411364356344132e-06,
+      "loss": 0.4035,
+      "step": 7801
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.409051427445836e-06,
+      "loss": 0.4116,
+      "step": 7802
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.406738585864445e-06,
+      "loss": 0.4403,
+      "step": 7803
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.404425831726895e-06,
+      "loss": 0.3961,
+      "step": 7804
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.402113165160118e-06,
+      "loss": 0.4389,
+      "step": 7805
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.399800586291038e-06,
+      "loss": 0.412,
+      "step": 7806
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.397488095246582e-06,
+      "loss": 0.4207,
+      "step": 7807
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.39517569215367e-06,
+      "loss": 0.4008,
+      "step": 7808
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.392863377139212e-06,
+      "loss": 0.4443,
+      "step": 7809
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.390551150330115e-06,
+      "loss": 0.4127,
+      "step": 7810
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.38823901185328e-06,
+      "loss": 0.4368,
+      "step": 7811
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.38592696183561e-06,
+      "loss": 0.3981,
+      "step": 7812
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.383615000403999e-06,
+      "loss": 0.4342,
+      "step": 7813
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.381303127685332e-06,
+      "loss": 0.4372,
+      "step": 7814
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.378991343806492e-06,
+      "loss": 0.4105,
+      "step": 7815
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.376679648894361e-06,
+      "loss": 0.3814,
+      "step": 7816
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.374368043075813e-06,
+      "loss": 0.4358,
+      "step": 7817
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.372056526477718e-06,
+      "loss": 0.422,
+      "step": 7818
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.369745099226937e-06,
+      "loss": 0.4049,
+      "step": 7819
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.367433761450332e-06,
+      "loss": 0.408,
+      "step": 7820
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.365122513274757e-06,
+      "loss": 0.4173,
+      "step": 7821
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.36281135482706e-06,
+      "loss": 0.4276,
+      "step": 7822
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.360500286234082e-06,
+      "loss": 0.395,
+      "step": 7823
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.35818930762267e-06,
+      "loss": 0.4255,
+      "step": 7824
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.355878419119657e-06,
+      "loss": 0.4097,
+      "step": 7825
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.353567620851873e-06,
+      "loss": 0.4111,
+      "step": 7826
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.35125691294614e-06,
+      "loss": 0.4147,
+      "step": 7827
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.348946295529277e-06,
+      "loss": 0.3948,
+      "step": 7828
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.346635768728104e-06,
+      "loss": 0.4085,
+      "step": 7829
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.344325332669428e-06,
+      "loss": 0.4121,
+      "step": 7830
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.342014987480047e-06,
+      "loss": 0.425,
+      "step": 7831
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.339704733286775e-06,
+      "loss": 0.4447,
+      "step": 7832
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.337394570216398e-06,
+      "loss": 0.4181,
+      "step": 7833
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.335084498395706e-06,
+      "loss": 0.4125,
+      "step": 7834
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.332774517951489e-06,
+      "loss": 0.4231,
+      "step": 7835
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.330464629010521e-06,
+      "loss": 0.4108,
+      "step": 7836
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.328154831699583e-06,
+      "loss": 0.4195,
+      "step": 7837
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.32584512614544e-06,
+      "loss": 0.3972,
+      "step": 7838
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.323535512474854e-06,
+      "loss": 0.4247,
+      "step": 7839
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.321225990814592e-06,
+      "loss": 0.4283,
+      "step": 7840
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.318916561291405e-06,
+      "loss": 0.4409,
+      "step": 7841
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.316607224032044e-06,
+      "loss": 0.3813,
+      "step": 7842
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.314297979163256e-06,
+      "loss": 0.4079,
+      "step": 7843
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.311988826811777e-06,
+      "loss": 0.4479,
+      "step": 7844
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.309679767104339e-06,
+      "loss": 0.4227,
+      "step": 7845
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.307370800167675e-06,
+      "loss": 0.4424,
+      "step": 7846
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.305061926128507e-06,
+      "loss": 0.4202,
+      "step": 7847
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.302753145113558e-06,
+      "loss": 0.4201,
+      "step": 7848
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.300444457249544e-06,
+      "loss": 0.3968,
+      "step": 7849
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.298135862663165e-06,
+      "loss": 0.4331,
+      "step": 7850
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.295827361481132e-06,
+      "loss": 0.4145,
+      "step": 7851
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.293518953830141e-06,
+      "loss": 0.4117,
+      "step": 7852
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.291210639836886e-06,
+      "loss": 0.4509,
+      "step": 7853
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.288902419628057e-06,
+      "loss": 0.4158,
+      "step": 7854
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.286594293330333e-06,
+      "loss": 0.4404,
+      "step": 7855
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.284286261070399e-06,
+      "loss": 0.4403,
+      "step": 7856
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.281978322974924e-06,
+      "loss": 0.4648,
+      "step": 7857
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.279670479170575e-06,
+      "loss": 0.4233,
+      "step": 7858
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.277362729784015e-06,
+      "loss": 0.431,
+      "step": 7859
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.275055074941904e-06,
+      "loss": 0.3965,
+      "step": 7860
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.27274751477089e-06,
+      "loss": 0.4198,
+      "step": 7861
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.270440049397626e-06,
+      "loss": 0.4135,
+      "step": 7862
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.268132678948744e-06,
+      "loss": 0.4144,
+      "step": 7863
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.265825403550892e-06,
+      "loss": 0.4191,
+      "step": 7864
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.3869,
+      "step": 7865
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.261211138414784e-06,
+      "loss": 0.3936,
+      "step": 7866
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.258904148929775e-06,
+      "loss": 0.4074,
+      "step": 7867
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.256597255002286e-06,
+      "loss": 0.4168,
+      "step": 7868
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.254290456758928e-06,
+      "loss": 0.4107,
+      "step": 7869
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.251983754326304e-06,
+      "loss": 0.4188,
+      "step": 7870
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.24967714783101e-06,
+      "loss": 0.3982,
+      "step": 7871
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.247370637399655e-06,
+      "loss": 0.437,
+      "step": 7872
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.245064223158816e-06,
+      "loss": 0.3957,
+      "step": 7873
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.242757905235082e-06,
+      "loss": 0.4069,
+      "step": 7874
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.24045168375503e-06,
+      "loss": 0.4339,
+      "step": 7875
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.238145558845235e-06,
+      "loss": 0.4082,
+      "step": 7876
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.235839530632265e-06,
+      "loss": 0.3659,
+      "step": 7877
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.233533599242681e-06,
+      "loss": 0.4344,
+      "step": 7878
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.231227764803042e-06,
+      "loss": 0.4254,
+      "step": 7879
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.228922027439902e-06,
+      "loss": 0.4296,
+      "step": 7880
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.226616387279808e-06,
+      "loss": 0.4004,
+      "step": 7881
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.2243108444493e-06,
+      "loss": 0.3938,
+      "step": 7882
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.222005399074913e-06,
+      "loss": 0.4218,
+      "step": 7883
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.219700051283183e-06,
+      "loss": 0.3735,
+      "step": 7884
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.217394801200632e-06,
+      "loss": 0.4435,
+      "step": 7885
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.215089648953783e-06,
+      "loss": 0.4238,
+      "step": 7886
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.212784594669144e-06,
+      "loss": 0.4485,
+      "step": 7887
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.210479638473228e-06,
+      "loss": 0.4067,
+      "step": 7888
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.208174780492545e-06,
+      "loss": 0.4735,
+      "step": 7889
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.20587002085359e-06,
+      "loss": 0.4356,
+      "step": 7890
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.203565359682854e-06,
+      "loss": 0.4043,
+      "step": 7891
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.201260797106828e-06,
+      "loss": 0.433,
+      "step": 7892
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.198956333251993e-06,
+      "loss": 0.4262,
+      "step": 7893
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.196651968244826e-06,
+      "loss": 0.3936,
+      "step": 7894
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.1943477022118e-06,
+      "loss": 0.4392,
+      "step": 7895
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.192043535279377e-06,
+      "loss": 0.421,
+      "step": 7896
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.189739467574022e-06,
+      "loss": 0.4052,
+      "step": 7897
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.187435499222194e-06,
+      "loss": 0.391,
+      "step": 7898
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.185131630350337e-06,
+      "loss": 0.3933,
+      "step": 7899
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.182827861084896e-06,
+      "loss": 0.4216,
+      "step": 7900
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.180524191552315e-06,
+      "loss": 0.4262,
+      "step": 7901
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.178220621879022e-06,
+      "loss": 0.4614,
+      "step": 7902
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.175917152191447e-06,
+      "loss": 0.4551,
+      "step": 7903
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.173613782616008e-06,
+      "loss": 0.411,
+      "step": 7904
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.17131051327913e-06,
+      "loss": 0.4136,
+      "step": 7905
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.169007344307223e-06,
+      "loss": 0.4323,
+      "step": 7906
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.16670427582669e-06,
+      "loss": 0.42,
+      "step": 7907
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.164401307963934e-06,
+      "loss": 0.4308,
+      "step": 7908
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.162098440845347e-06,
+      "loss": 0.3966,
+      "step": 7909
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.159795674597321e-06,
+      "loss": 0.4096,
+      "step": 7910
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.157493009346243e-06,
+      "loss": 0.4548,
+      "step": 7911
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.15519044521848e-06,
+      "loss": 0.416,
+      "step": 7912
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.152887982340418e-06,
+      "loss": 0.4705,
+      "step": 7913
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.150585620838418e-06,
+      "loss": 0.4447,
+      "step": 7914
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.148283360838844e-06,
+      "loss": 0.4549,
+      "step": 7915
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.145981202468052e-06,
+      "loss": 0.4234,
+      "step": 7916
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.14367914585239e-06,
+      "loss": 0.4127,
+      "step": 7917
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.141377191118206e-06,
+      "loss": 0.4116,
+      "step": 7918
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.139075338391836e-06,
+      "loss": 0.4464,
+      "step": 7919
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.136773587799614e-06,
+      "loss": 0.3558,
+      "step": 7920
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.134471939467874e-06,
+      "loss": 0.4057,
+      "step": 7921
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.132170393522933e-06,
+      "loss": 0.4394,
+      "step": 7922
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.129868950091113e-06,
+      "loss": 0.4285,
+      "step": 7923
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.127567609298722e-06,
+      "loss": 0.4402,
+      "step": 7924
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.125266371272063e-06,
+      "loss": 0.3998,
+      "step": 7925
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.122965236137442e-06,
+      "loss": 0.4162,
+      "step": 7926
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.12066420402115e-06,
+      "loss": 0.3832,
+      "step": 7927
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.118363275049473e-06,
+      "loss": 0.3969,
+      "step": 7928
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.116062449348701e-06,
+      "loss": 0.4311,
+      "step": 7929
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.113761727045106e-06,
+      "loss": 0.46,
+      "step": 7930
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.111461108264965e-06,
+      "loss": 0.4415,
+      "step": 7931
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.10916059313454e-06,
+      "loss": 0.424,
+      "step": 7932
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.106860181780093e-06,
+      "loss": 0.4138,
+      "step": 7933
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.104559874327876e-06,
+      "loss": 0.3944,
+      "step": 7934
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.102259670904142e-06,
+      "loss": 0.4113,
+      "step": 7935
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.099959571635128e-06,
+      "loss": 0.4209,
+      "step": 7936
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.097659576647081e-06,
+      "loss": 0.4279,
+      "step": 7937
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.095359686066228e-06,
+      "loss": 0.4059,
+      "step": 7938
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.093059900018793e-06,
+      "loss": 0.4087,
+      "step": 7939
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.090760218630999e-06,
+      "loss": 0.4244,
+      "step": 7940
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.088460642029064e-06,
+      "loss": 0.4326,
+      "step": 7941
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.086161170339189e-06,
+      "loss": 0.4372,
+      "step": 7942
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.083861803687583e-06,
+      "loss": 0.4192,
+      "step": 7943
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.081562542200438e-06,
+      "loss": 0.3933,
+      "step": 7944
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.079263386003952e-06,
+      "loss": 0.4298,
+      "step": 7945
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.076964335224311e-06,
+      "loss": 0.3904,
+      "step": 7946
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.07466538998769e-06,
+      "loss": 0.4123,
+      "step": 7947
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.072366550420266e-06,
+      "loss": 0.3985,
+      "step": 7948
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.070067816648209e-06,
+      "loss": 0.4129,
+      "step": 7949
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.067769188797679e-06,
+      "loss": 0.4068,
+      "step": 7950
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.06547066699483e-06,
+      "loss": 0.4149,
+      "step": 7951
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.063172251365819e-06,
+      "loss": 0.4032,
+      "step": 7952
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.060873942036787e-06,
+      "loss": 0.3867,
+      "step": 7953
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.058575739133877e-06,
+      "loss": 0.4281,
+      "step": 7954
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.056277642783222e-06,
+      "loss": 0.4453,
+      "step": 7955
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.053979653110946e-06,
+      "loss": 0.441,
+      "step": 7956
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.051681770243176e-06,
+      "loss": 0.4325,
+      "step": 7957
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.049383994306023e-06,
+      "loss": 0.4313,
+      "step": 7958
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.0470863254256e-06,
+      "loss": 0.4247,
+      "step": 7959
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.044788763728011e-06,
+      "loss": 0.3875,
+      "step": 7960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.042491309339351e-06,
+      "loss": 0.389,
+      "step": 7961
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.040193962385718e-06,
+      "loss": 0.3916,
+      "step": 7962
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.037896722993197e-06,
+      "loss": 0.422,
+      "step": 7963
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.035599591287865e-06,
+      "loss": 0.3881,
+      "step": 7964
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.033302567395803e-06,
+      "loss": 0.4616,
+      "step": 7965
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.031005651443074e-06,
+      "loss": 0.438,
+      "step": 7966
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.028708843555743e-06,
+      "loss": 0.3939,
+      "step": 7967
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.026412143859865e-06,
+      "loss": 0.4119,
+      "step": 7968
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.024115552481491e-06,
+      "loss": 0.4314,
+      "step": 7969
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.021819069546672e-06,
+      "loss": 0.4318,
+      "step": 7970
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.019522695181445e-06,
+      "loss": 0.4339,
+      "step": 7971
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.01722642951184e-06,
+      "loss": 0.4475,
+      "step": 7972
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.014930272663884e-06,
+      "loss": 0.4221,
+      "step": 7973
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.0126342247636e-06,
+      "loss": 0.4119,
+      "step": 7974
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.010338285937006e-06,
+      "loss": 0.3983,
+      "step": 7975
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.008042456310105e-06,
+      "loss": 0.4174,
+      "step": 7976
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.005746736008901e-06,
+      "loss": 0.4078,
+      "step": 7977
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.003451125159399e-06,
+      "loss": 0.4029,
+      "step": 7978
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.001155623887585e-06,
+      "loss": 0.396,
+      "step": 7979
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.998860232319446e-06,
+      "loss": 0.4279,
+      "step": 7980
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.996564950580955e-06,
+      "loss": 0.4416,
+      "step": 7981
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.994269778798093e-06,
+      "loss": 0.454,
+      "step": 7982
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.991974717096824e-06,
+      "loss": 0.3875,
+      "step": 7983
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.989679765603108e-06,
+      "loss": 0.4168,
+      "step": 7984
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.9873849244429e-06,
+      "loss": 0.4196,
+      "step": 7985
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.985090193742151e-06,
+      "loss": 0.3902,
+      "step": 7986
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.982795573626804e-06,
+      "loss": 0.4296,
+      "step": 7987
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.980501064222797e-06,
+      "loss": 0.4116,
+      "step": 7988
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.978206665656057e-06,
+      "loss": 0.4183,
+      "step": 7989
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.975912378052511e-06,
+      "loss": 0.4303,
+      "step": 7990
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.97361820153808e-06,
+      "loss": 0.4126,
+      "step": 7991
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.971324136238672e-06,
+      "loss": 0.4283,
+      "step": 7992
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.969030182280192e-06,
+      "loss": 0.3873,
+      "step": 7993
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.966736339788545e-06,
+      "loss": 0.4153,
+      "step": 7994
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.964442608889627e-06,
+      "loss": 0.4253,
+      "step": 7995
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.962148989709322e-06,
+      "loss": 0.4357,
+      "step": 7996
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.959855482373513e-06,
+      "loss": 0.3854,
+      "step": 7997
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.957562087008076e-06,
+      "loss": 0.4261,
+      "step": 7998
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.95526880373888e-06,
+      "loss": 0.4279,
+      "step": 7999
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.95297563269179e-06,
+      "loss": 0.4387,
+      "step": 8000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.950682573992657e-06,
+      "loss": 0.4105,
+      "step": 8001
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.948389627767344e-06,
+      "loss": 0.4307,
+      "step": 8002
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.946096794141685e-06,
+      "loss": 0.4591,
+      "step": 8003
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.943804073241527e-06,
+      "loss": 0.4112,
+      "step": 8004
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.941511465192699e-06,
+      "loss": 0.4084,
+      "step": 8005
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.939218970121025e-06,
+      "loss": 0.3872,
+      "step": 8006
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.936926588152331e-06,
+      "loss": 0.4125,
+      "step": 8007
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.934634319412426e-06,
+      "loss": 0.4324,
+      "step": 8008
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.932342164027117e-06,
+      "loss": 0.4131,
+      "step": 8009
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.930050122122212e-06,
+      "loss": 0.4542,
+      "step": 8010
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.927758193823502e-06,
+      "loss": 0.4304,
+      "step": 8011
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.925466379256777e-06,
+      "loss": 0.415,
+      "step": 8012
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.92317467854782e-06,
+      "loss": 0.3976,
+      "step": 8013
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.92088309182241e-06,
+      "loss": 0.4252,
+      "step": 8014
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.91859161920631e-06,
+      "loss": 0.4182,
+      "step": 8015
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.916300260825293e-06,
+      "loss": 0.4025,
+      "step": 8016
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.914009016805107e-06,
+      "loss": 0.4218,
+      "step": 8017
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.911717887271516e-06,
+      "loss": 0.438,
+      "step": 8018
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.909426872350256e-06,
+      "loss": 0.404,
+      "step": 8019
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.90713597216707e-06,
+      "loss": 0.3882,
+      "step": 8020
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.90484518684769e-06,
+      "loss": 0.4063,
+      "step": 8021
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.902554516517842e-06,
+      "loss": 0.4584,
+      "step": 8022
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.900263961303243e-06,
+      "loss": 0.428,
+      "step": 8023
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.897973521329615e-06,
+      "loss": 0.4005,
+      "step": 8024
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.895683196722658e-06,
+      "loss": 0.4122,
+      "step": 8025
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.893392987608073e-06,
+      "loss": 0.4203,
+      "step": 8026
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.89110289411156e-06,
+      "loss": 0.4063,
+      "step": 8027
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.888812916358804e-06,
+      "loss": 0.4222,
+      "step": 8028
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.88652305447549e-06,
+      "loss": 0.4331,
+      "step": 8029
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.884233308587291e-06,
+      "loss": 0.4153,
+      "step": 8030
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.881943678819877e-06,
+      "loss": 0.4248,
+      "step": 8031
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.87965416529891e-06,
+      "loss": 0.4226,
+      "step": 8032
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.877364768150048e-06,
+      "loss": 0.3971,
+      "step": 8033
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.875075487498939e-06,
+      "loss": 0.4146,
+      "step": 8034
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.872786323471232e-06,
+      "loss": 0.4237,
+      "step": 8035
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.870497276192559e-06,
+      "loss": 0.3946,
+      "step": 8036
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.868208345788557e-06,
+      "loss": 0.4039,
+      "step": 8037
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.865919532384845e-06,
+      "loss": 0.3956,
+      "step": 8038
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.863630836107043e-06,
+      "loss": 0.4371,
+      "step": 8039
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.861342257080761e-06,
+      "loss": 0.4413,
+      "step": 8040
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.859053795431609e-06,
+      "loss": 0.403,
+      "step": 8041
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.85676545128518e-06,
+      "loss": 0.391,
+      "step": 8042
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.854477224767072e-06,
+      "loss": 0.4245,
+      "step": 8043
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.852189116002869e-06,
+      "loss": 0.4332,
+      "step": 8044
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.849901125118147e-06,
+      "loss": 0.4193,
+      "step": 8045
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.847613252238487e-06,
+      "loss": 0.4296,
+      "step": 8046
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.84532549748945e-06,
+      "loss": 0.4077,
+      "step": 8047
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.843037860996593e-06,
+      "loss": 0.3819,
+      "step": 8048
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.840750342885476e-06,
+      "loss": 0.3943,
+      "step": 8049
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.838462943281639e-06,
+      "loss": 0.4093,
+      "step": 8050
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.836175662310633e-06,
+      "loss": 0.4172,
+      "step": 8051
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.833888500097986e-06,
+      "loss": 0.4278,
+      "step": 8052
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.831601456769223e-06,
+      "loss": 0.4094,
+      "step": 8053
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.82931453244987e-06,
+      "loss": 0.4406,
+      "step": 8054
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.82702772726544e-06,
+      "loss": 0.4091,
+      "step": 8055
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.82474104134144e-06,
+      "loss": 0.4447,
+      "step": 8056
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.82245447480337e-06,
+      "loss": 0.4188,
+      "step": 8057
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.820168027776725e-06,
+      "loss": 0.4064,
+      "step": 8058
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.817881700386999e-06,
+      "loss": 0.4099,
+      "step": 8059
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.815595492759669e-06,
+      "loss": 0.4157,
+      "step": 8060
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.813309405020211e-06,
+      "loss": 0.4063,
+      "step": 8061
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.811023437294093e-06,
+      "loss": 0.439,
+      "step": 8062
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.808737589706779e-06,
+      "loss": 0.4064,
+      "step": 8063
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.806451862383724e-06,
+      "loss": 0.4285,
+      "step": 8064
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.804166255450372e-06,
+      "loss": 0.4432,
+      "step": 8065
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.801880769032171e-06,
+      "loss": 0.389,
+      "step": 8066
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.799595403254553e-06,
+      "loss": 0.423,
+      "step": 8067
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.797310158242952e-06,
+      "loss": 0.4309,
+      "step": 8068
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.795025034122786e-06,
+      "loss": 0.4096,
+      "step": 8069
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.792740031019472e-06,
+      "loss": 0.4377,
+      "step": 8070
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.790455149058418e-06,
+      "loss": 0.4011,
+      "step": 8071
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.788170388365029e-06,
+      "loss": 0.4505,
+      "step": 8072
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.785885749064698e-06,
+      "loss": 0.4265,
+      "step": 8073
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.783601231282813e-06,
+      "loss": 0.439,
+      "step": 8074
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.781316835144758e-06,
+      "loss": 0.4038,
+      "step": 8075
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.779032560775913e-06,
+      "loss": 0.4182,
+      "step": 8076
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.776748408301644e-06,
+      "loss": 0.38,
+      "step": 8077
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.774464377847313e-06,
+      "loss": 0.4247,
+      "step": 8078
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.772180469538273e-06,
+      "loss": 0.4165,
+      "step": 8079
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.769896683499878e-06,
+      "loss": 0.4637,
+      "step": 8080
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.767613019857467e-06,
+      "loss": 0.412,
+      "step": 8081
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.765329478736374e-06,
+      "loss": 0.4128,
+      "step": 8082
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.763046060261931e-06,
+      "loss": 0.445,
+      "step": 8083
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.760762764559461e-06,
+      "loss": 0.4048,
+      "step": 8084
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.758479591754277e-06,
+      "loss": 0.4027,
+      "step": 8085
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.756196541971691e-06,
+      "loss": 0.4093,
+      "step": 8086
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.753913615337e-06,
+      "loss": 0.4218,
+      "step": 8087
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.751630811975501e-06,
+      "loss": 0.4412,
+      "step": 8088
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.749348132012484e-06,
+      "loss": 0.4258,
+      "step": 8089
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.74706557557323e-06,
+      "loss": 0.386,
+      "step": 8090
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.744783142783008e-06,
+      "loss": 0.3994,
+      "step": 8091
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.742500833767094e-06,
+      "loss": 0.3741,
+      "step": 8092
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.740218648650748e-06,
+      "loss": 0.4163,
+      "step": 8093
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.737936587559223e-06,
+      "loss": 0.4353,
+      "step": 8094
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.735654650617764e-06,
+      "loss": 0.4349,
+      "step": 8095
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.733372837951615e-06,
+      "loss": 0.4783,
+      "step": 8096
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.73109114968601e-06,
+      "loss": 0.4308,
+      "step": 8097
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.728809585946177e-06,
+      "loss": 0.4208,
+      "step": 8098
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.726528146857329e-06,
+      "loss": 0.3845,
+      "step": 8099
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.724246832544688e-06,
+      "loss": 0.434,
+      "step": 8100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.721965643133458e-06,
+      "loss": 0.4088,
+      "step": 8101
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.719684578748839e-06,
+      "loss": 0.4037,
+      "step": 8102
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.717403639516025e-06,
+      "loss": 0.4454,
+      "step": 8103
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.715122825560198e-06,
+      "loss": 0.4369,
+      "step": 8104
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.712842137006539e-06,
+      "loss": 0.423,
+      "step": 8105
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.710561573980224e-06,
+      "loss": 0.4184,
+      "step": 8106
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.70828113660641e-06,
+      "loss": 0.4194,
+      "step": 8107
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.706000825010266e-06,
+      "loss": 0.4194,
+      "step": 8108
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.703720639316933e-06,
+      "loss": 0.4061,
+      "step": 8109
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.701440579651566e-06,
+      "loss": 0.3934,
+      "step": 8110
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.699160646139297e-06,
+      "loss": 0.4206,
+      "step": 8111
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.696880838905256e-06,
+      "loss": 0.3837,
+      "step": 8112
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.69460115807457e-06,
+      "loss": 0.4288,
+      "step": 8113
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.692321603772354e-06,
+      "loss": 0.4025,
+      "step": 8114
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.690042176123716e-06,
+      "loss": 0.4014,
+      "step": 8115
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.687762875253765e-06,
+      "loss": 0.4525,
+      "step": 8116
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.685483701287591e-06,
+      "loss": 0.4076,
+      "step": 8117
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.68320465435029e-06,
+      "loss": 0.3992,
+      "step": 8118
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.680925734566938e-06,
+      "loss": 0.4187,
+      "step": 8119
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.678646942062613e-06,
+      "loss": 0.4033,
+      "step": 8120
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.67636827696238e-06,
+      "loss": 0.411,
+      "step": 8121
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.674089739391306e-06,
+      "loss": 0.397,
+      "step": 8122
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.671811329474437e-06,
+      "loss": 0.3918,
+      "step": 8123
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.66953304733683e-06,
+      "loss": 0.4404,
+      "step": 8124
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.66725489310352e-06,
+      "loss": 0.4619,
+      "step": 8125
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.664976866899539e-06,
+      "loss": 0.4191,
+      "step": 8126
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.662698968849915e-06,
+      "loss": 0.4013,
+      "step": 8127
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.66042119907967e-06,
+      "loss": 0.3953,
+      "step": 8128
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.658143557713809e-06,
+      "loss": 0.415,
+      "step": 8129
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.655866044877343e-06,
+      "loss": 0.4443,
+      "step": 8130
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.653588660695263e-06,
+      "loss": 0.4161,
+      "step": 8131
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.651311405292572e-06,
+      "loss": 0.4345,
+      "step": 8132
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.649034278794246e-06,
+      "loss": 0.3932,
+      "step": 8133
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.64675728132526e-06,
+      "loss": 0.4168,
+      "step": 8134
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.644480413010586e-06,
+      "loss": 0.4091,
+      "step": 8135
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.64220367397519e-06,
+      "loss": 0.4246,
+      "step": 8136
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.639927064344023e-06,
+      "loss": 0.4175,
+      "step": 8137
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.637650584242034e-06,
+      "loss": 0.3985,
+      "step": 8138
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.635374233794162e-06,
+      "loss": 0.4003,
+      "step": 8139
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.633098013125347e-06,
+      "loss": 0.4123,
+      "step": 8140
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.630821922360516e-06,
+      "loss": 0.4095,
+      "step": 8141
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.628545961624585e-06,
+      "loss": 0.3915,
+      "step": 8142
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.626270131042467e-06,
+      "loss": 0.4056,
+      "step": 8143
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.62399443073907e-06,
+      "loss": 0.4116,
+      "step": 8144
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.621718860839292e-06,
+      "loss": 0.4289,
+      "step": 8145
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.619443421468021e-06,
+      "loss": 0.4031,
+      "step": 8146
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.617168112750145e-06,
+      "loss": 0.4434,
+      "step": 8147
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.614892934810538e-06,
+      "loss": 0.4238,
+      "step": 8148
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.612617887774076e-06,
+      "loss": 0.3933,
+      "step": 8149
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.610342971765616e-06,
+      "loss": 0.4285,
+      "step": 8150
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.608068186910014e-06,
+      "loss": 0.415,
+      "step": 8151
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.60579353333212e-06,
+      "loss": 0.4404,
+      "step": 8152
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.603519011156774e-06,
+      "loss": 0.4427,
+      "step": 8153
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6012446205088074e-06,
+      "loss": 0.4285,
+      "step": 8154
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.598970361513052e-06,
+      "loss": 0.4395,
+      "step": 8155
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.596696234294322e-06,
+      "loss": 0.446,
+      "step": 8156
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5944222389774354e-06,
+      "loss": 0.4271,
+      "step": 8157
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.592148375687194e-06,
+      "loss": 0.3898,
+      "step": 8158
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5898746445483936e-06,
+      "loss": 0.3914,
+      "step": 8159
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5876010456858285e-06,
+      "loss": 0.424,
+      "step": 8160
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.58532757922428e-06,
+      "loss": 0.4076,
+      "step": 8161
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5830542452885225e-06,
+      "loss": 0.4014,
+      "step": 8162
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.580781044003324e-06,
+      "loss": 0.4295,
+      "step": 8163
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.578507975493448e-06,
+      "loss": 0.4254,
+      "step": 8164
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.57623503988365e-06,
+      "loss": 0.4541,
+      "step": 8165
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.573962237298677e-06,
+      "loss": 0.4388,
+      "step": 8166
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.571689567863266e-06,
+      "loss": 0.4412,
+      "step": 8167
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.569417031702148e-06,
+      "loss": 0.4108,
+      "step": 8168
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.567144628940052e-06,
+      "loss": 0.4346,
+      "step": 8169
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.564872359701693e-06,
+      "loss": 0.4243,
+      "step": 8170
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.562600224111779e-06,
+      "loss": 0.4199,
+      "step": 8171
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.560328222295016e-06,
+      "loss": 0.4361,
+      "step": 8172
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.558056354376097e-06,
+      "loss": 0.3932,
+      "step": 8173
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.555784620479716e-06,
+      "loss": 0.4402,
+      "step": 8174
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.55351302073055e-06,
+      "loss": 0.4179,
+      "step": 8175
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.551241555253268e-06,
+      "loss": 0.3992,
+      "step": 8176
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5489702241725445e-06,
+      "loss": 0.4077,
+      "step": 8177
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.546699027613033e-06,
+      "loss": 0.4169,
+      "step": 8178
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.544427965699386e-06,
+      "loss": 0.4005,
+      "step": 8179
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.542157038556244e-06,
+      "loss": 0.3832,
+      "step": 8180
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.539886246308248e-06,
+      "loss": 0.4348,
+      "step": 8181
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.537615589080027e-06,
+      "loss": 0.4329,
+      "step": 8182
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.535345066996203e-06,
+      "loss": 0.4183,
+      "step": 8183
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.533074680181389e-06,
+      "loss": 0.4155,
+      "step": 8184
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.53080442876019e-06,
+      "loss": 0.4501,
+      "step": 8185
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.52853431285721e-06,
+      "loss": 0.4383,
+      "step": 8186
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.526264332597037e-06,
+      "loss": 0.4093,
+      "step": 8187
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5239944881042534e-06,
+      "loss": 0.3996,
+      "step": 8188
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.521724779503446e-06,
+      "loss": 0.3935,
+      "step": 8189
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.519455206919173e-06,
+      "loss": 0.3973,
+      "step": 8190
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.517185770476005e-06,
+      "loss": 0.4086,
+      "step": 8191
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.514916470298494e-06,
+      "loss": 0.4369,
+      "step": 8192
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.512647306511186e-06,
+      "loss": 0.4138,
+      "step": 8193
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5103782792386225e-06,
+      "loss": 0.4044,
+      "step": 8194
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.508109388605335e-06,
+      "loss": 0.4224,
+      "step": 8195
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.505840634735845e-06,
+      "loss": 0.4066,
+      "step": 8196
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.503572017754676e-06,
+      "loss": 0.4049,
+      "step": 8197
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.501303537786333e-06,
+      "loss": 0.391,
+      "step": 8198
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.499035194955322e-06,
+      "loss": 0.3982,
+      "step": 8199
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4967669893861364e-06,
+      "loss": 0.4348,
+      "step": 8200
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4944989212032614e-06,
+      "loss": 0.4233,
+      "step": 8201
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.492230990531176e-06,
+      "loss": 0.3787,
+      "step": 8202
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.489963197494357e-06,
+      "loss": 0.4192,
+      "step": 8203
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.487695542217262e-06,
+      "loss": 0.4515,
+      "step": 8204
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.485428024824355e-06,
+      "loss": 0.3878,
+      "step": 8205
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.483160645440084e-06,
+      "loss": 0.4023,
+      "step": 8206
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.480893404188887e-06,
+      "loss": 0.388,
+      "step": 8207
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.478626301195202e-06,
+      "loss": 0.4364,
+      "step": 8208
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.476359336583454e-06,
+      "loss": 0.4064,
+      "step": 8209
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.474092510478061e-06,
+      "loss": 0.4253,
+      "step": 8210
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.471825823003437e-06,
+      "loss": 0.4608,
+      "step": 8211
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4695592742839795e-06,
+      "loss": 0.4232,
+      "step": 8212
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.467292864444096e-06,
+      "loss": 0.4448,
+      "step": 8213
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.465026593608169e-06,
+      "loss": 0.4377,
+      "step": 8214
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.462760461900576e-06,
+      "loss": 0.4126,
+      "step": 8215
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.460494469445696e-06,
+      "loss": 0.4194,
+      "step": 8216
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.458228616367893e-06,
+      "loss": 0.4183,
+      "step": 8217
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.455962902791522e-06,
+      "loss": 0.4385,
+      "step": 8218
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4536973288409385e-06,
+      "loss": 0.3837,
+      "step": 8219
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.451431894640477e-06,
+      "loss": 0.4561,
+      "step": 8220
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4491666003144844e-06,
+      "loss": 0.4346,
+      "step": 8221
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.446901445987281e-06,
+      "loss": 0.4103,
+      "step": 8222
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.444636431783186e-06,
+      "loss": 0.4215,
+      "step": 8223
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.442371557826515e-06,
+      "loss": 0.4429,
+      "step": 8224
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.440106824241568e-06,
+      "loss": 0.4267,
+      "step": 8225
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.437842231152648e-06,
+      "loss": 0.4326,
+      "step": 8226
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.435577778684033e-06,
+      "loss": 0.4095,
+      "step": 8227
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.433313466960016e-06,
+      "loss": 0.4206,
+      "step": 8228
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.43104929610486e-06,
+      "loss": 0.3886,
+      "step": 8229
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4287852662428414e-06,
+      "loss": 0.4186,
+      "step": 8230
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.426521377498213e-06,
+      "loss": 0.4034,
+      "step": 8231
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.424257629995221e-06,
+      "loss": 0.4172,
+      "step": 8232
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.421994023858114e-06,
+      "loss": 0.4109,
+      "step": 8233
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.419730559211125e-06,
+      "loss": 0.4205,
+      "step": 8234
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.417467236178479e-06,
+      "loss": 0.4236,
+      "step": 8235
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.415204054884399e-06,
+      "loss": 0.4198,
+      "step": 8236
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.41294101545309e-06,
+      "loss": 0.4015,
+      "step": 8237
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.410678118008764e-06,
+      "loss": 0.4511,
+      "step": 8238
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.408415362675612e-06,
+      "loss": 0.4441,
+      "step": 8239
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.406152749577822e-06,
+      "loss": 0.3974,
+      "step": 8240
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.403890278839575e-06,
+      "loss": 0.4537,
+      "step": 8241
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.401627950585044e-06,
+      "loss": 0.4371,
+      "step": 8242
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.399365764938393e-06,
+      "loss": 0.419,
+      "step": 8243
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3971037220237775e-06,
+      "loss": 0.4104,
+      "step": 8244
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.394841821965345e-06,
+      "loss": 0.4129,
+      "step": 8245
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.392580064887244e-06,
+      "loss": 0.4134,
+      "step": 8246
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.390318450913601e-06,
+      "loss": 0.4315,
+      "step": 8247
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.388056980168545e-06,
+      "loss": 0.4175,
+      "step": 8248
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.385795652776191e-06,
+      "loss": 0.4308,
+      "step": 8249
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3835344688606505e-06,
+      "loss": 0.4141,
+      "step": 8250
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.381273428546024e-06,
+      "loss": 0.4103,
+      "step": 8251
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.379012531956404e-06,
+      "loss": 0.4333,
+      "step": 8252
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.376751779215878e-06,
+      "loss": 0.4242,
+      "step": 8253
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.374491170448525e-06,
+      "loss": 0.4275,
+      "step": 8254
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.372230705778418e-06,
+      "loss": 0.4259,
+      "step": 8255
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.369970385329615e-06,
+      "loss": 0.4262,
+      "step": 8256
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3677102092261695e-06,
+      "loss": 0.4331,
+      "step": 8257
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.365450177592133e-06,
+      "loss": 0.4415,
+      "step": 8258
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.36319029055154e-06,
+      "loss": 0.4128,
+      "step": 8259
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.360930548228422e-06,
+      "loss": 0.4073,
+      "step": 8260
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.358670950746798e-06,
+      "loss": 0.411,
+      "step": 8261
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.356411498230689e-06,
+      "loss": 0.4488,
+      "step": 8262
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3541521908041005e-06,
+      "loss": 0.4264,
+      "step": 8263
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.35189302859103e-06,
+      "loss": 0.3837,
+      "step": 8264
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.34963401171547e-06,
+      "loss": 0.3958,
+      "step": 8265
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.347375140301398e-06,
+      "loss": 0.3966,
+      "step": 8266
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.345116414472796e-06,
+      "loss": 0.455,
+      "step": 8267
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3428578343536275e-06,
+      "loss": 0.4055,
+      "step": 8268
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.340599400067847e-06,
+      "loss": 0.4213,
+      "step": 8269
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.338341111739412e-06,
+      "loss": 0.3926,
+      "step": 8270
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.336082969492266e-06,
+      "loss": 0.4337,
+      "step": 8271
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.333824973450341e-06,
+      "loss": 0.4187,
+      "step": 8272
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.331567123737564e-06,
+      "loss": 0.3977,
+      "step": 8273
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.32930942047785e-06,
+      "loss": 0.4556,
+      "step": 8274
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.327051863795118e-06,
+      "loss": 0.4377,
+      "step": 8275
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.324794453813264e-06,
+      "loss": 0.4481,
+      "step": 8276
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.322537190656182e-06,
+      "loss": 0.4072,
+      "step": 8277
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3202800744477666e-06,
+      "loss": 0.3816,
+      "step": 8278
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.318023105311889e-06,
+      "loss": 0.422,
+      "step": 8279
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.315766283372423e-06,
+      "loss": 0.4058,
+      "step": 8280
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.313509608753231e-06,
+      "loss": 0.4138,
+      "step": 8281
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.311253081578162e-06,
+      "loss": 0.4285,
+      "step": 8282
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.308996701971072e-06,
+      "loss": 0.405,
+      "step": 8283
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.306740470055791e-06,
+      "loss": 0.4189,
+      "step": 8284
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.304484385956149e-06,
+      "loss": 0.4345,
+      "step": 8285
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.302228449795974e-06,
+      "loss": 0.4584,
+      "step": 8286
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.299972661699074e-06,
+      "loss": 0.4014,
+      "step": 8287
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.297717021789259e-06,
+      "loss": 0.4401,
+      "step": 8288
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.295461530190326e-06,
+      "loss": 0.427,
+      "step": 8289
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.29320618702606e-06,
+      "loss": 0.4108,
+      "step": 8290
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.290950992420247e-06,
+      "loss": 0.4111,
+      "step": 8291
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2886959464966574e-06,
+      "loss": 0.4243,
+      "step": 8292
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2864410493790536e-06,
+      "loss": 0.4235,
+      "step": 8293
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2841863011912006e-06,
+      "loss": 0.4056,
+      "step": 8294
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2819317020568415e-06,
+      "loss": 0.4177,
+      "step": 8295
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.279677252099715e-06,
+      "loss": 0.4339,
+      "step": 8296
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.277422951443558e-06,
+      "loss": 0.4271,
+      "step": 8297
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.275168800212092e-06,
+      "loss": 0.3895,
+      "step": 8298
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.27291479852903e-06,
+      "loss": 0.4222,
+      "step": 8299
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2706609465180844e-06,
+      "loss": 0.4398,
+      "step": 8300
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.268407244302955e-06,
+      "loss": 0.4334,
+      "step": 8301
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.266153692007325e-06,
+      "loss": 0.4366,
+      "step": 8302
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.263900289754887e-06,
+      "loss": 0.4225,
+      "step": 8303
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2616470376693105e-06,
+      "loss": 0.4118,
+      "step": 8304
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.259393935874265e-06,
+      "loss": 0.4447,
+      "step": 8305
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.257140984493407e-06,
+      "loss": 0.3885,
+      "step": 8306
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2548881836503856e-06,
+      "loss": 0.4171,
+      "step": 8307
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.252635533468844e-06,
+      "loss": 0.4233,
+      "step": 8308
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.250383034072415e-06,
+      "loss": 0.4402,
+      "step": 8309
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.24813068558472e-06,
+      "loss": 0.419,
+      "step": 8310
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.245878488129383e-06,
+      "loss": 0.4135,
+      "step": 8311
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.243626441830009e-06,
+      "loss": 0.4038,
+      "step": 8312
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2413745468101985e-06,
+      "loss": 0.4138,
+      "step": 8313
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.239122803193543e-06,
+      "loss": 0.4143,
+      "step": 8314
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2368712111036285e-06,
+      "loss": 0.424,
+      "step": 8315
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.234619770664026e-06,
+      "loss": 0.4004,
+      "step": 8316
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.232368481998309e-06,
+      "loss": 0.4475,
+      "step": 8317
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2301173452300255e-06,
+      "loss": 0.4251,
+      "step": 8318
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.227866360482738e-06,
+      "loss": 0.4016,
+      "step": 8319
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.225615527879983e-06,
+      "loss": 0.4493,
+      "step": 8320
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.223364847545292e-06,
+      "loss": 0.4256,
+      "step": 8321
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.221114319602195e-06,
+      "loss": 0.3884,
+      "step": 8322
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.218863944174207e-06,
+      "loss": 0.4115,
+      "step": 8323
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.216613721384836e-06,
+      "loss": 0.4284,
+      "step": 8324
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.214363651357581e-06,
+      "loss": 0.4309,
+      "step": 8325
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2121137342159316e-06,
+      "loss": 0.4071,
+      "step": 8326
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.209863970083381e-06,
+      "loss": 0.4115,
+      "step": 8327
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.207614359083397e-06,
+      "loss": 0.4091,
+      "step": 8328
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.205364901339445e-06,
+      "loss": 0.4455,
+      "step": 8329
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.20311559697499e-06,
+      "loss": 0.4118,
+      "step": 8330
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.200866446113475e-06,
+      "loss": 0.3754,
+      "step": 8331
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.198617448878344e-06,
+      "loss": 0.4255,
+      "step": 8332
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.196368605393029e-06,
+      "loss": 0.4105,
+      "step": 8333
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.194119915780952e-06,
+      "loss": 0.3928,
+      "step": 8334
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.191871380165538e-06,
+      "loss": 0.3962,
+      "step": 8335
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.189622998670189e-06,
+      "loss": 0.3907,
+      "step": 8336
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.187374771418303e-06,
+      "loss": 0.4301,
+      "step": 8337
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.18512669853327e-06,
+      "loss": 0.4347,
+      "step": 8338
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1828787801384755e-06,
+      "loss": 0.4186,
+      "step": 8339
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.180631016357293e-06,
+      "loss": 0.3492,
+      "step": 8340
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.178383407313086e-06,
+      "loss": 0.3703,
+      "step": 8341
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.176135953129208e-06,
+      "loss": 0.419,
+      "step": 8342
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.173888653929013e-06,
+      "loss": 0.3829,
+      "step": 8343
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.171641509835841e-06,
+      "loss": 0.4168,
+      "step": 8344
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1693945209730206e-06,
+      "loss": 0.437,
+      "step": 8345
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.167147687463874e-06,
+      "loss": 0.4147,
+      "step": 8346
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.164901009431717e-06,
+      "loss": 0.4226,
+      "step": 8347
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.162654486999856e-06,
+      "loss": 0.3897,
+      "step": 8348
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.160408120291587e-06,
+      "loss": 0.4116,
+      "step": 8349
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.158161909430192e-06,
+      "loss": 0.4479,
+      "step": 8350
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.155915854538961e-06,
+      "loss": 0.399,
+      "step": 8351
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1536699557411625e-06,
+      "loss": 0.3844,
+      "step": 8352
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.151424213160061e-06,
+      "loss": 0.3748,
+      "step": 8353
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1491786269189065e-06,
+      "loss": 0.4312,
+      "step": 8354
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1469331971409465e-06,
+      "loss": 0.4338,
+      "step": 8355
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1446879239494205e-06,
+      "loss": 0.3716,
+      "step": 8356
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.142442807467553e-06,
+      "loss": 0.398,
+      "step": 8357
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.140197847818564e-06,
+      "loss": 0.4062,
+      "step": 8358
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.13795304512567e-06,
+      "loss": 0.409,
+      "step": 8359
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.135708399512068e-06,
+      "loss": 0.3653,
+      "step": 8360
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1334639111009575e-06,
+      "loss": 0.4242,
+      "step": 8361
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.13121958001552e-06,
+      "loss": 0.3919,
+      "step": 8362
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.128975406378933e-06,
+      "loss": 0.408,
+      "step": 8363
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.126731390314367e-06,
+      "loss": 0.3961,
+      "step": 8364
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.12448753194498e-06,
+      "loss": 0.4676,
+      "step": 8365
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1222438313939225e-06,
+      "loss": 0.4075,
+      "step": 8366
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.120000288784332e-06,
+      "loss": 0.3919,
+      "step": 8367
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1177569042393506e-06,
+      "loss": 0.4663,
+      "step": 8368
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1155136778820995e-06,
+      "loss": 0.4306,
+      "step": 8369
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1132706098356985e-06,
+      "loss": 0.3721,
+      "step": 8370
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.111027700223249e-06,
+      "loss": 0.3702,
+      "step": 8371
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1087849491678505e-06,
+      "loss": 0.434,
+      "step": 8372
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.106542356792599e-06,
+      "loss": 0.4245,
+      "step": 8373
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.104299923220571e-06,
+      "loss": 0.4055,
+      "step": 8374
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.102057648574836e-06,
+      "loss": 0.4134,
+      "step": 8375
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.099815532978466e-06,
+      "loss": 0.3857,
+      "step": 8376
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.097573576554513e-06,
+      "loss": 0.4119,
+      "step": 8377
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.095331779426023e-06,
+      "loss": 0.377,
+      "step": 8378
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.093090141716035e-06,
+      "loss": 0.3774,
+      "step": 8379
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.090848663547574e-06,
+      "loss": 0.4156,
+      "step": 8380
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0886073450436655e-06,
+      "loss": 0.4224,
+      "step": 8381
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.086366186327319e-06,
+      "loss": 0.4273,
+      "step": 8382
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.084125187521533e-06,
+      "loss": 0.3864,
+      "step": 8383
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0818843487493085e-06,
+      "loss": 0.4245,
+      "step": 8384
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.079643670133627e-06,
+      "loss": 0.3756,
+      "step": 8385
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.077403151797467e-06,
+      "loss": 0.4356,
+      "step": 8386
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0751627938637945e-06,
+      "loss": 0.4241,
+      "step": 8387
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0729225964555655e-06,
+      "loss": 0.4481,
+      "step": 8388
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.070682559695737e-06,
+      "loss": 0.4115,
+      "step": 8389
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.068442683707244e-06,
+      "loss": 0.4594,
+      "step": 8390
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.066202968613019e-06,
+      "loss": 0.4148,
+      "step": 8391
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.063963414535989e-06,
+      "loss": 0.4369,
+      "step": 8392
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.061724021599066e-06,
+      "loss": 0.4108,
+      "step": 8393
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0594847899251596e-06,
+      "loss": 0.4193,
+      "step": 8394
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.057245719637164e-06,
+      "loss": 0.4483,
+      "step": 8395
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0550068108579674e-06,
+      "loss": 0.4363,
+      "step": 8396
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.052768063710447e-06,
+      "loss": 0.3871,
+      "step": 8397
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.050529478317476e-06,
+      "loss": 0.4001,
+      "step": 8398
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.048291054801912e-06,
+      "loss": 0.4079,
+      "step": 8399
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.046052793286615e-06,
+      "loss": 0.4044,
+      "step": 8400
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.043814693894426e-06,
+      "loss": 0.3982,
+      "step": 8401
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0415767567481745e-06,
+      "loss": 0.45,
+      "step": 8402
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.039338981970693e-06,
+      "loss": 0.4564,
+      "step": 8403
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.037101369684796e-06,
+      "loss": 0.381,
+      "step": 8404
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.03486392001329e-06,
+      "loss": 0.4128,
+      "step": 8405
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.032626633078976e-06,
+      "loss": 0.407,
+      "step": 8406
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0303895090046405e-06,
+      "loss": 0.4268,
+      "step": 8407
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.028152547913073e-06,
+      "loss": 0.401,
+      "step": 8408
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.025915749927041e-06,
+      "loss": 0.3959,
+      "step": 8409
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.023679115169305e-06,
+      "loss": 0.424,
+      "step": 8410
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0214426437626244e-06,
+      "loss": 0.4084,
+      "step": 8411
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0192063358297444e-06,
+      "loss": 0.4087,
+      "step": 8412
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.016970191493397e-06,
+      "loss": 0.4573,
+      "step": 8413
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0147342108763125e-06,
+      "loss": 0.4431,
+      "step": 8414
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.012498394101208e-06,
+      "loss": 0.3963,
+      "step": 8415
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.010262741290799e-06,
+      "loss": 0.3886,
+      "step": 8416
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0080272525677796e-06,
+      "loss": 0.4148,
+      "step": 8417
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.005791928054844e-06,
+      "loss": 0.3964,
+      "step": 8418
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.003556767874673e-06,
+      "loss": 0.4038,
+      "step": 8419
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.001321772149943e-06,
+      "loss": 0.3886,
+      "step": 8420
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.999086941003315e-06,
+      "loss": 0.4084,
+      "step": 8421
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.996852274557446e-06,
+      "loss": 0.3921,
+      "step": 8422
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.994617772934982e-06,
+      "loss": 0.3851,
+      "step": 8423
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.992383436258561e-06,
+      "loss": 0.4288,
+      "step": 8424
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.990149264650814e-06,
+      "loss": 0.4151,
+      "step": 8425
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.987915258234357e-06,
+      "loss": 0.404,
+      "step": 8426
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9856814171318e-06,
+      "loss": 0.4376,
+      "step": 8427
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9834477414657455e-06,
+      "loss": 0.4222,
+      "step": 8428
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.981214231358786e-06,
+      "loss": 0.4168,
+      "step": 8429
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.978980886933502e-06,
+      "loss": 0.4307,
+      "step": 8430
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.976747708312466e-06,
+      "loss": 0.4097,
+      "step": 8431
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.974514695618247e-06,
+      "loss": 0.4322,
+      "step": 8432
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.972281848973401e-06,
+      "loss": 0.4291,
+      "step": 8433
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.970049168500474e-06,
+      "loss": 0.41,
+      "step": 8434
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.967816654322001e-06,
+      "loss": 0.3795,
+      "step": 8435
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.965584306560511e-06,
+      "loss": 0.3691,
+      "step": 8436
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.963352125338525e-06,
+      "loss": 0.4082,
+      "step": 8437
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.961120110778552e-06,
+      "loss": 0.3808,
+      "step": 8438
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.95888826300309e-06,
+      "loss": 0.4329,
+      "step": 8439
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9566565821346334e-06,
+      "loss": 0.4618,
+      "step": 8440
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.954425068295665e-06,
+      "loss": 0.4311,
+      "step": 8441
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.95219372160866e-06,
+      "loss": 0.4205,
+      "step": 8442
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.949962542196081e-06,
+      "loss": 0.433,
+      "step": 8443
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.947731530180379e-06,
+      "loss": 0.4005,
+      "step": 8444
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.945500685684006e-06,
+      "loss": 0.3733,
+      "step": 8445
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.943270008829397e-06,
+      "loss": 0.3967,
+      "step": 8446
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.941039499738975e-06,
+      "loss": 0.4084,
+      "step": 8447
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.938809158535164e-06,
+      "loss": 0.4255,
+      "step": 8448
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9365789853403684e-06,
+      "loss": 0.4184,
+      "step": 8449
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.934348980276995e-06,
+      "loss": 0.3965,
+      "step": 8450
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.932119143467428e-06,
+      "loss": 0.4183,
+      "step": 8451
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9298894750340486e-06,
+      "loss": 0.383,
+      "step": 8452
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.927659975099235e-06,
+      "loss": 0.3969,
+      "step": 8453
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.925430643785344e-06,
+      "loss": 0.387,
+      "step": 8454
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.923201481214732e-06,
+      "loss": 0.389,
+      "step": 8455
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.920972487509737e-06,
+      "loss": 0.409,
+      "step": 8456
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.918743662792704e-06,
+      "loss": 0.4007,
+      "step": 8457
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.916515007185956e-06,
+      "loss": 0.3891,
+      "step": 8458
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.914286520811806e-06,
+      "loss": 0.4107,
+      "step": 8459
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.912058203792566e-06,
+      "loss": 0.3912,
+      "step": 8460
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.4521,
+      "step": 8461
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.907602078307984e-06,
+      "loss": 0.4331,
+      "step": 8462
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.905374270087215e-06,
+      "loss": 0.3978,
+      "step": 8463
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.903146631710486e-06,
+      "loss": 0.4314,
+      "step": 8464
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9009191633000636e-06,
+      "loss": 0.3675,
+      "step": 8465
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.898691864978195e-06,
+      "loss": 0.4308,
+      "step": 8466
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.896464736867125e-06,
+      "loss": 0.4296,
+      "step": 8467
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.894237779089086e-06,
+      "loss": 0.4127,
+      "step": 8468
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.892010991766298e-06,
+      "loss": 0.4416,
+      "step": 8469
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.8897843750209796e-06,
+      "loss": 0.4198,
+      "step": 8470
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.887557928975333e-06,
+      "loss": 0.3955,
+      "step": 8471
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.885331653751549e-06,
+      "loss": 0.3657,
+      "step": 8472
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.883105549471823e-06,
+      "loss": 0.4097,
+      "step": 8473
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.880879616258323e-06,
+      "loss": 0.3975,
+      "step": 8474
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.878653854233223e-06,
+      "loss": 0.4209,
+      "step": 8475
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.876428263518678e-06,
+      "loss": 0.4315,
+      "step": 8476
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8742028442368346e-06,
+      "loss": 0.4241,
+      "step": 8477
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.871977596509831e-06,
+      "loss": 0.4152,
+      "step": 8478
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.869752520459803e-06,
+      "loss": 0.4036,
+      "step": 8479
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.867527616208859e-06,
+      "loss": 0.4044,
+      "step": 8480
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.865302883879122e-06,
+      "loss": 0.3869,
+      "step": 8481
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.863078323592689e-06,
+      "loss": 0.4095,
+      "step": 8482
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8608539354716475e-06,
+      "loss": 0.4078,
+      "step": 8483
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.858629719638086e-06,
+      "loss": 0.4092,
+      "step": 8484
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.856405676214073e-06,
+      "loss": 0.4703,
+      "step": 8485
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.854181805321674e-06,
+      "loss": 0.4191,
+      "step": 8486
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.851958107082942e-06,
+      "loss": 0.4274,
+      "step": 8487
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.849734581619919e-06,
+      "loss": 0.4154,
+      "step": 8488
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.847511229054646e-06,
+      "loss": 0.4166,
+      "step": 8489
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.845288049509147e-06,
+      "loss": 0.4275,
+      "step": 8490
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.843065043105432e-06,
+      "loss": 0.424,
+      "step": 8491
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.840842209965515e-06,
+      "loss": 0.4091,
+      "step": 8492
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.83861955021139e-06,
+      "loss": 0.452,
+      "step": 8493
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.836397063965044e-06,
+      "loss": 0.4015,
+      "step": 8494
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8341747513484526e-06,
+      "loss": 0.3947,
+      "step": 8495
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.831952612483583e-06,
+      "loss": 0.3893,
+      "step": 8496
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.829730647492404e-06,
+      "loss": 0.4102,
+      "step": 8497
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.827508856496859e-06,
+      "loss": 0.3893,
+      "step": 8498
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.825287239618886e-06,
+      "loss": 0.4047,
+      "step": 8499
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.823065796980415e-06,
+      "loss": 0.4016,
+      "step": 8500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.820844528703371e-06,
+      "loss": 0.3778,
+      "step": 8501
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.818623434909661e-06,
+      "loss": 0.424,
+      "step": 8502
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.816402515721188e-06,
+      "loss": 0.4459,
+      "step": 8503
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.814181771259844e-06,
+      "loss": 0.396,
+      "step": 8504
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.811961201647508e-06,
+      "loss": 0.4016,
+      "step": 8505
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.80974080700606e-06,
+      "loss": 0.3855,
+      "step": 8506
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.807520587457359e-06,
+      "loss": 0.409,
+      "step": 8507
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.805300543123257e-06,
+      "loss": 0.4455,
+      "step": 8508
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.803080674125601e-06,
+      "loss": 0.3942,
+      "step": 8509
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.800860980586224e-06,
+      "loss": 0.4161,
+      "step": 8510
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.798641462626947e-06,
+      "loss": 0.4026,
+      "step": 8511
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.796422120369592e-06,
+      "loss": 0.4087,
+      "step": 8512
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.794202953935956e-06,
+      "loss": 0.417,
+      "step": 8513
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.791983963447845e-06,
+      "loss": 0.4403,
+      "step": 8514
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.789765149027039e-06,
+      "loss": 0.4539,
+      "step": 8515
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.787546510795313e-06,
+      "loss": 0.3991,
+      "step": 8516
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.785328048874438e-06,
+      "loss": 0.4023,
+      "step": 8517
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.783109763386168e-06,
+      "loss": 0.4118,
+      "step": 8518
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7808916544522505e-06,
+      "loss": 0.4079,
+      "step": 8519
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.778673722194425e-06,
+      "loss": 0.4018,
+      "step": 8520
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.776455966734414e-06,
+      "loss": 0.4407,
+      "step": 8521
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7742383881939445e-06,
+      "loss": 0.3868,
+      "step": 8522
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7720209866947206e-06,
+      "loss": 0.4343,
+      "step": 8523
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.769803762358442e-06,
+      "loss": 0.3978,
+      "step": 8524
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.767586715306796e-06,
+      "loss": 0.4259,
+      "step": 8525
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.765369845661464e-06,
+      "loss": 0.3643,
+      "step": 8526
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.763153153544115e-06,
+      "loss": 0.4212,
+      "step": 8527
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.760936639076408e-06,
+      "loss": 0.3856,
+      "step": 8528
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.758720302379993e-06,
+      "loss": 0.4162,
+      "step": 8529
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.756504143576511e-06,
+      "loss": 0.3813,
+      "step": 8530
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7542881627875965e-06,
+      "loss": 0.3905,
+      "step": 8531
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.752072360134866e-06,
+      "loss": 0.4152,
+      "step": 8532
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.749856735739929e-06,
+      "loss": 0.3846,
+      "step": 8533
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7476412897243925e-06,
+      "loss": 0.437,
+      "step": 8534
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.745426022209844e-06,
+      "loss": 0.4061,
+      "step": 8535
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7432109333178676e-06,
+      "loss": 0.3875,
+      "step": 8536
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.740996023170029e-06,
+      "loss": 0.4048,
+      "step": 8537
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.738781291887897e-06,
+      "loss": 0.4084,
+      "step": 8538
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7365667395930244e-06,
+      "loss": 0.4327,
+      "step": 8539
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.734352366406951e-06,
+      "loss": 0.4172,
+      "step": 8540
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7321381724512105e-06,
+      "loss": 0.3683,
+      "step": 8541
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.729924157847323e-06,
+      "loss": 0.4016,
+      "step": 8542
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.727710322716805e-06,
+      "loss": 0.4315,
+      "step": 8543
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.725496667181159e-06,
+      "loss": 0.4182,
+      "step": 8544
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.723283191361874e-06,
+      "loss": 0.4299,
+      "step": 8545
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7210698953804386e-06,
+      "loss": 0.3967,
+      "step": 8546
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.718856779358324e-06,
+      "loss": 0.462,
+      "step": 8547
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.716643843416997e-06,
+      "loss": 0.4255,
+      "step": 8548
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.714431087677906e-06,
+      "loss": 0.3991,
+      "step": 8549
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.712218512262498e-06,
+      "loss": 0.4005,
+      "step": 8550
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.71000611729221e-06,
+      "loss": 0.3687,
+      "step": 8551
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.70779390288846e-06,
+      "loss": 0.4162,
+      "step": 8552
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.705581869172663e-06,
+      "loss": 0.3649,
+      "step": 8553
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.703370016266228e-06,
+      "loss": 0.4052,
+      "step": 8554
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.701158344290546e-06,
+      "loss": 0.4446,
+      "step": 8555
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.698946853367002e-06,
+      "loss": 0.3789,
+      "step": 8556
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.6967355436169725e-06,
+      "loss": 0.4311,
+      "step": 8557
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.694524415161818e-06,
+      "loss": 0.4361,
+      "step": 8558
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.692313468122896e-06,
+      "loss": 0.4332,
+      "step": 8559
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.690102702621548e-06,
+      "loss": 0.3862,
+      "step": 8560
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.687892118779109e-06,
+      "loss": 0.4273,
+      "step": 8561
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.68568171671691e-06,
+      "loss": 0.3981,
+      "step": 8562
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.68347149655626e-06,
+      "loss": 0.418,
+      "step": 8563
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.681261458418464e-06,
+      "loss": 0.4431,
+      "step": 8564
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.679051602424821e-06,
+      "loss": 0.3989,
+      "step": 8565
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6768419286966116e-06,
+      "loss": 0.4581,
+      "step": 8566
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.674632437355108e-06,
+      "loss": 0.4082,
+      "step": 8567
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.672423128521582e-06,
+      "loss": 0.4007,
+      "step": 8568
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.670214002317279e-06,
+      "loss": 0.4168,
+      "step": 8569
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.668005058863453e-06,
+      "loss": 0.4237,
+      "step": 8570
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.665796298281337e-06,
+      "loss": 0.4001,
+      "step": 8571
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.66358772069215e-06,
+      "loss": 0.4391,
+      "step": 8572
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.661379326217112e-06,
+      "loss": 0.4252,
+      "step": 8573
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.659171114977427e-06,
+      "loss": 0.388,
+      "step": 8574
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6569630870942845e-06,
+      "loss": 0.4094,
+      "step": 8575
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.654755242688875e-06,
+      "loss": 0.3649,
+      "step": 8576
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.652547581882371e-06,
+      "loss": 0.4018,
+      "step": 8577
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.650340104795933e-06,
+      "loss": 0.4181,
+      "step": 8578
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.64813281155072e-06,
+      "loss": 0.4079,
+      "step": 8579
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.645925702267875e-06,
+      "loss": 0.414,
+      "step": 8580
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.643718777068532e-06,
+      "loss": 0.4128,
+      "step": 8581
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.641512036073814e-06,
+      "loss": 0.3747,
+      "step": 8582
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.639305479404836e-06,
+      "loss": 0.4047,
+      "step": 8583
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.637099107182699e-06,
+      "loss": 0.4117,
+      "step": 8584
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6348929195284995e-06,
+      "loss": 0.4258,
+      "step": 8585
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6326869165633155e-06,
+      "loss": 0.4434,
+      "step": 8586
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.630481098408228e-06,
+      "loss": 0.4377,
+      "step": 8587
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.628275465184297e-06,
+      "loss": 0.4158,
+      "step": 8588
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.626070017012573e-06,
+      "loss": 0.4248,
+      "step": 8589
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.623864754014104e-06,
+      "loss": 0.4052,
+      "step": 8590
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.621659676309919e-06,
+      "loss": 0.4254,
+      "step": 8591
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.619454784021038e-06,
+      "loss": 0.4025,
+      "step": 8592
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.61725007726848e-06,
+      "loss": 0.4148,
+      "step": 8593
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.615045556173238e-06,
+      "loss": 0.4246,
+      "step": 8594
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.612841220856314e-06,
+      "loss": 0.4385,
+      "step": 8595
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.610637071438686e-06,
+      "loss": 0.4272,
+      "step": 8596
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.608433108041321e-06,
+      "loss": 0.437,
+      "step": 8597
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.606229330785187e-06,
+      "loss": 0.3891,
+      "step": 8598
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.604025739791232e-06,
+      "loss": 0.4038,
+      "step": 8599
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.601822335180397e-06,
+      "loss": 0.4003,
+      "step": 8600
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.59961911707361e-06,
+      "loss": 0.3881,
+      "step": 8601
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.597416085591794e-06,
+      "loss": 0.4083,
+      "step": 8602
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.595213240855862e-06,
+      "loss": 0.3845,
+      "step": 8603
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.593010582986712e-06,
+      "loss": 0.4049,
+      "step": 8604
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.590808112105232e-06,
+      "loss": 0.3814,
+      "step": 8605
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5886058283323015e-06,
+      "loss": 0.385,
+      "step": 8606
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.586403731788792e-06,
+      "loss": 0.4074,
+      "step": 8607
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.58420182259556e-06,
+      "loss": 0.424,
+      "step": 8608
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.582000100873455e-06,
+      "loss": 0.4083,
+      "step": 8609
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.4194,
+      "step": 8610
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.577597220325966e-06,
+      "loss": 0.4114,
+      "step": 8611
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.575396061742232e-06,
+      "loss": 0.4023,
+      "step": 8612
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.573195091112915e-06,
+      "loss": 0.45,
+      "step": 8613
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.570994308558813e-06,
+      "loss": 0.4393,
+      "step": 8614
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.568793714200714e-06,
+      "loss": 0.4046,
+      "step": 8615
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5665933081593945e-06,
+      "loss": 0.3854,
+      "step": 8616
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.564393090555618e-06,
+      "loss": 0.405,
+      "step": 8617
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.562193061510141e-06,
+      "loss": 0.4389,
+      "step": 8618
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.55999322114371e-06,
+      "loss": 0.4433,
+      "step": 8619
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.557793569577062e-06,
+      "loss": 0.4178,
+      "step": 8620
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5555941069309205e-06,
+      "loss": 0.3863,
+      "step": 8621
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.553394833325996e-06,
+      "loss": 0.3951,
+      "step": 8622
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.551195748882997e-06,
+      "loss": 0.3888,
+      "step": 8623
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.548996853722616e-06,
+      "loss": 0.3815,
+      "step": 8624
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.546798147965536e-06,
+      "loss": 0.4063,
+      "step": 8625
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.544599631732426e-06,
+      "loss": 0.3956,
+      "step": 8626
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.542401305143953e-06,
+      "loss": 0.4001,
+      "step": 8627
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.540203168320769e-06,
+      "loss": 0.4526,
+      "step": 8628
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5380052213835155e-06,
+      "loss": 0.3983,
+      "step": 8629
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.535807464452821e-06,
+      "loss": 0.4006,
+      "step": 8630
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.533609897649307e-06,
+      "loss": 0.4357,
+      "step": 8631
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.531412521093586e-06,
+      "loss": 0.3915,
+      "step": 8632
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.529215334906257e-06,
+      "loss": 0.4235,
+      "step": 8633
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5270183392079035e-06,
+      "loss": 0.4254,
+      "step": 8634
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.524821534119113e-06,
+      "loss": 0.4385,
+      "step": 8635
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5226249197604496e-06,
+      "loss": 0.3972,
+      "step": 8636
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.520428496252474e-06,
+      "loss": 0.4345,
+      "step": 8637
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.518232263715731e-06,
+      "loss": 0.4106,
+      "step": 8638
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.516036222270759e-06,
+      "loss": 0.4011,
+      "step": 8639
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.513840372038082e-06,
+      "loss": 0.3837,
+      "step": 8640
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.511644713138221e-06,
+      "loss": 0.4315,
+      "step": 8641
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.509449245691679e-06,
+      "loss": 0.38,
+      "step": 8642
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.507253969818945e-06,
+      "loss": 0.467,
+      "step": 8643
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.5050588856405115e-06,
+      "loss": 0.3739,
+      "step": 8644
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.502863993276854e-06,
+      "loss": 0.4154,
+      "step": 8645
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.500669292848429e-06,
+      "loss": 0.4096,
+      "step": 8646
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.498474784475696e-06,
+      "loss": 0.446,
+      "step": 8647
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.49628046827909e-06,
+      "loss": 0.4763,
+      "step": 8648
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4940863443790504e-06,
+      "loss": 0.4217,
+      "step": 8649
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4918924128959945e-06,
+      "loss": 0.4044,
+      "step": 8650
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.489698673950329e-06,
+      "loss": 0.39,
+      "step": 8651
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.487505127662463e-06,
+      "loss": 0.4293,
+      "step": 8652
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.48531177415278e-06,
+      "loss": 0.4385,
+      "step": 8653
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.483118613541663e-06,
+      "loss": 0.4342,
+      "step": 8654
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.480925645949477e-06,
+      "loss": 0.4338,
+      "step": 8655
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4787328714965805e-06,
+      "loss": 0.3935,
+      "step": 8656
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.476540290303323e-06,
+      "loss": 0.3962,
+      "step": 8657
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.474347902490039e-06,
+      "loss": 0.4292,
+      "step": 8658
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.472155708177052e-06,
+      "loss": 0.4387,
+      "step": 8659
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.469963707484683e-06,
+      "loss": 0.4106,
+      "step": 8660
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.467771900533233e-06,
+      "loss": 0.4038,
+      "step": 8661
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4655802874430005e-06,
+      "loss": 0.417,
+      "step": 8662
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4633888683342634e-06,
+      "loss": 0.45,
+      "step": 8663
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4611976433272996e-06,
+      "loss": 0.4478,
+      "step": 8664
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.459006612542365e-06,
+      "loss": 0.3995,
+      "step": 8665
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.456815776099716e-06,
+      "loss": 0.425,
+      "step": 8666
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.45462513411959e-06,
+      "loss": 0.3893,
+      "step": 8667
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4524346867222236e-06,
+      "loss": 0.3995,
+      "step": 8668
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4502444340278315e-06,
+      "loss": 0.3896,
+      "step": 8669
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.448054376156621e-06,
+      "loss": 0.4203,
+      "step": 8670
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.445864513228795e-06,
+      "loss": 0.4195,
+      "step": 8671
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.443674845364539e-06,
+      "loss": 0.4328,
+      "step": 8672
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4414853726840265e-06,
+      "loss": 0.4004,
+      "step": 8673
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.439296095307429e-06,
+      "loss": 0.3998,
+      "step": 8674
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4371070133548955e-06,
+      "loss": 0.393,
+      "step": 8675
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.434918126946577e-06,
+      "loss": 0.3741,
+      "step": 8676
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.432729436202605e-06,
+      "loss": 0.3859,
+      "step": 8677
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4305409412431e-06,
+      "loss": 0.4059,
+      "step": 8678
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.428352642188178e-06,
+      "loss": 0.4167,
+      "step": 8679
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.42616453915794e-06,
+      "loss": 0.4001,
+      "step": 8680
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4239766322724725e-06,
+      "loss": 0.3842,
+      "step": 8681
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.421788921651862e-06,
+      "loss": 0.3498,
+      "step": 8682
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.419601407416172e-06,
+      "loss": 0.415,
+      "step": 8683
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.417414089685466e-06,
+      "loss": 0.4068,
+      "step": 8684
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.415226968579791e-06,
+      "loss": 0.4002,
+      "step": 8685
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.413040044219181e-06,
+      "loss": 0.3866,
+      "step": 8686
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4108533167236655e-06,
+      "loss": 0.4535,
+      "step": 8687
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.408666786213258e-06,
+      "loss": 0.3589,
+      "step": 8688
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.406480452807964e-06,
+      "loss": 0.3937,
+      "step": 8689
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4042943166277725e-06,
+      "loss": 0.3976,
+      "step": 8690
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.402108377792671e-06,
+      "loss": 0.4026,
+      "step": 8691
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.399922636422633e-06,
+      "loss": 0.4204,
+      "step": 8692
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.3977370926376185e-06,
+      "loss": 0.4143,
+      "step": 8693
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.3955517465575775e-06,
+      "loss": 0.4314,
+      "step": 8694
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.3933665983024465e-06,
+      "loss": 0.3767,
+      "step": 8695
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.391181647992159e-06,
+      "loss": 0.3836,
+      "step": 8696
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.388996895746631e-06,
+      "loss": 0.3929,
+      "step": 8697
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3868123416857665e-06,
+      "loss": 0.419,
+      "step": 8698
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.384627985929465e-06,
+      "loss": 0.3788,
+      "step": 8699
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.382443828597611e-06,
+      "loss": 0.4352,
+      "step": 8700
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.380259869810082e-06,
+      "loss": 0.4013,
+      "step": 8701
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.378076109686735e-06,
+      "loss": 0.3833,
+      "step": 8702
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3758925483474265e-06,
+      "loss": 0.4133,
+      "step": 8703
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.373709185911998e-06,
+      "loss": 0.4188,
+      "step": 8704
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.371526022500279e-06,
+      "loss": 0.3645,
+      "step": 8705
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3693430582320915e-06,
+      "loss": 0.4358,
+      "step": 8706
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3671602932272366e-06,
+      "loss": 0.4097,
+      "step": 8707
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.364977727605521e-06,
+      "loss": 0.4262,
+      "step": 8708
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.362795361486731e-06,
+      "loss": 0.3877,
+      "step": 8709
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.360613194990639e-06,
+      "loss": 0.3583,
+      "step": 8710
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.358431228237013e-06,
+      "loss": 0.3315,
+      "step": 8711
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.356249461345603e-06,
+      "loss": 0.3764,
+      "step": 8712
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3540678944361556e-06,
+      "loss": 0.4216,
+      "step": 8713
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.351886527628404e-06,
+      "loss": 0.4389,
+      "step": 8714
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.349705361042064e-06,
+      "loss": 0.4448,
+      "step": 8715
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.347524394796849e-06,
+      "loss": 0.3911,
+      "step": 8716
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.345343629012458e-06,
+      "loss": 0.4183,
+      "step": 8717
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.343163063808582e-06,
+      "loss": 0.4099,
+      "step": 8718
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.340982699304895e-06,
+      "loss": 0.3704,
+      "step": 8719
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.338802535621062e-06,
+      "loss": 0.4007,
+      "step": 8720
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.336622572876742e-06,
+      "loss": 0.3985,
+      "step": 8721
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.334442811191577e-06,
+      "loss": 0.4047,
+      "step": 8722
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3322632506852e-06,
+      "loss": 0.4365,
+      "step": 8723
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3300838914772275e-06,
+      "loss": 0.3978,
+      "step": 8724
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.32790473368728e-06,
+      "loss": 0.4254,
+      "step": 8725
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.325725777434955e-06,
+      "loss": 0.4392,
+      "step": 8726
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.323547022839838e-06,
+      "loss": 0.4321,
+      "step": 8727
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.321368470021509e-06,
+      "loss": 0.4187,
+      "step": 8728
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.319190119099533e-06,
+      "loss": 0.4055,
+      "step": 8729
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.31701197019347e-06,
+      "loss": 0.4042,
+      "step": 8730
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.314834023422859e-06,
+      "loss": 0.4139,
+      "step": 8731
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.312656278907232e-06,
+      "loss": 0.4053,
+      "step": 8732
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.310478736766122e-06,
+      "loss": 0.4348,
+      "step": 8733
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.308301397119028e-06,
+      "loss": 0.4212,
+      "step": 8734
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.306124260085459e-06,
+      "loss": 0.4028,
+      "step": 8735
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3039473257849e-06,
+      "loss": 0.4236,
+      "step": 8736
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.301770594336828e-06,
+      "loss": 0.4282,
+      "step": 8737
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.299594065860712e-06,
+      "loss": 0.4237,
+      "step": 8738
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.297417740476008e-06,
+      "loss": 0.4212,
+      "step": 8739
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.295241618302156e-06,
+      "loss": 0.4332,
+      "step": 8740
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.293065699458596e-06,
+      "loss": 0.3995,
+      "step": 8741
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.290889984064743e-06,
+      "loss": 0.395,
+      "step": 8742
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2887144722400165e-06,
+      "loss": 0.4327,
+      "step": 8743
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.286539164103809e-06,
+      "loss": 0.4398,
+      "step": 8744
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.284364059775511e-06,
+      "loss": 0.4305,
+      "step": 8745
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.282189159374502e-06,
+      "loss": 0.3982,
+      "step": 8746
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.280014463020148e-06,
+      "loss": 0.3962,
+      "step": 8747
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.277839970831796e-06,
+      "loss": 0.4175,
+      "step": 8748
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2756656829288035e-06,
+      "loss": 0.3935,
+      "step": 8749
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2734915994304925e-06,
+      "loss": 0.4235,
+      "step": 8750
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.271317720456191e-06,
+      "loss": 0.419,
+      "step": 8751
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.269144046125207e-06,
+      "loss": 0.4049,
+      "step": 8752
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.266970576556839e-06,
+      "loss": 0.4112,
+      "step": 8753
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.264797311870372e-06,
+      "loss": 0.3923,
+      "step": 8754
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.262624252185088e-06,
+      "loss": 0.4659,
+      "step": 8755
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.260451397620244e-06,
+      "loss": 0.4024,
+      "step": 8756
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.258278748295103e-06,
+      "loss": 0.3775,
+      "step": 8757
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.256106304328906e-06,
+      "loss": 0.3929,
+      "step": 8758
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.25393406584088e-06,
+      "loss": 0.4448,
+      "step": 8759
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.251762032950249e-06,
+      "loss": 0.4136,
+      "step": 8760
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.249590205776221e-06,
+      "loss": 0.4283,
+      "step": 8761
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.247418584437991e-06,
+      "loss": 0.4104,
+      "step": 8762
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.245247169054749e-06,
+      "loss": 0.3822,
+      "step": 8763
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2430759597456636e-06,
+      "loss": 0.4098,
+      "step": 8764
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.240904956629909e-06,
+      "loss": 0.3876,
+      "step": 8765
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.238734159826631e-06,
+      "loss": 0.37,
+      "step": 8766
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.236563569454969e-06,
+      "loss": 0.4013,
+      "step": 8767
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.234393185634058e-06,
+      "loss": 0.4146,
+      "step": 8768
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.232223008483013e-06,
+      "loss": 0.386,
+      "step": 8769
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.230053038120941e-06,
+      "loss": 0.3873,
+      "step": 8770
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.227883274666937e-06,
+      "loss": 0.4547,
+      "step": 8771
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.225713718240084e-06,
+      "loss": 0.3871,
+      "step": 8772
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.223544368959462e-06,
+      "loss": 0.4336,
+      "step": 8773
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.221375226944127e-06,
+      "loss": 0.3882,
+      "step": 8774
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.219206292313131e-06,
+      "loss": 0.3932,
+      "step": 8775
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.217037565185511e-06,
+      "loss": 0.454,
+      "step": 8776
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.214869045680296e-06,
+      "loss": 0.4051,
+      "step": 8777
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.212700733916501e-06,
+      "loss": 0.382,
+      "step": 8778
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.210532630013131e-06,
+      "loss": 0.4154,
+      "step": 8779
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.208364734089181e-06,
+      "loss": 0.3889,
+      "step": 8780
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.206197046263627e-06,
+      "loss": 0.464,
+      "step": 8781
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.204029566655447e-06,
+      "loss": 0.381,
+      "step": 8782
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.201862295383598e-06,
+      "loss": 0.396,
+      "step": 8783
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.199695232567025e-06,
+      "loss": 0.4204,
+      "step": 8784
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.197528378324664e-06,
+      "loss": 0.4003,
+      "step": 8785
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.195361732775445e-06,
+      "loss": 0.4086,
+      "step": 8786
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.193195296038275e-06,
+      "loss": 0.4019,
+      "step": 8787
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.191029068232055e-06,
+      "loss": 0.4147,
+      "step": 8788
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.188863049475677e-06,
+      "loss": 0.4095,
+      "step": 8789
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1866972398880245e-06,
+      "loss": 0.3819,
+      "step": 8790
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.184531639587963e-06,
+      "loss": 0.4197,
+      "step": 8791
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.182366248694345e-06,
+      "loss": 0.3741,
+      "step": 8792
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.180201067326015e-06,
+      "loss": 0.4255,
+      "step": 8793
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1780360956018094e-06,
+      "loss": 0.4201,
+      "step": 8794
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.175871333640546e-06,
+      "loss": 0.4298,
+      "step": 8795
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1737067815610355e-06,
+      "loss": 0.43,
+      "step": 8796
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.171542439482075e-06,
+      "loss": 0.391,
+      "step": 8797
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.169378307522456e-06,
+      "loss": 0.4339,
+      "step": 8798
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.167214385800952e-06,
+      "loss": 0.39,
+      "step": 8799
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.165050674436324e-06,
+      "loss": 0.4617,
+      "step": 8800
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.162887173547324e-06,
+      "loss": 0.4042,
+      "step": 8801
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1607238832526975e-06,
+      "loss": 0.414,
+      "step": 8802
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.158560803671168e-06,
+      "loss": 0.4596,
+      "step": 8803
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1563979349214545e-06,
+      "loss": 0.3974,
+      "step": 8804
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1542352771222635e-06,
+      "loss": 0.4235,
+      "step": 8805
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.15207283039229e-06,
+      "loss": 0.429,
+      "step": 8806
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.149910594850217e-06,
+      "loss": 0.4037,
+      "step": 8807
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.147748570614716e-06,
+      "loss": 0.3909,
+      "step": 8808
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1455867578044435e-06,
+      "loss": 0.4173,
+      "step": 8809
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.143425156538051e-06,
+      "loss": 0.388,
+      "step": 8810
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.141263766934173e-06,
+      "loss": 0.4003,
+      "step": 8811
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.139102589111435e-06,
+      "loss": 0.4019,
+      "step": 8812
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.136941623188446e-06,
+      "loss": 0.4545,
+      "step": 8813
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.134780869283812e-06,
+      "loss": 0.371,
+      "step": 8814
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.132620327516125e-06,
+      "loss": 0.436,
+      "step": 8815
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.130459998003959e-06,
+      "loss": 0.4393,
+      "step": 8816
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1282998808658835e-06,
+      "loss": 0.4243,
+      "step": 8817
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.126139976220449e-06,
+      "loss": 0.394,
+      "step": 8818
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.123980284186202e-06,
+      "loss": 0.416,
+      "step": 8819
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.121820804881675e-06,
+      "loss": 0.4189,
+      "step": 8820
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.119661538425381e-06,
+      "loss": 0.3962,
+      "step": 8821
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.117502484935838e-06,
+      "loss": 0.4268,
+      "step": 8822
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.115343644531536e-06,
+      "loss": 0.3909,
+      "step": 8823
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.113185017330964e-06,
+      "loss": 0.4268,
+      "step": 8824
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.111026603452592e-06,
+      "loss": 0.4231,
+      "step": 8825
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1088684030148814e-06,
+      "loss": 0.4151,
+      "step": 8826
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.106710416136286e-06,
+      "loss": 0.4183,
+      "step": 8827
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.104552642935238e-06,
+      "loss": 0.4236,
+      "step": 8828
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.102395083530164e-06,
+      "loss": 0.4377,
+      "step": 8829
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1002377380394835e-06,
+      "loss": 0.412,
+      "step": 8830
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.098080606581596e-06,
+      "loss": 0.3652,
+      "step": 8831
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.095923689274894e-06,
+      "loss": 0.4328,
+      "step": 8832
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.0937669862377565e-06,
+      "loss": 0.4183,
+      "step": 8833
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.09161049758855e-06,
+      "loss": 0.3968,
+      "step": 8834
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.08945422344563e-06,
+      "loss": 0.4376,
+      "step": 8835
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.087298163927342e-06,
+      "loss": 0.4189,
+      "step": 8836
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.085142319152013e-06,
+      "loss": 0.4108,
+      "step": 8837
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.082986689237972e-06,
+      "loss": 0.4005,
+      "step": 8838
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.080831274303523e-06,
+      "loss": 0.388,
+      "step": 8839
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.078676074466963e-06,
+      "loss": 0.3946,
+      "step": 8840
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.076521089846576e-06,
+      "loss": 0.4205,
+      "step": 8841
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.07436632056064e-06,
+      "loss": 0.3939,
+      "step": 8842
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.07221176672741e-06,
+      "loss": 0.4313,
+      "step": 8843
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.070057428465138e-06,
+      "loss": 0.3802,
+      "step": 8844
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.06790330589206e-06,
+      "loss": 0.4031,
+      "step": 8845
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.065749399126407e-06,
+      "loss": 0.4031,
+      "step": 8846
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.063595708286392e-06,
+      "loss": 0.4155,
+      "step": 8847
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.061442233490211e-06,
+      "loss": 0.4204,
+      "step": 8848
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.059288974856061e-06,
+      "loss": 0.3842,
+      "step": 8849
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.057135932502118e-06,
+      "loss": 0.421,
+      "step": 8850
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0549831065465505e-06,
+      "loss": 0.4224,
+      "step": 8851
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.052830497107507e-06,
+      "loss": 0.4389,
+      "step": 8852
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.050678104303138e-06,
+      "loss": 0.4051,
+      "step": 8853
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.048525928251566e-06,
+      "loss": 0.384,
+      "step": 8854
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.04637396907092e-06,
+      "loss": 0.4073,
+      "step": 8855
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.044222226879302e-06,
+      "loss": 0.3788,
+      "step": 8856
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.042070701794806e-06,
+      "loss": 0.4011,
+      "step": 8857
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.039919393935521e-06,
+      "loss": 0.4375,
+      "step": 8858
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.037768303419511e-06,
+      "loss": 0.3935,
+      "step": 8859
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.035617430364839e-06,
+      "loss": 0.4136,
+      "step": 8860
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.033466774889554e-06,
+      "loss": 0.472,
+      "step": 8861
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.031316337111686e-06,
+      "loss": 0.4164,
+      "step": 8862
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0291661171492654e-06,
+      "loss": 0.4308,
+      "step": 8863
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.027016115120303e-06,
+      "loss": 0.3666,
+      "step": 8864
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.024866331142793e-06,
+      "loss": 0.3885,
+      "step": 8865
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.022716765334731e-06,
+      "loss": 0.4425,
+      "step": 8866
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.020567417814087e-06,
+      "loss": 0.4288,
+      "step": 8867
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.018418288698824e-06,
+      "loss": 0.4261,
+      "step": 8868
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.016269378106901e-06,
+      "loss": 0.3581,
+      "step": 8869
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.014120686156246e-06,
+      "loss": 0.3893,
+      "step": 8870
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.011972212964801e-06,
+      "loss": 0.4239,
+      "step": 8871
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0098239586504725e-06,
+      "loss": 0.4422,
+      "step": 8872
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.007675923331166e-06,
+      "loss": 0.4115,
+      "step": 8873
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.005528107124776e-06,
+      "loss": 0.3857,
+      "step": 8874
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.003380510149179e-06,
+      "loss": 0.4366,
+      "step": 8875
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.001233132522246e-06,
+      "loss": 0.4073,
+      "step": 8876
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.999085974361827e-06,
+      "loss": 0.4517,
+      "step": 8877
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9969390357857674e-06,
+      "loss": 0.3615,
+      "step": 8878
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.994792316911906e-06,
+      "loss": 0.393,
+      "step": 8879
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.992645817858056e-06,
+      "loss": 0.4149,
+      "step": 8880
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.990499538742027e-06,
+      "loss": 0.3915,
+      "step": 8881
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.98835347968161e-06,
+      "loss": 0.4583,
+      "step": 8882
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.986207640794596e-06,
+      "loss": 0.3461,
+      "step": 8883
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.98406202219875e-06,
+      "loss": 0.3736,
+      "step": 8884
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.981916624011832e-06,
+      "loss": 0.3672,
+      "step": 8885
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.97977144635159e-06,
+      "loss": 0.4145,
+      "step": 8886
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.97762648933576e-06,
+      "loss": 0.3986,
+      "step": 8887
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.975481753082066e-06,
+      "loss": 0.3717,
+      "step": 8888
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.973337237708216e-06,
+      "loss": 0.3598,
+      "step": 8889
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.971192943331907e-06,
+      "loss": 0.3976,
+      "step": 8890
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.96904887007083e-06,
+      "loss": 0.4149,
+      "step": 8891
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.966905018042659e-06,
+      "loss": 0.3972,
+      "step": 8892
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.964761387365052e-06,
+      "loss": 0.4395,
+      "step": 8893
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.96261797815566e-06,
+      "loss": 0.4199,
+      "step": 8894
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.960474790532122e-06,
+      "loss": 0.3885,
+      "step": 8895
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.958331824612069e-06,
+      "loss": 0.3605,
+      "step": 8896
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.956189080513108e-06,
+      "loss": 0.427,
+      "step": 8897
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9540465583528415e-06,
+      "loss": 0.3495,
+      "step": 8898
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.95190425824886e-06,
+      "loss": 0.3974,
+      "step": 8899
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.949762180318739e-06,
+      "loss": 0.3999,
+      "step": 8900
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9476203246800455e-06,
+      "loss": 0.3957,
+      "step": 8901
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.945478691450326e-06,
+      "loss": 0.4272,
+      "step": 8902
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.943337280747131e-06,
+      "loss": 0.3942,
+      "step": 8903
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.941196092687982e-06,
+      "loss": 0.4653,
+      "step": 8904
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.939055127390397e-06,
+      "loss": 0.4494,
+      "step": 8905
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.93691438497188e-06,
+      "loss": 0.4387,
+      "step": 8906
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.93477386554992e-06,
+      "loss": 0.4305,
+      "step": 8907
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.932633569242e-06,
+      "loss": 0.4076,
+      "step": 8908
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.930493496165586e-06,
+      "loss": 0.4038,
+      "step": 8909
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9283536464381266e-06,
+      "loss": 0.4092,
+      "step": 8910
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9262140201770745e-06,
+      "loss": 0.3701,
+      "step": 8911
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.924074617499853e-06,
+      "loss": 0.3665,
+      "step": 8912
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.921935438523885e-06,
+      "loss": 0.3794,
+      "step": 8913
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.919796483366572e-06,
+      "loss": 0.3811,
+      "step": 8914
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.917657752145308e-06,
+      "loss": 0.386,
+      "step": 8915
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.915519244977477e-06,
+      "loss": 0.3878,
+      "step": 8916
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9133809619804464e-06,
+      "loss": 0.4332,
+      "step": 8917
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.911242903271572e-06,
+      "loss": 0.4397,
+      "step": 8918
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.909105068968195e-06,
+      "loss": 0.4337,
+      "step": 8919
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9069674591876535e-06,
+      "loss": 0.413,
+      "step": 8920
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9048300740472655e-06,
+      "loss": 0.3882,
+      "step": 8921
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9026929136643375e-06,
+      "loss": 0.4112,
+      "step": 8922
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.900555978156164e-06,
+      "loss": 0.4145,
+      "step": 8923
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8984192676400256e-06,
+      "loss": 0.443,
+      "step": 8924
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.896282782233197e-06,
+      "loss": 0.3727,
+      "step": 8925
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.894146522052934e-06,
+      "loss": 0.392,
+      "step": 8926
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.892010487216477e-06,
+      "loss": 0.4412,
+      "step": 8927
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.889874677841069e-06,
+      "loss": 0.4253,
+      "step": 8928
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.887739094043923e-06,
+      "loss": 0.4135,
+      "step": 8929
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8856037359422536e-06,
+      "loss": 0.4274,
+      "step": 8930
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.883468603653253e-06,
+      "loss": 0.4217,
+      "step": 8931
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.881333697294103e-06,
+      "loss": 0.4289,
+      "step": 8932
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.87919901698198e-06,
+      "loss": 0.4027,
+      "step": 8933
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.87706456283404e-06,
+      "loss": 0.4239,
+      "step": 8934
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8749303349674255e-06,
+      "loss": 0.4217,
+      "step": 8935
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.872796333499278e-06,
+      "loss": 0.4054,
+      "step": 8936
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8706625585467135e-06,
+      "loss": 0.4152,
+      "step": 8937
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.868529010226845e-06,
+      "loss": 0.397,
+      "step": 8938
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.866395688656769e-06,
+      "loss": 0.4084,
+      "step": 8939
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.864262593953567e-06,
+      "loss": 0.4098,
+      "step": 8940
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8621297262343094e-06,
+      "loss": 0.4138,
+      "step": 8941
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.859997085616062e-06,
+      "loss": 0.4193,
+      "step": 8942
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.857864672215863e-06,
+      "loss": 0.4233,
+      "step": 8943
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.855732486150753e-06,
+      "loss": 0.4213,
+      "step": 8944
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8536005275377576e-06,
+      "loss": 0.44,
+      "step": 8945
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.851468796493876e-06,
+      "loss": 0.4553,
+      "step": 8946
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.849337293136113e-06,
+      "loss": 0.4225,
+      "step": 8947
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.847206017581451e-06,
+      "loss": 0.4268,
+      "step": 8948
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.845074969946858e-06,
+      "loss": 0.3985,
+      "step": 8949
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.842944150349299e-06,
+      "loss": 0.4112,
+      "step": 8950
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.840813558905716e-06,
+      "loss": 0.4145,
+      "step": 8951
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.83868319573305e-06,
+      "loss": 0.4351,
+      "step": 8952
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.836553060948217e-06,
+      "loss": 0.4297,
+      "step": 8953
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.83442315466813e-06,
+      "loss": 0.4131,
+      "step": 8954
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.832293477009685e-06,
+      "loss": 0.3912,
+      "step": 8955
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.830164028089766e-06,
+      "loss": 0.4224,
+      "step": 8956
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.828034808025243e-06,
+      "loss": 0.3866,
+      "step": 8957
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8259058169329775e-06,
+      "loss": 0.3891,
+      "step": 8958
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.823777054929811e-06,
+      "loss": 0.3773,
+      "step": 8959
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.821648522132586e-06,
+      "loss": 0.3865,
+      "step": 8960
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.819520218658114e-06,
+      "loss": 0.4072,
+      "step": 8961
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.817392144623214e-06,
+      "loss": 0.3998,
+      "step": 8962
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.815264300144679e-06,
+      "loss": 0.3961,
+      "step": 8963
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.813136685339289e-06,
+      "loss": 0.3943,
+      "step": 8964
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8110093003238175e-06,
+      "loss": 0.3904,
+      "step": 8965
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.808882145215025e-06,
+      "loss": 0.4237,
+      "step": 8966
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.806755220129648e-06,
+      "loss": 0.4025,
+      "step": 8967
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8046285251844324e-06,
+      "loss": 0.3828,
+      "step": 8968
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.802502060496089e-06,
+      "loss": 0.3826,
+      "step": 8969
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.800375826181335e-06,
+      "loss": 0.4159,
+      "step": 8970
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.798249822356859e-06,
+      "loss": 0.4378,
+      "step": 8971
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.796124049139347e-06,
+      "loss": 0.4232,
+      "step": 8972
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.793998506645464e-06,
+      "loss": 0.4032,
+      "step": 8973
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7918731949918726e-06,
+      "loss": 0.4234,
+      "step": 8974
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.789748114295212e-06,
+      "loss": 0.3846,
+      "step": 8975
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.78762326467212e-06,
+      "loss": 0.4017,
+      "step": 8976
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.785498646239213e-06,
+      "loss": 0.4013,
+      "step": 8977
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.783374259113096e-06,
+      "loss": 0.3934,
+      "step": 8978
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.781250103410367e-06,
+      "loss": 0.4011,
+      "step": 8979
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.779126179247607e-06,
+      "loss": 0.4279,
+      "step": 8980
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.777002486741381e-06,
+      "loss": 0.3771,
+      "step": 8981
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.774879026008246e-06,
+      "loss": 0.414,
+      "step": 8982
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.772755797164743e-06,
+      "loss": 0.4196,
+      "step": 8983
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.770632800327408e-06,
+      "loss": 0.4337,
+      "step": 8984
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7685100356127555e-06,
+      "loss": 0.3896,
+      "step": 8985
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.766387503137286e-06,
+      "loss": 0.4026,
+      "step": 8986
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.764265203017501e-06,
+      "loss": 0.4044,
+      "step": 8987
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.762143135369873e-06,
+      "loss": 0.3994,
+      "step": 8988
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.760021300310871e-06,
+      "loss": 0.3741,
+      "step": 8989
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.757899697956948e-06,
+      "loss": 0.4217,
+      "step": 8990
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.755778328424546e-06,
+      "loss": 0.4293,
+      "step": 8991
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.753657191830086e-06,
+      "loss": 0.3711,
+      "step": 8992
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.751536288289997e-06,
+      "loss": 0.3704,
+      "step": 8993
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.749415617920673e-06,
+      "loss": 0.4135,
+      "step": 8994
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.747295180838503e-06,
+      "loss": 0.3872,
+      "step": 8995
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.74517497715987e-06,
+      "loss": 0.3945,
+      "step": 8996
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.743055007001135e-06,
+      "loss": 0.4316,
+      "step": 8997
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.740935270478649e-06,
+      "loss": 0.399,
+      "step": 8998
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7388157677087516e-06,
+      "loss": 0.4201,
+      "step": 8999
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.736696498807766e-06,
+      "loss": 0.367,
+      "step": 9000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.734577463892008e-06,
+      "loss": 0.4008,
+      "step": 9001
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.73245866307778e-06,
+      "loss": 0.4089,
+      "step": 9002
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.730340096481362e-06,
+      "loss": 0.4408,
+      "step": 9003
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.728221764219035e-06,
+      "loss": 0.4035,
+      "step": 9004
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.726103666407062e-06,
+      "loss": 0.3609,
+      "step": 9005
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.723985803161684e-06,
+      "loss": 0.4027,
+      "step": 9006
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7218681745991455e-06,
+      "loss": 0.4548,
+      "step": 9007
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.719750780835658e-06,
+      "loss": 0.422,
+      "step": 9008
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.717633621987444e-06,
+      "loss": 0.3943,
+      "step": 9009
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.715516698170696e-06,
+      "loss": 0.4178,
+      "step": 9010
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7134000095015915e-06,
+      "loss": 0.4073,
+      "step": 9011
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.711283556096314e-06,
+      "loss": 0.4601,
+      "step": 9012
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7091673380710155e-06,
+      "loss": 0.4086,
+      "step": 9013
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.707051355541842e-06,
+      "loss": 0.4057,
+      "step": 9014
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.704935608624926e-06,
+      "loss": 0.4078,
+      "step": 9015
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.702820097436383e-06,
+      "loss": 0.4226,
+      "step": 9016
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7007048220923285e-06,
+      "loss": 0.4149,
+      "step": 9017
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.698589782708852e-06,
+      "loss": 0.4437,
+      "step": 9018
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.696474979402036e-06,
+      "loss": 0.3875,
+      "step": 9019
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.69436041228794e-06,
+      "loss": 0.3973,
+      "step": 9020
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.692246081482631e-06,
+      "loss": 0.4088,
+      "step": 9021
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.690131987102146e-06,
+      "loss": 0.4445,
+      "step": 9022
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.688018129262514e-06,
+      "loss": 0.4213,
+      "step": 9023
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6859045080797455e-06,
+      "loss": 0.3729,
+      "step": 9024
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.683791123669854e-06,
+      "loss": 0.4075,
+      "step": 9025
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.681677976148824e-06,
+      "loss": 0.4238,
+      "step": 9026
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.679565065632633e-06,
+      "loss": 0.4009,
+      "step": 9027
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.677452392237239e-06,
+      "loss": 0.4165,
+      "step": 9028
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.675339956078605e-06,
+      "loss": 0.422,
+      "step": 9029
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6732277572726615e-06,
+      "loss": 0.4201,
+      "step": 9030
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.671115795935336e-06,
+      "loss": 0.3931,
+      "step": 9031
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6690040721825345e-06,
+      "loss": 0.3987,
+      "step": 9032
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.666892586130166e-06,
+      "loss": 0.4123,
+      "step": 9033
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.664781337894109e-06,
+      "loss": 0.4486,
+      "step": 9034
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.662670327590239e-06,
+      "loss": 0.3991,
+      "step": 9035
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.660559555334415e-06,
+      "loss": 0.4083,
+      "step": 9036
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.658449021242481e-06,
+      "loss": 0.3813,
+      "step": 9037
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.656338725430276e-06,
+      "loss": 0.4361,
+      "step": 9038
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.654228668013617e-06,
+      "loss": 0.4295,
+      "step": 9039
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.65211884910831e-06,
+      "loss": 0.4092,
+      "step": 9040
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.650009268830155e-06,
+      "loss": 0.4656,
+      "step": 9041
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.647899927294931e-06,
+      "loss": 0.4035,
+      "step": 9042
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.645790824618403e-06,
+      "loss": 0.4325,
+      "step": 9043
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.643681960916329e-06,
+      "loss": 0.4464,
+      "step": 9044
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.641573336304447e-06,
+      "loss": 0.4054,
+      "step": 9045
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.639464950898491e-06,
+      "loss": 0.4197,
+      "step": 9046
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.637356804814176e-06,
+      "loss": 0.428,
+      "step": 9047
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6352488981671975e-06,
+      "loss": 0.3991,
+      "step": 9048
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.633141231073256e-06,
+      "loss": 0.404,
+      "step": 9049
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.631033803648021e-06,
+      "loss": 0.402,
+      "step": 9050
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.628926616007156e-06,
+      "loss": 0.4259,
+      "step": 9051
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.626819668266314e-06,
+      "loss": 0.4013,
+      "step": 9052
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.624712960541128e-06,
+      "loss": 0.3983,
+      "step": 9053
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.622606492947218e-06,
+      "loss": 0.3888,
+      "step": 9054
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.620500265600206e-06,
+      "loss": 0.4295,
+      "step": 9055
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.618394278615681e-06,
+      "loss": 0.4014,
+      "step": 9056
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.616288532109225e-06,
+      "loss": 0.441,
+      "step": 9057
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.614183026196417e-06,
+      "loss": 0.4373,
+      "step": 9058
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6120777609928105e-06,
+      "loss": 0.4375,
+      "step": 9059
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.609972736613949e-06,
+      "loss": 0.4335,
+      "step": 9060
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.607867953175364e-06,
+      "loss": 0.4299,
+      "step": 9061
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.605763410792571e-06,
+      "loss": 0.4542,
+      "step": 9062
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.603659109581081e-06,
+      "loss": 0.419,
+      "step": 9063
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.601555049656382e-06,
+      "loss": 0.4205,
+      "step": 9064
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.599451231133947e-06,
+      "loss": 0.3984,
+      "step": 9065
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.597347654129253e-06,
+      "loss": 0.423,
+      "step": 9066
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.595244318757744e-06,
+      "loss": 0.416,
+      "step": 9067
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.59314122513486e-06,
+      "loss": 0.3899,
+      "step": 9068
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5910383733760255e-06,
+      "loss": 0.4277,
+      "step": 9069
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.58893576359665e-06,
+      "loss": 0.3825,
+      "step": 9070
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.586833395912138e-06,
+      "loss": 0.4217,
+      "step": 9071
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.584731270437873e-06,
+      "loss": 0.4613,
+      "step": 9072
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.58262938728922e-06,
+      "loss": 0.4003,
+      "step": 9073
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.58052774658155e-06,
+      "loss": 0.4377,
+      "step": 9074
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.578426348430201e-06,
+      "loss": 0.4126,
+      "step": 9075
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.576325192950508e-06,
+      "loss": 0.4503,
+      "step": 9076
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.574224280257788e-06,
+      "loss": 0.3573,
+      "step": 9077
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.572123610467347e-06,
+      "loss": 0.3703,
+      "step": 9078
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.570023183694473e-06,
+      "loss": 0.4207,
+      "step": 9079
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.567923000054454e-06,
+      "loss": 0.4126,
+      "step": 9080
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.565823059662545e-06,
+      "loss": 0.407,
+      "step": 9081
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.563723362634008e-06,
+      "loss": 0.4105,
+      "step": 9082
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5616239090840795e-06,
+      "loss": 0.3746,
+      "step": 9083
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.559524699127981e-06,
+      "loss": 0.4023,
+      "step": 9084
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.557425732880928e-06,
+      "loss": 0.3587,
+      "step": 9085
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.555327010458118e-06,
+      "loss": 0.3915,
+      "step": 9086
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5532285319747315e-06,
+      "loss": 0.4135,
+      "step": 9087
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.55113029754595e-06,
+      "loss": 0.4093,
+      "step": 9088
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.549032307286922e-06,
+      "loss": 0.4235,
+      "step": 9089
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.546934561312802e-06,
+      "loss": 0.4061,
+      "step": 9090
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.544837059738719e-06,
+      "loss": 0.4081,
+      "step": 9091
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.542739802679789e-06,
+      "loss": 0.3762,
+      "step": 9092
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5406427902511185e-06,
+      "loss": 0.4099,
+      "step": 9093
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.538546022567799e-06,
+      "loss": 0.4444,
+      "step": 9094
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.536449499744907e-06,
+      "loss": 0.4059,
+      "step": 9095
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.534353221897504e-06,
+      "loss": 0.4217,
+      "step": 9096
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.532257189140644e-06,
+      "loss": 0.4324,
+      "step": 9097
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5301614015893715e-06,
+      "loss": 0.4172,
+      "step": 9098
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.528065859358705e-06,
+      "loss": 0.4078,
+      "step": 9099
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.525970562563656e-06,
+      "loss": 0.4361,
+      "step": 9100
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.523875511319221e-06,
+      "loss": 0.4301,
+      "step": 9101
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.521780705740383e-06,
+      "loss": 0.4421,
+      "step": 9102
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.519686145942114e-06,
+      "loss": 0.4503,
+      "step": 9103
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.517591832039367e-06,
+      "loss": 0.4193,
+      "step": 9104
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.515497764147093e-06,
+      "loss": 0.4293,
+      "step": 9105
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.513403942380213e-06,
+      "loss": 0.4174,
+      "step": 9106
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.511310366853653e-06,
+      "loss": 0.4213,
+      "step": 9107
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.509217037682308e-06,
+      "loss": 0.4388,
+      "step": 9108
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.507123954981073e-06,
+      "loss": 0.3966,
+      "step": 9109
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.505031118864821e-06,
+      "loss": 0.4177,
+      "step": 9110
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.502938529448414e-06,
+      "loss": 0.4304,
+      "step": 9111
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.5008461868467004e-06,
+      "loss": 0.4027,
+      "step": 9112
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.498754091174512e-06,
+      "loss": 0.4115,
+      "step": 9113
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4966622425466745e-06,
+      "loss": 0.4373,
+      "step": 9114
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4945706410779994e-06,
+      "loss": 0.3903,
+      "step": 9115
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.492479286883279e-06,
+      "loss": 0.4053,
+      "step": 9116
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4903881800772925e-06,
+      "loss": 0.4397,
+      "step": 9117
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4882973207748065e-06,
+      "loss": 0.4177,
+      "step": 9118
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.486206709090578e-06,
+      "loss": 0.4017,
+      "step": 9119
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.484116345139345e-06,
+      "loss": 0.3867,
+      "step": 9120
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.482026229035828e-06,
+      "loss": 0.3916,
+      "step": 9121
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4799363608947484e-06,
+      "loss": 0.4124,
+      "step": 9122
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.477846740830808e-06,
+      "loss": 0.3941,
+      "step": 9123
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.475757368958688e-06,
+      "loss": 0.4167,
+      "step": 9124
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.473668245393058e-06,
+      "loss": 0.4387,
+      "step": 9125
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.471579370248581e-06,
+      "loss": 0.3995,
+      "step": 9126
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4694907436399e-06,
+      "loss": 0.3955,
+      "step": 9127
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.467402365681646e-06,
+      "loss": 0.4454,
+      "step": 9128
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.465314236488433e-06,
+      "loss": 0.4103,
+      "step": 9129
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4632263561748724e-06,
+      "loss": 0.4099,
+      "step": 9130
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.461138724855546e-06,
+      "loss": 0.4052,
+      "step": 9131
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.459051342645041e-06,
+      "loss": 0.4232,
+      "step": 9132
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.456964209657913e-06,
+      "loss": 0.3862,
+      "step": 9133
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.454877326008714e-06,
+      "loss": 0.3894,
+      "step": 9134
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4527906918119776e-06,
+      "loss": 0.3977,
+      "step": 9135
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.450704307182229e-06,
+      "loss": 0.4148,
+      "step": 9136
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.44861817223397e-06,
+      "loss": 0.4249,
+      "step": 9137
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.446532287081698e-06,
+      "loss": 0.4128,
+      "step": 9138
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.444446651839894e-06,
+      "loss": 0.4193,
+      "step": 9139
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.442361266623029e-06,
+      "loss": 0.4008,
+      "step": 9140
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.440276131545555e-06,
+      "loss": 0.3692,
+      "step": 9141
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.438191246721908e-06,
+      "loss": 0.397,
+      "step": 9142
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4361066122665166e-06,
+      "loss": 0.4116,
+      "step": 9143
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.434022228293791e-06,
+      "loss": 0.3994,
+      "step": 9144
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.431938094918132e-06,
+      "loss": 0.4238,
+      "step": 9145
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.429854212253917e-06,
+      "loss": 0.4372,
+      "step": 9146
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.42777058041553e-06,
+      "loss": 0.4237,
+      "step": 9147
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.425687199517315e-06,
+      "loss": 0.4219,
+      "step": 9148
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4236040696736244e-06,
+      "loss": 0.4479,
+      "step": 9149
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.421521190998785e-06,
+      "loss": 0.4177,
+      "step": 9150
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.419438563607112e-06,
+      "loss": 0.4115,
+      "step": 9151
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4173561876129075e-06,
+      "loss": 0.3649,
+      "step": 9152
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.41527406313046e-06,
+      "loss": 0.4359,
+      "step": 9153
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4131921902740405e-06,
+      "loss": 0.3882,
+      "step": 9154
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.411110569157916e-06,
+      "loss": 0.4529,
+      "step": 9155
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.409029199896325e-06,
+      "loss": 0.4075,
+      "step": 9156
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4069480826035105e-06,
+      "loss": 0.4053,
+      "step": 9157
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.404867217393685e-06,
+      "loss": 0.3854,
+      "step": 9158
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.402786604381056e-06,
+      "loss": 0.4133,
+      "step": 9159
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.400706243679814e-06,
+      "loss": 0.4376,
+      "step": 9160
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.398626135404137e-06,
+      "loss": 0.409,
+      "step": 9161
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.396546279668184e-06,
+      "loss": 0.4485,
+      "step": 9162
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.394466676586114e-06,
+      "loss": 0.4096,
+      "step": 9163
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.392387326272058e-06,
+      "loss": 0.4291,
+      "step": 9164
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.390308228840135e-06,
+      "loss": 0.3823,
+      "step": 9165
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.38822938440446e-06,
+      "loss": 0.4251,
+      "step": 9166
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.386150793079124e-06,
+      "loss": 0.4172,
+      "step": 9167
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.384072454978206e-06,
+      "loss": 0.4325,
+      "step": 9168
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.381994370215777e-06,
+      "loss": 0.4189,
+      "step": 9169
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.37991653890588e-06,
+      "loss": 0.4124,
+      "step": 9170
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3778389611625666e-06,
+      "loss": 0.4148,
+      "step": 9171
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.375761637099854e-06,
+      "loss": 0.4173,
+      "step": 9172
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.373684566831752e-06,
+      "loss": 0.4031,
+      "step": 9173
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.371607750472263e-06,
+      "loss": 0.4193,
+      "step": 9174
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.369531188135369e-06,
+      "loss": 0.4001,
+      "step": 9175
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.367454879935036e-06,
+      "loss": 0.4426,
+      "step": 9176
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3653788259852205e-06,
+      "loss": 0.41,
+      "step": 9177
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.363303026399859e-06,
+      "loss": 0.4249,
+      "step": 9178
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.36122748129289e-06,
+      "loss": 0.4221,
+      "step": 9179
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.359152190778218e-06,
+      "loss": 0.3874,
+      "step": 9180
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.357077154969742e-06,
+      "loss": 0.446,
+      "step": 9181
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3550023739813525e-06,
+      "loss": 0.4115,
+      "step": 9182
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.35292784792692e-06,
+      "loss": 0.3931,
+      "step": 9183
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.350853576920299e-06,
+      "loss": 0.4191,
+      "step": 9184
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3487795610753345e-06,
+      "loss": 0.4094,
+      "step": 9185
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.346705800505851e-06,
+      "loss": 0.4397,
+      "step": 9186
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.344632295325673e-06,
+      "loss": 0.3492,
+      "step": 9187
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.342559045648595e-06,
+      "loss": 0.4073,
+      "step": 9188
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.340486051588409e-06,
+      "loss": 0.4075,
+      "step": 9189
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.338413313258879e-06,
+      "loss": 0.4092,
+      "step": 9190
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.336340830773777e-06,
+      "loss": 0.3977,
+      "step": 9191
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.334268604246841e-06,
+      "loss": 0.4112,
+      "step": 9192
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.332196633791802e-06,
+      "loss": 0.4095,
+      "step": 9193
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.330124919522379e-06,
+      "loss": 0.4136,
+      "step": 9194
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.328053461552272e-06,
+      "loss": 0.4363,
+      "step": 9195
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.325982259995175e-06,
+      "loss": 0.414,
+      "step": 9196
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.323911314964759e-06,
+      "loss": 0.445,
+      "step": 9197
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.321840626574683e-06,
+      "loss": 0.4421,
+      "step": 9198
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3197701949386e-06,
+      "loss": 0.4134,
+      "step": 9199
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.31770002017014e-06,
+      "loss": 0.4141,
+      "step": 9200
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.315630102382921e-06,
+      "loss": 0.4424,
+      "step": 9201
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.313560441690547e-06,
+      "loss": 0.4389,
+      "step": 9202
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3114910382066055e-06,
+      "loss": 0.3698,
+      "step": 9203
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.309421892044679e-06,
+      "loss": 0.3742,
+      "step": 9204
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.307353003318325e-06,
+      "loss": 0.4273,
+      "step": 9205
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.305284372141095e-06,
+      "loss": 0.3887,
+      "step": 9206
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3032159986265165e-06,
+      "loss": 0.4082,
+      "step": 9207
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3011478828881155e-06,
+      "loss": 0.3991,
+      "step": 9208
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.299080025039398e-06,
+      "loss": 0.3945,
+      "step": 9209
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.29701242519385e-06,
+      "loss": 0.3981,
+      "step": 9210
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.29494508346495e-06,
+      "loss": 0.386,
+      "step": 9211
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.292877999966165e-06,
+      "loss": 0.4278,
+      "step": 9212
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.290811174810941e-06,
+      "loss": 0.4356,
+      "step": 9213
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2887446081127146e-06,
+      "loss": 0.4148,
+      "step": 9214
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2866782999849e-06,
+      "loss": 0.4177,
+      "step": 9215
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.284612250540912e-06,
+      "loss": 0.3897,
+      "step": 9216
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.28254645989414e-06,
+      "loss": 0.3977,
+      "step": 9217
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.280480928157961e-06,
+      "loss": 0.397,
+      "step": 9218
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.278415655445734e-06,
+      "loss": 0.4204,
+      "step": 9219
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.276350641870818e-06,
+      "loss": 0.4442,
+      "step": 9220
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.274285887546542e-06,
+      "loss": 0.3961,
+      "step": 9221
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.27222139258623e-06,
+      "loss": 0.357,
+      "step": 9222
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.270157157103186e-06,
+      "loss": 0.4139,
+      "step": 9223
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2680931812107004e-06,
+      "loss": 0.3939,
+      "step": 9224
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.26602946502206e-06,
+      "loss": 0.3719,
+      "step": 9225
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.263966008650523e-06,
+      "loss": 0.4062,
+      "step": 9226
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2619028122093366e-06,
+      "loss": 0.4233,
+      "step": 9227
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.259839875811743e-06,
+      "loss": 0.396,
+      "step": 9228
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.257777199570961e-06,
+      "loss": 0.4039,
+      "step": 9229
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.255714783600198e-06,
+      "loss": 0.4582,
+      "step": 9230
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.253652628012644e-06,
+      "loss": 0.4363,
+      "step": 9231
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.251590732921476e-06,
+      "loss": 0.4347,
+      "step": 9232
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.249529098439864e-06,
+      "loss": 0.386,
+      "step": 9233
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.247467724680956e-06,
+      "loss": 0.3779,
+      "step": 9234
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.245406611757882e-06,
+      "loss": 0.3971,
+      "step": 9235
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.243345759783771e-06,
+      "loss": 0.3721,
+      "step": 9236
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.241285168871728e-06,
+      "loss": 0.4038,
+      "step": 9237
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.239224839134843e-06,
+      "loss": 0.4304,
+      "step": 9238
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.237164770686196e-06,
+      "loss": 0.4087,
+      "step": 9239
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.235104963638846e-06,
+      "loss": 0.4044,
+      "step": 9240
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.233045418105851e-06,
+      "loss": 0.4466,
+      "step": 9241
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.23098613420024e-06,
+      "loss": 0.3841,
+      "step": 9242
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.228927112035033e-06,
+      "loss": 0.4353,
+      "step": 9243
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.226868351723244e-06,
+      "loss": 0.4097,
+      "step": 9244
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.224809853377859e-06,
+      "loss": 0.4133,
+      "step": 9245
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.222751617111858e-06,
+      "loss": 0.4017,
+      "step": 9246
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.220693643038202e-06,
+      "loss": 0.3632,
+      "step": 9247
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.218635931269841e-06,
+      "loss": 0.4094,
+      "step": 9248
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.216578481919706e-06,
+      "loss": 0.3828,
+      "step": 9249
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.214521295100725e-06,
+      "loss": 0.4073,
+      "step": 9250
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.212464370925796e-06,
+      "loss": 0.4131,
+      "step": 9251
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.210407709507817e-06,
+      "loss": 0.4009,
+      "step": 9252
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.208351310959662e-06,
+      "loss": 0.3735,
+      "step": 9253
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.206295175394193e-06,
+      "loss": 0.4018,
+      "step": 9254
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.204239302924257e-06,
+      "loss": 0.4392,
+      "step": 9255
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.20218369366269e-06,
+      "loss": 0.3937,
+      "step": 9256
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.2001283477223065e-06,
+      "loss": 0.4204,
+      "step": 9257
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.198073265215917e-06,
+      "loss": 0.3901,
+      "step": 9258
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1960184462563055e-06,
+      "loss": 0.432,
+      "step": 9259
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.193963890956255e-06,
+      "loss": 0.4318,
+      "step": 9260
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.191909599428524e-06,
+      "loss": 0.4235,
+      "step": 9261
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.189855571785859e-06,
+      "loss": 0.3226,
+      "step": 9262
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.187801808140991e-06,
+      "loss": 0.3964,
+      "step": 9263
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1857483086066396e-06,
+      "loss": 0.3988,
+      "step": 9264
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.183695073295507e-06,
+      "loss": 0.4317,
+      "step": 9265
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.181642102320278e-06,
+      "loss": 0.4395,
+      "step": 9266
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.179589395793636e-06,
+      "loss": 0.4192,
+      "step": 9267
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.17753695382823e-06,
+      "loss": 0.4327,
+      "step": 9268
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.175484776536719e-06,
+      "loss": 0.4266,
+      "step": 9269
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.173432864031723e-06,
+      "loss": 0.3882,
+      "step": 9270
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1713812164258625e-06,
+      "loss": 0.417,
+      "step": 9271
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.169329833831738e-06,
+      "loss": 0.4154,
+      "step": 9272
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.167278716361936e-06,
+      "loss": 0.3994,
+      "step": 9273
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.165227864129028e-06,
+      "loss": 0.441,
+      "step": 9274
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1631772772455765e-06,
+      "loss": 0.4056,
+      "step": 9275
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.161126955824118e-06,
+      "loss": 0.4059,
+      "step": 9276
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.159076899977191e-06,
+      "loss": 0.3917,
+      "step": 9277
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.157027109817303e-06,
+      "loss": 0.3937,
+      "step": 9278
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1549775854569574e-06,
+      "loss": 0.393,
+      "step": 9279
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.152928327008635e-06,
+      "loss": 0.4243,
+      "step": 9280
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.150879334584811e-06,
+      "loss": 0.3905,
+      "step": 9281
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.148830608297938e-06,
+      "loss": 0.4401,
+      "step": 9282
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.146782148260454e-06,
+      "loss": 0.4218,
+      "step": 9283
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.144733954584792e-06,
+      "loss": 0.4166,
+      "step": 9284
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1426860273833656e-06,
+      "loss": 0.4369,
+      "step": 9285
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.140638366768569e-06,
+      "loss": 0.4062,
+      "step": 9286
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1385909728527835e-06,
+      "loss": 0.4296,
+      "step": 9287
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.136543845748382e-06,
+      "loss": 0.4137,
+      "step": 9288
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.134496985567714e-06,
+      "loss": 0.3937,
+      "step": 9289
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.132450392423118e-06,
+      "loss": 0.4243,
+      "step": 9290
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1304040664269175e-06,
+      "loss": 0.4235,
+      "step": 9291
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1283580076914285e-06,
+      "loss": 0.3739,
+      "step": 9292
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.126312216328938e-06,
+      "loss": 0.3962,
+      "step": 9293
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.124266692451734e-06,
+      "loss": 0.4049,
+      "step": 9294
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.122221436172079e-06,
+      "loss": 0.4475,
+      "step": 9295
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.120176447602222e-06,
+      "loss": 0.4201,
+      "step": 9296
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.118131726854401e-06,
+      "loss": 0.3694,
+      "step": 9297
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.116087274040837e-06,
+      "loss": 0.4431,
+      "step": 9298
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1140430892737335e-06,
+      "loss": 0.3972,
+      "step": 9299
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.11199917266529e-06,
+      "loss": 0.4201,
+      "step": 9300
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.109955524327676e-06,
+      "loss": 0.4193,
+      "step": 9301
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.107912144373062e-06,
+      "loss": 0.4482,
+      "step": 9302
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.105869032913591e-06,
+      "loss": 0.4511,
+      "step": 9303
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1038261900613985e-06,
+      "loss": 0.4271,
+      "step": 9304
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.1017836159286e-06,
+      "loss": 0.4127,
+      "step": 9305
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.099741310627302e-06,
+      "loss": 0.366,
+      "step": 9306
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.097699274269593e-06,
+      "loss": 0.4212,
+      "step": 9307
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.095657506967542e-06,
+      "loss": 0.368,
+      "step": 9308
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.093616008833213e-06,
+      "loss": 0.4292,
+      "step": 9309
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.091574779978655e-06,
+      "loss": 0.4072,
+      "step": 9310
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.089533820515893e-06,
+      "loss": 0.3859,
+      "step": 9311
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.087493130556944e-06,
+      "loss": 0.4083,
+      "step": 9312
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.085452710213805e-06,
+      "loss": 0.3772,
+      "step": 9313
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.083412559598465e-06,
+      "loss": 0.4042,
+      "step": 9314
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.081372678822892e-06,
+      "loss": 0.3861,
+      "step": 9315
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.07933306799904e-06,
+      "loss": 0.4007,
+      "step": 9316
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.077293727238857e-06,
+      "loss": 0.4672,
+      "step": 9317
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.075254656654261e-06,
+      "loss": 0.4017,
+      "step": 9318
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.073215856357172e-06,
+      "loss": 0.3764,
+      "step": 9319
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.071177326459483e-06,
+      "loss": 0.3963,
+      "step": 9320
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.069139067073074e-06,
+      "loss": 0.4314,
+      "step": 9321
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.067101078309815e-06,
+      "loss": 0.4048,
+      "step": 9322
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.065063360281555e-06,
+      "loss": 0.4129,
+      "step": 9323
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.063025913100129e-06,
+      "loss": 0.3925,
+      "step": 9324
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.060988736877366e-06,
+      "loss": 0.4228,
+      "step": 9325
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.058951831725067e-06,
+      "loss": 0.3861,
+      "step": 9326
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0569151977550305e-06,
+      "loss": 0.4398,
+      "step": 9327
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0548788350790325e-06,
+      "loss": 0.4104,
+      "step": 9328
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.052842743808834e-06,
+      "loss": 0.4034,
+      "step": 9329
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.050806924056184e-06,
+      "loss": 0.3607,
+      "step": 9330
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.048771375932815e-06,
+      "loss": 0.4166,
+      "step": 9331
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.046736099550445e-06,
+      "loss": 0.4246,
+      "step": 9332
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.044701095020774e-06,
+      "loss": 0.4341,
+      "step": 9333
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0426663624554985e-06,
+      "loss": 0.371,
+      "step": 9334
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.040631901966282e-06,
+      "loss": 0.4182,
+      "step": 9335
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.038597713664792e-06,
+      "loss": 0.4241,
+      "step": 9336
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0365637976626695e-06,
+      "loss": 0.4295,
+      "step": 9337
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.034530154071539e-06,
+      "loss": 0.4108,
+      "step": 9338
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.032496783003019e-06,
+      "loss": 0.4451,
+      "step": 9339
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.030463684568705e-06,
+      "loss": 0.4515,
+      "step": 9340
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.028430858880176e-06,
+      "loss": 0.4442,
+      "step": 9341
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.026398306049012e-06,
+      "loss": 0.4053,
+      "step": 9342
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0243660261867554e-06,
+      "loss": 0.4267,
+      "step": 9343
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0223340194049555e-06,
+      "loss": 0.4449,
+      "step": 9344
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.020302285815131e-06,
+      "loss": 0.4033,
+      "step": 9345
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0182708255287885e-06,
+      "loss": 0.4391,
+      "step": 9346
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.016239638657425e-06,
+      "loss": 0.4162,
+      "step": 9347
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0142087253125175e-06,
+      "loss": 0.4164,
+      "step": 9348
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.012178085605527e-06,
+      "loss": 0.4241,
+      "step": 9349
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.010147719647909e-06,
+      "loss": 0.4506,
+      "step": 9350
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.008117627551093e-06,
+      "loss": 0.4373,
+      "step": 9351
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.006087809426497e-06,
+      "loss": 0.4112,
+      "step": 9352
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.004058265385528e-06,
+      "loss": 0.4266,
+      "step": 9353
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.002028995539572e-06,
+      "loss": 0.4106,
+      "step": 9354
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.4203,
+      "step": 9355
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.99797127887818e-06,
+      "loss": 0.4054,
+      "step": 9356
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.995942832285441e-06,
+      "loss": 0.3758,
+      "step": 9357
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.9939146603331245e-06,
+      "loss": 0.4085,
+      "step": 9358
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.991886763132538e-06,
+      "loss": 0.4125,
+      "step": 9359
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.989859140794976e-06,
+      "loss": 0.4191,
+      "step": 9360
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.987831793431731e-06,
+      "loss": 0.3927,
+      "step": 9361
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.985804721154065e-06,
+      "loss": 0.419,
+      "step": 9362
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.983777924073232e-06,
+      "loss": 0.4557,
+      "step": 9363
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.981751402300469e-06,
+      "loss": 0.4049,
+      "step": 9364
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.979725155946996e-06,
+      "loss": 0.3887,
+      "step": 9365
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.977699185124027e-06,
+      "loss": 0.3995,
+      "step": 9366
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.975673489942753e-06,
+      "loss": 0.412,
+      "step": 9367
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.973648070514344e-06,
+      "loss": 0.4309,
+      "step": 9368
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.971622926949972e-06,
+      "loss": 0.3479,
+      "step": 9369
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9695980593607815e-06,
+      "loss": 0.4035,
+      "step": 9370
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.967573467857902e-06,
+      "loss": 0.371,
+      "step": 9371
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9655491525524515e-06,
+      "loss": 0.448,
+      "step": 9372
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.963525113555527e-06,
+      "loss": 0.4257,
+      "step": 9373
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.961501350978223e-06,
+      "loss": 0.4007,
+      "step": 9374
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.959477864931607e-06,
+      "loss": 0.4043,
+      "step": 9375
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.957454655526736e-06,
+      "loss": 0.4088,
+      "step": 9376
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.955431722874645e-06,
+      "loss": 0.3814,
+      "step": 9377
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.953409067086368e-06,
+      "loss": 0.4069,
+      "step": 9378
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.951386688272915e-06,
+      "loss": 0.4203,
+      "step": 9379
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.949364586545276e-06,
+      "loss": 0.4128,
+      "step": 9380
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.947342762014431e-06,
+      "loss": 0.3974,
+      "step": 9381
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.945321214791351e-06,
+      "loss": 0.4336,
+      "step": 9382
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.943299944986983e-06,
+      "loss": 0.4324,
+      "step": 9383
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.941278952712259e-06,
+      "loss": 0.3884,
+      "step": 9384
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.939258238078099e-06,
+      "loss": 0.4205,
+      "step": 9385
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.937237801195409e-06,
+      "loss": 0.4061,
+      "step": 9386
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.935217642175078e-06,
+      "loss": 0.4288,
+      "step": 9387
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.933197761127978e-06,
+      "loss": 0.3654,
+      "step": 9388
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.931178158164964e-06,
+      "loss": 0.3993,
+      "step": 9389
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.929158833396887e-06,
+      "loss": 0.3665,
+      "step": 9390
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.927139786934569e-06,
+      "loss": 0.4125,
+      "step": 9391
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.925121018888823e-06,
+      "loss": 0.4112,
+      "step": 9392
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.923102529370447e-06,
+      "loss": 0.4232,
+      "step": 9393
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.92108431849022e-06,
+      "loss": 0.4366,
+      "step": 9394
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.919066386358914e-06,
+      "loss": 0.4089,
+      "step": 9395
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.917048733087278e-06,
+      "loss": 0.4334,
+      "step": 9396
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.915031358786043e-06,
+      "loss": 0.4148,
+      "step": 9397
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.913014263565938e-06,
+      "loss": 0.3799,
+      "step": 9398
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.910997447537665e-06,
+      "loss": 0.394,
+      "step": 9399
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.908980910811914e-06,
+      "loss": 0.4059,
+      "step": 9400
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9069646534993575e-06,
+      "loss": 0.4425,
+      "step": 9401
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.904948675710654e-06,
+      "loss": 0.4341,
+      "step": 9402
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.902932977556453e-06,
+      "loss": 0.3692,
+      "step": 9403
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.90091755914738e-06,
+      "loss": 0.3882,
+      "step": 9404
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.898902420594049e-06,
+      "loss": 0.4068,
+      "step": 9405
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.896887562007054e-06,
+      "loss": 0.388,
+      "step": 9406
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.894872983496984e-06,
+      "loss": 0.4267,
+      "step": 9407
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.892858685174404e-06,
+      "loss": 0.4659,
+      "step": 9408
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.890844667149866e-06,
+      "loss": 0.4166,
+      "step": 9409
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.888830929533903e-06,
+      "loss": 0.4028,
+      "step": 9410
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.886817472437036e-06,
+      "loss": 0.4475,
+      "step": 9411
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8848042959697784e-06,
+      "loss": 0.3692,
+      "step": 9412
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.882791400242615e-06,
+      "loss": 0.4147,
+      "step": 9413
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.880778785366016e-06,
+      "loss": 0.4093,
+      "step": 9414
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.878766451450451e-06,
+      "loss": 0.3911,
+      "step": 9415
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.876754398606358e-06,
+      "loss": 0.3827,
+      "step": 9416
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.874742626944167e-06,
+      "loss": 0.3783,
+      "step": 9417
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.87273113657429e-06,
+      "loss": 0.4197,
+      "step": 9418
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8707199276071225e-06,
+      "loss": 0.3672,
+      "step": 9419
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.868709000153052e-06,
+      "loss": 0.3961,
+      "step": 9420
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.866698354322444e-06,
+      "loss": 0.4136,
+      "step": 9421
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.864687990225646e-06,
+      "loss": 0.4031,
+      "step": 9422
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.862677907973e-06,
+      "loss": 0.3839,
+      "step": 9423
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.860668107674823e-06,
+      "loss": 0.4103,
+      "step": 9424
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.85865858944142e-06,
+      "loss": 0.4352,
+      "step": 9425
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.856649353383082e-06,
+      "loss": 0.4054,
+      "step": 9426
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.854640399610077e-06,
+      "loss": 0.4577,
+      "step": 9427
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.852631728232672e-06,
+      "loss": 0.4128,
+      "step": 9428
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.850623339361107e-06,
+      "loss": 0.4221,
+      "step": 9429
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.848615233105604e-06,
+      "loss": 0.4408,
+      "step": 9430
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8466074095763835e-06,
+      "loss": 0.4099,
+      "step": 9431
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.844599868883639e-06,
+      "loss": 0.4157,
+      "step": 9432
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.842592611137551e-06,
+      "loss": 0.3925,
+      "step": 9433
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8405856364482836e-06,
+      "loss": 0.3778,
+      "step": 9434
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.838578944925989e-06,
+      "loss": 0.4114,
+      "step": 9435
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.836572536680796e-06,
+      "loss": 0.4108,
+      "step": 9436
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.834566411822831e-06,
+      "loss": 0.3995,
+      "step": 9437
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.83256057046219e-06,
+      "loss": 0.4073,
+      "step": 9438
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.83055501270897e-06,
+      "loss": 0.4206,
+      "step": 9439
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.828549738673236e-06,
+      "loss": 0.4334,
+      "step": 9440
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.826544748465048e-06,
+      "loss": 0.4352,
+      "step": 9441
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8245400421944435e-06,
+      "loss": 0.4228,
+      "step": 9442
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.82253561997145e-06,
+      "loss": 0.4055,
+      "step": 9443
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8205314819060745e-06,
+      "loss": 0.4242,
+      "step": 9444
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.818527628108317e-06,
+      "loss": 0.3871,
+      "step": 9445
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.816524058688149e-06,
+      "loss": 0.4371,
+      "step": 9446
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.814520773755541e-06,
+      "loss": 0.398,
+      "step": 9447
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.812517773420437e-06,
+      "loss": 0.3802,
+      "step": 9448
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.810515057792768e-06,
+      "loss": 0.4029,
+      "step": 9449
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.80851262698245e-06,
+      "loss": 0.392,
+      "step": 9450
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.806510481099386e-06,
+      "loss": 0.4369,
+      "step": 9451
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8045086202534575e-06,
+      "loss": 0.4079,
+      "step": 9452
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.802507044554532e-06,
+      "loss": 0.3942,
+      "step": 9453
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.800505754112467e-06,
+      "loss": 0.4185,
+      "step": 9454
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.798504749037102e-06,
+      "loss": 0.418,
+      "step": 9455
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.796504029438258e-06,
+      "loss": 0.4258,
+      "step": 9456
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.79450359542574e-06,
+      "loss": 0.4129,
+      "step": 9457
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.79250344710934e-06,
+      "loss": 0.3676,
+      "step": 9458
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.790503584598832e-06,
+      "loss": 0.3877,
+      "step": 9459
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.788504008003978e-06,
+      "loss": 0.4102,
+      "step": 9460
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.786504717434515e-06,
+      "loss": 0.407,
+      "step": 9461
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.784505713000181e-06,
+      "loss": 0.392,
+      "step": 9462
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.782506994810679e-06,
+      "loss": 0.4251,
+      "step": 9463
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.780508562975716e-06,
+      "loss": 0.4195,
+      "step": 9464
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.778510417604965e-06,
+      "loss": 0.3919,
+      "step": 9465
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.776512558808095e-06,
+      "loss": 0.4309,
+      "step": 9466
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.774514986694755e-06,
+      "loss": 0.4577,
+      "step": 9467
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7725177013745785e-06,
+      "loss": 0.4028,
+      "step": 9468
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.770520702957182e-06,
+      "loss": 0.4219,
+      "step": 9469
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.768523991552166e-06,
+      "loss": 0.4445,
+      "step": 9470
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.766527567269122e-06,
+      "loss": 0.4271,
+      "step": 9471
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.764531430217621e-06,
+      "loss": 0.4203,
+      "step": 9472
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.762535580507217e-06,
+      "loss": 0.4144,
+      "step": 9473
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.760540018247449e-06,
+      "loss": 0.3934,
+      "step": 9474
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.758544743547839e-06,
+      "loss": 0.4081,
+      "step": 9475
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.756549756517898e-06,
+      "loss": 0.4492,
+      "step": 9476
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.754555057267114e-06,
+      "loss": 0.4132,
+      "step": 9477
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.752560645904962e-06,
+      "loss": 0.4146,
+      "step": 9478
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.750566522540906e-06,
+      "loss": 0.4143,
+      "step": 9479
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.748572687284393e-06,
+      "loss": 0.3813,
+      "step": 9480
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.746579140244848e-06,
+      "loss": 0.4212,
+      "step": 9481
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.744585881531686e-06,
+      "loss": 0.3998,
+      "step": 9482
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7425929112543025e-06,
+      "loss": 0.4234,
+      "step": 9483
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.740600229522079e-06,
+      "loss": 0.4108,
+      "step": 9484
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.738607836444381e-06,
+      "loss": 0.3709,
+      "step": 9485
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.736615732130555e-06,
+      "loss": 0.4311,
+      "step": 9486
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.734623916689941e-06,
+      "loss": 0.3894,
+      "step": 9487
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.732632390231851e-06,
+      "loss": 0.4475,
+      "step": 9488
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.730641152865592e-06,
+      "loss": 0.4306,
+      "step": 9489
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.728650204700449e-06,
+      "loss": 0.4362,
+      "step": 9490
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.726659545845691e-06,
+      "loss": 0.4132,
+      "step": 9491
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.724669176410572e-06,
+      "loss": 0.4315,
+      "step": 9492
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.722679096504332e-06,
+      "loss": 0.4547,
+      "step": 9493
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.720689306236194e-06,
+      "loss": 0.4149,
+      "step": 9494
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7186998057153586e-06,
+      "loss": 0.3689,
+      "step": 9495
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7167105950510215e-06,
+      "loss": 0.3867,
+      "step": 9496
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.714721674352363e-06,
+      "loss": 0.3932,
+      "step": 9497
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.712733043728536e-06,
+      "loss": 0.4212,
+      "step": 9498
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.710744703288685e-06,
+      "loss": 0.4328,
+      "step": 9499
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.708756653141936e-06,
+      "loss": 0.4104,
+      "step": 9500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.706768893397402e-06,
+      "loss": 0.3748,
+      "step": 9501
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.704781424164177e-06,
+      "loss": 0.3971,
+      "step": 9502
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.702794245551337e-06,
+      "loss": 0.4175,
+      "step": 9503
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.700807357667953e-06,
+      "loss": 0.3999,
+      "step": 9504
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.698820760623064e-06,
+      "loss": 0.3856,
+      "step": 9505
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6968344545257105e-06,
+      "loss": 0.4277,
+      "step": 9506
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.694848439484904e-06,
+      "loss": 0.453,
+      "step": 9507
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.692862715609643e-06,
+      "loss": 0.388,
+      "step": 9508
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.690877283008911e-06,
+      "loss": 0.4324,
+      "step": 9509
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.688892141791678e-06,
+      "loss": 0.3953,
+      "step": 9510
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.686907292066889e-06,
+      "loss": 0.4144,
+      "step": 9511
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.684922733943489e-06,
+      "loss": 0.4074,
+      "step": 9512
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.682938467530389e-06,
+      "loss": 0.4401,
+      "step": 9513
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6809544929365e-06,
+      "loss": 0.3847,
+      "step": 9514
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.678970810270707e-06,
+      "loss": 0.3722,
+      "step": 9515
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.67698741964188e-06,
+      "loss": 0.3873,
+      "step": 9516
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6750043211588755e-06,
+      "loss": 0.3916,
+      "step": 9517
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.673021514930533e-06,
+      "loss": 0.4418,
+      "step": 9518
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.671039001065672e-06,
+      "loss": 0.386,
+      "step": 9519
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.669056779673108e-06,
+      "loss": 0.3976,
+      "step": 9520
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.667074850861627e-06,
+      "loss": 0.4187,
+      "step": 9521
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.665093214740003e-06,
+      "loss": 0.4074,
+      "step": 9522
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.663111871417e-06,
+      "loss": 0.3715,
+      "step": 9523
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.661130821001359e-06,
+      "loss": 0.3616,
+      "step": 9524
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.659150063601806e-06,
+      "loss": 0.3915,
+      "step": 9525
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.657169599327053e-06,
+      "loss": 0.421,
+      "step": 9526
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6551894282857916e-06,
+      "loss": 0.4397,
+      "step": 9527
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.653209550586707e-06,
+      "loss": 0.4354,
+      "step": 9528
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.651229966338458e-06,
+      "loss": 0.4009,
+      "step": 9529
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.649250675649689e-06,
+      "loss": 0.4192,
+      "step": 9530
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6472716786290375e-06,
+      "loss": 0.4082,
+      "step": 9531
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.645292975385112e-06,
+      "loss": 0.4088,
+      "step": 9532
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.643314566026514e-06,
+      "loss": 0.4585,
+      "step": 9533
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.641336450661822e-06,
+      "loss": 0.4673,
+      "step": 9534
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.639358629399602e-06,
+      "loss": 0.4091,
+      "step": 9535
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6373811023484085e-06,
+      "loss": 0.4433,
+      "step": 9536
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.635403869616774e-06,
+      "loss": 0.4231,
+      "step": 9537
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.63342693131321e-06,
+      "loss": 0.3945,
+      "step": 9538
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6314502875462255e-06,
+      "loss": 0.4213,
+      "step": 9539
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.629473938424304e-06,
+      "loss": 0.4036,
+      "step": 9540
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.627497884055912e-06,
+      "loss": 0.3953,
+      "step": 9541
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.625522124549505e-06,
+      "loss": 0.4086,
+      "step": 9542
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.623546660013518e-06,
+      "loss": 0.3968,
+      "step": 9543
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.621571490556368e-06,
+      "loss": 0.384,
+      "step": 9544
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.619596616286467e-06,
+      "loss": 0.4345,
+      "step": 9545
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.617622037312198e-06,
+      "loss": 0.4242,
+      "step": 9546
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.615647753741932e-06,
+      "loss": 0.43,
+      "step": 9547
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.61367376568403e-06,
+      "loss": 0.3901,
+      "step": 9548
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.611700073246829e-06,
+      "loss": 0.3918,
+      "step": 9549
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.609726676538652e-06,
+      "loss": 0.4249,
+      "step": 9550
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.607753575667804e-06,
+      "loss": 0.3486,
+      "step": 9551
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.605780770742575e-06,
+      "loss": 0.3864,
+      "step": 9552
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.603808261871245e-06,
+      "loss": 0.4255,
+      "step": 9553
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.601836049162069e-06,
+      "loss": 0.4362,
+      "step": 9554
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.599864132723286e-06,
+      "loss": 0.3989,
+      "step": 9555
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.597892512663129e-06,
+      "loss": 0.422,
+      "step": 9556
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5959211890898024e-06,
+      "loss": 0.3733,
+      "step": 9557
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5939501621115015e-06,
+      "loss": 0.4542,
+      "step": 9558
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.591979431836402e-06,
+      "loss": 0.447,
+      "step": 9559
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.590008998372661e-06,
+      "loss": 0.4203,
+      "step": 9560
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5880388618284315e-06,
+      "loss": 0.4291,
+      "step": 9561
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.586069022311836e-06,
+      "loss": 0.42,
+      "step": 9562
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.584099479930987e-06,
+      "loss": 0.3823,
+      "step": 9563
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.582130234793976e-06,
+      "loss": 0.4232,
+      "step": 9564
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.580161287008892e-06,
+      "loss": 0.4337,
+      "step": 9565
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.578192636683791e-06,
+      "loss": 0.4462,
+      "step": 9566
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.576224283926719e-06,
+      "loss": 0.4095,
+      "step": 9567
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.574256228845706e-06,
+      "loss": 0.4221,
+      "step": 9568
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.57228847154877e-06,
+      "loss": 0.463,
+      "step": 9569
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.570321012143908e-06,
+      "loss": 0.3864,
+      "step": 9570
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5683538507390975e-06,
+      "loss": 0.3969,
+      "step": 9571
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.566386987442303e-06,
+      "loss": 0.4064,
+      "step": 9572
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5644204223614786e-06,
+      "loss": 0.4235,
+      "step": 9573
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.562454155604552e-06,
+      "loss": 0.4158,
+      "step": 9574
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.560488187279441e-06,
+      "loss": 0.396,
+      "step": 9575
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.558522517494039e-06,
+      "loss": 0.4443,
+      "step": 9576
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5565571463562365e-06,
+      "loss": 0.405,
+      "step": 9577
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.554592073973898e-06,
+      "loss": 0.4115,
+      "step": 9578
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5526273004548725e-06,
+      "loss": 0.4076,
+      "step": 9579
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.550662825906994e-06,
+      "loss": 0.4118,
+      "step": 9580
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.548698650438077e-06,
+      "loss": 0.391,
+      "step": 9581
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.546734774155928e-06,
+      "loss": 0.4057,
+      "step": 9582
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.54477119716833e-06,
+      "loss": 0.4247,
+      "step": 9583
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5428079195830445e-06,
+      "loss": 0.3783,
+      "step": 9584
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.540844941507833e-06,
+      "loss": 0.3837,
+      "step": 9585
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.538882263050426e-06,
+      "loss": 0.4007,
+      "step": 9586
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.536919884318542e-06,
+      "loss": 0.3961,
+      "step": 9587
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.534957805419885e-06,
+      "loss": 0.4275,
+      "step": 9588
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5329960264621355e-06,
+      "loss": 0.3863,
+      "step": 9589
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.531034547552971e-06,
+      "loss": 0.3865,
+      "step": 9590
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.52907336880004e-06,
+      "loss": 0.3739,
+      "step": 9591
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.527112490310975e-06,
+      "loss": 0.3879,
+      "step": 9592
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.525151912193405e-06,
+      "loss": 0.3649,
+      "step": 9593
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.523191634554929e-06,
+      "loss": 0.4214,
+      "step": 9594
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5212316575031325e-06,
+      "loss": 0.4128,
+      "step": 9595
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.519271981145589e-06,
+      "loss": 0.3773,
+      "step": 9596
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5173126055898464e-06,
+      "loss": 0.3905,
+      "step": 9597
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.51535353094345e-06,
+      "loss": 0.3925,
+      "step": 9598
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5133947573139165e-06,
+      "loss": 0.4033,
+      "step": 9599
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.51143628480875e-06,
+      "loss": 0.3977,
+      "step": 9600
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.509478113535441e-06,
+      "loss": 0.4035,
+      "step": 9601
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.507520243601461e-06,
+      "loss": 0.3961,
+      "step": 9602
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.505562675114263e-06,
+      "loss": 0.3856,
+      "step": 9603
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.503605408181287e-06,
+      "loss": 0.399,
+      "step": 9604
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.501648442909952e-06,
+      "loss": 0.4168,
+      "step": 9605
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.49969177940766e-06,
+      "loss": 0.4286,
+      "step": 9606
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.497735417781809e-06,
+      "loss": 0.3973,
+      "step": 9607
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.495779358139764e-06,
+      "loss": 0.3895,
+      "step": 9608
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4938236005888845e-06,
+      "loss": 0.4193,
+      "step": 9609
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.491868145236509e-06,
+      "loss": 0.4166,
+      "step": 9610
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.489912992189958e-06,
+      "loss": 0.4048,
+      "step": 9611
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.487958141556538e-06,
+      "loss": 0.3897,
+      "step": 9612
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4860035934435375e-06,
+      "loss": 0.4308,
+      "step": 9613
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.484049347958227e-06,
+      "loss": 0.386,
+      "step": 9614
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.482095405207867e-06,
+      "loss": 0.4372,
+      "step": 9615
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.480141765299696e-06,
+      "loss": 0.4001,
+      "step": 9616
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.478188428340931e-06,
+      "loss": 0.4582,
+      "step": 9617
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.476235394438787e-06,
+      "loss": 0.3808,
+      "step": 9618
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.474282663700449e-06,
+      "loss": 0.4099,
+      "step": 9619
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4723302362330886e-06,
+      "loss": 0.418,
+      "step": 9620
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.470378112143865e-06,
+      "loss": 0.4468,
+      "step": 9621
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.468426291539915e-06,
+      "loss": 0.3778,
+      "step": 9622
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.466474774528358e-06,
+      "loss": 0.4158,
+      "step": 9623
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.464523561216307e-06,
+      "loss": 0.4382,
+      "step": 9624
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.462572651710847e-06,
+      "loss": 0.4188,
+      "step": 9625
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.460622046119056e-06,
+      "loss": 0.4327,
+      "step": 9626
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4586717445479865e-06,
+      "loss": 0.4126,
+      "step": 9627
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.45672174710468e-06,
+      "loss": 0.4076,
+      "step": 9628
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.454772053896156e-06,
+      "loss": 0.4066,
+      "step": 9629
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.452822665029423e-06,
+      "loss": 0.4293,
+      "step": 9630
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.450873580611466e-06,
+      "loss": 0.4612,
+      "step": 9631
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.448924800749263e-06,
+      "loss": 0.4097,
+      "step": 9632
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.446976325549766e-06,
+      "loss": 0.4209,
+      "step": 9633
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.445028155119922e-06,
+      "loss": 0.407,
+      "step": 9634
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.443080289566646e-06,
+      "loss": 0.4238,
+      "step": 9635
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.441132728996845e-06,
+      "loss": 0.4241,
+      "step": 9636
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.439185473517411e-06,
+      "loss": 0.4101,
+      "step": 9637
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.437238523235212e-06,
+      "loss": 0.432,
+      "step": 9638
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.435291878257108e-06,
+      "loss": 0.4443,
+      "step": 9639
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.43334553868993e-06,
+      "loss": 0.417,
+      "step": 9640
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4313995046405055e-06,
+      "loss": 0.4029,
+      "step": 9641
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.429453776215644e-06,
+      "loss": 0.4247,
+      "step": 9642
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.427508353522129e-06,
+      "loss": 0.3684,
+      "step": 9643
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.425563236666733e-06,
+      "loss": 0.3882,
+      "step": 9644
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.423618425756212e-06,
+      "loss": 0.4357,
+      "step": 9645
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4216739208973024e-06,
+      "loss": 0.3963,
+      "step": 9646
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.419729722196725e-06,
+      "loss": 0.3933,
+      "step": 9647
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.417785829761183e-06,
+      "loss": 0.4009,
+      "step": 9648
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.415842243697369e-06,
+      "loss": 0.4218,
+      "step": 9649
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.413898964111949e-06,
+      "loss": 0.4375,
+      "step": 9650
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.411955991111583e-06,
+      "loss": 0.4007,
+      "step": 9651
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.410013324802904e-06,
+      "loss": 0.4216,
+      "step": 9652
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.408070965292534e-06,
+      "loss": 0.4064,
+      "step": 9653
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.406128912687074e-06,
+      "loss": 0.4049,
+      "step": 9654
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.404187167093113e-06,
+      "loss": 0.4564,
+      "step": 9655
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.402245728617218e-06,
+      "loss": 0.3908,
+      "step": 9656
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4003045973659475e-06,
+      "loss": 0.438,
+      "step": 9657
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.398363773445831e-06,
+      "loss": 0.427,
+      "step": 9658
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.3964232569633946e-06,
+      "loss": 0.4098,
+      "step": 9659
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.394483048025138e-06,
+      "loss": 0.4305,
+      "step": 9660
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.392543146737546e-06,
+      "loss": 0.398,
+      "step": 9661
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.390603553207088e-06,
+      "loss": 0.4033,
+      "step": 9662
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.388664267540215e-06,
+      "loss": 0.3889,
+      "step": 9663
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.386725289843362e-06,
+      "loss": 0.4183,
+      "step": 9664
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3847866202229426e-06,
+      "loss": 0.4109,
+      "step": 9665
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.382848258785364e-06,
+      "loss": 0.3565,
+      "step": 9666
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.380910205637012e-06,
+      "loss": 0.4098,
+      "step": 9667
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.378972460884251e-06,
+      "loss": 0.4052,
+      "step": 9668
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3770350246334305e-06,
+      "loss": 0.4104,
+      "step": 9669
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.375097896990884e-06,
+      "loss": 0.3749,
+      "step": 9670
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.373161078062928e-06,
+      "loss": 0.4213,
+      "step": 9671
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.371224567955863e-06,
+      "loss": 0.4451,
+      "step": 9672
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.369288366775967e-06,
+      "loss": 0.4013,
+      "step": 9673
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.367352474629512e-06,
+      "loss": 0.4218,
+      "step": 9674
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.365416891622742e-06,
+      "loss": 0.4406,
+      "step": 9675
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.363481617861894e-06,
+      "loss": 0.4382,
+      "step": 9676
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.361546653453179e-06,
+      "loss": 0.4146,
+      "step": 9677
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.359611998502795e-06,
+      "loss": 0.3992,
+      "step": 9678
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.357677653116923e-06,
+      "loss": 0.4032,
+      "step": 9679
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.355743617401726e-06,
+      "loss": 0.4125,
+      "step": 9680
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.353809891463352e-06,
+      "loss": 0.4282,
+      "step": 9681
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.351876475407927e-06,
+      "loss": 0.3841,
+      "step": 9682
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.349943369341566e-06,
+      "loss": 0.3849,
+      "step": 9683
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.348010573370369e-06,
+      "loss": 0.4138,
+      "step": 9684
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.346078087600411e-06,
+      "loss": 0.4174,
+      "step": 9685
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.344145912137755e-06,
+      "loss": 0.3806,
+      "step": 9686
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.342214047088443e-06,
+      "loss": 0.4029,
+      "step": 9687
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.340282492558505e-06,
+      "loss": 0.4183,
+      "step": 9688
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.338351248653952e-06,
+      "loss": 0.4066,
+      "step": 9689
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.336420315480771e-06,
+      "loss": 0.3647,
+      "step": 9690
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.334489693144946e-06,
+      "loss": 0.4382,
+      "step": 9691
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.332559381752433e-06,
+      "loss": 0.4084,
+      "step": 9692
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.330629381409178e-06,
+      "loss": 0.4193,
+      "step": 9693
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3286996922211034e-06,
+      "loss": 0.3943,
+      "step": 9694
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.326770314294119e-06,
+      "loss": 0.4326,
+      "step": 9695
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.324841247734113e-06,
+      "loss": 0.4173,
+      "step": 9696
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.322912492646963e-06,
+      "loss": 0.4012,
+      "step": 9697
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.320984049138518e-06,
+      "loss": 0.4012,
+      "step": 9698
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.319055917314629e-06,
+      "loss": 0.3972,
+      "step": 9699
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.317128097281109e-06,
+      "loss": 0.3751,
+      "step": 9700
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.315200589143772e-06,
+      "loss": 0.3556,
+      "step": 9701
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.313273393008404e-06,
+      "loss": 0.4442,
+      "step": 9702
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.311346508980773e-06,
+      "loss": 0.4209,
+      "step": 9703
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3094199371666355e-06,
+      "loss": 0.4286,
+      "step": 9704
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.307493677671729e-06,
+      "loss": 0.3949,
+      "step": 9705
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.305567730601768e-06,
+      "loss": 0.4261,
+      "step": 9706
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3036420960624644e-06,
+      "loss": 0.4083,
+      "step": 9707
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.301716774159497e-06,
+      "loss": 0.3803,
+      "step": 9708
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2997917649985384e-06,
+      "loss": 0.4432,
+      "step": 9709
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.29786706868524e-06,
+      "loss": 0.3883,
+      "step": 9710
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.295942685325234e-06,
+      "loss": 0.4343,
+      "step": 9711
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.294018615024137e-06,
+      "loss": 0.4073,
+      "step": 9712
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.29209485788755e-06,
+      "loss": 0.4004,
+      "step": 9713
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.290171414021051e-06,
+      "loss": 0.4473,
+      "step": 9714
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2882482835302145e-06,
+      "loss": 0.3886,
+      "step": 9715
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.286325466520582e-06,
+      "loss": 0.3998,
+      "step": 9716
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.284402963097684e-06,
+      "loss": 0.3779,
+      "step": 9717
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.282480773367039e-06,
+      "loss": 0.4347,
+      "step": 9718
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.280558897434141e-06,
+      "loss": 0.4131,
+      "step": 9719
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.278637335404472e-06,
+      "loss": 0.4004,
+      "step": 9720
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2767160873834904e-06,
+      "loss": 0.4051,
+      "step": 9721
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.274795153476639e-06,
+      "loss": 0.4222,
+      "step": 9722
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.272874533789352e-06,
+      "loss": 0.3848,
+      "step": 9723
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2709542284270366e-06,
+      "loss": 0.3945,
+      "step": 9724
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.269034237495082e-06,
+      "loss": 0.4105,
+      "step": 9725
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.267114561098873e-06,
+      "loss": 0.4054,
+      "step": 9726
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2651951993437636e-06,
+      "loss": 0.3948,
+      "step": 9727
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.263276152335094e-06,
+      "loss": 0.4137,
+      "step": 9728
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.261357420178191e-06,
+      "loss": 0.3959,
+      "step": 9729
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.259439002978354e-06,
+      "loss": 0.3995,
+      "step": 9730
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.257520900840884e-06,
+      "loss": 0.4247,
+      "step": 9731
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.255603113871045e-06,
+      "loss": 0.4002,
+      "step": 9732
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2536856421740964e-06,
+      "loss": 0.3993,
+      "step": 9733
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.251768485855271e-06,
+      "loss": 0.3955,
+      "step": 9734
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.249851645019795e-06,
+      "loss": 0.3727,
+      "step": 9735
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.247935119772868e-06,
+      "loss": 0.3814,
+      "step": 9736
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.246018910219677e-06,
+      "loss": 0.3879,
+      "step": 9737
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.244103016465387e-06,
+      "loss": 0.4112,
+      "step": 9738
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.242187438615153e-06,
+      "loss": 0.3865,
+      "step": 9739
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.240272176774108e-06,
+      "loss": 0.4308,
+      "step": 9740
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.238357231047369e-06,
+      "loss": 0.4019,
+      "step": 9741
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.23644260154003e-06,
+      "loss": 0.4039,
+      "step": 9742
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.234528288357179e-06,
+      "loss": 0.444,
+      "step": 9743
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.232614291603878e-06,
+      "loss": 0.3882,
+      "step": 9744
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2307006113851745e-06,
+      "loss": 0.4005,
+      "step": 9745
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.228787247806093e-06,
+      "loss": 0.4121,
+      "step": 9746
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.226874200971655e-06,
+      "loss": 0.4195,
+      "step": 9747
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.224961470986849e-06,
+      "loss": 0.433,
+      "step": 9748
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.223049057956653e-06,
+      "loss": 0.4179,
+      "step": 9749
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.221136961986028e-06,
+      "loss": 0.3854,
+      "step": 9750
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.219225183179914e-06,
+      "loss": 0.3904,
+      "step": 9751
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2173137216432414e-06,
+      "loss": 0.3945,
+      "step": 9752
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.215402577480915e-06,
+      "loss": 0.3765,
+      "step": 9753
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.213491750797825e-06,
+      "loss": 0.3941,
+      "step": 9754
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.211581241698841e-06,
+      "loss": 0.413,
+      "step": 9755
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.209671050288826e-06,
+      "loss": 0.3817,
+      "step": 9756
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.207761176672615e-06,
+      "loss": 0.4297,
+      "step": 9757
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.205851620955026e-06,
+      "loss": 0.4021,
+      "step": 9758
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.203942383240862e-06,
+      "loss": 0.3742,
+      "step": 9759
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2020334636349136e-06,
+      "loss": 0.4058,
+      "step": 9760
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.200124862241947e-06,
+      "loss": 0.3884,
+      "step": 9761
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.198216579166712e-06,
+      "loss": 0.3914,
+      "step": 9762
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.196308614513937e-06,
+      "loss": 0.379,
+      "step": 9763
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1944009683883495e-06,
+      "loss": 0.4294,
+      "step": 9764
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.19249364089464e-06,
+      "loss": 0.3721,
+      "step": 9765
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.190586632137491e-06,
+      "loss": 0.4119,
+      "step": 9766
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.188679942221563e-06,
+      "loss": 0.4177,
+      "step": 9767
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.186773571251506e-06,
+      "loss": 0.4067,
+      "step": 9768
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.18486751933195e-06,
+      "loss": 0.4332,
+      "step": 9769
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1829617865675024e-06,
+      "loss": 0.3781,
+      "step": 9770
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.181056373062753e-06,
+      "loss": 0.4374,
+      "step": 9771
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.179151278922285e-06,
+      "loss": 0.4352,
+      "step": 9772
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.177246504250654e-06,
+      "loss": 0.4303,
+      "step": 9773
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.175342049152401e-06,
+      "loss": 0.3986,
+      "step": 9774
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.173437913732048e-06,
+      "loss": 0.4315,
+      "step": 9775
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.171534098094097e-06,
+      "loss": 0.3665,
+      "step": 9776
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1696306023430455e-06,
+      "loss": 0.4281,
+      "step": 9777
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.167727426583359e-06,
+      "loss": 0.4402,
+      "step": 9778
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.165824570919486e-06,
+      "loss": 0.4377,
+      "step": 9779
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.163922035455871e-06,
+      "loss": 0.3904,
+      "step": 9780
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.162019820296928e-06,
+      "loss": 0.3855,
+      "step": 9781
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.160117925547056e-06,
+      "loss": 0.4271,
+      "step": 9782
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.158216351310639e-06,
+      "loss": 0.4083,
+      "step": 9783
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.156315097692037e-06,
+      "loss": 0.4081,
+      "step": 9784
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.154414164795607e-06,
+      "loss": 0.4304,
+      "step": 9785
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.152513552725674e-06,
+      "loss": 0.4166,
+      "step": 9786
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.150613261586547e-06,
+      "loss": 0.4184,
+      "step": 9787
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.148713291482528e-06,
+      "loss": 0.4052,
+      "step": 9788
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1468136425178895e-06,
+      "loss": 0.4325,
+      "step": 9789
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.144914314796893e-06,
+      "loss": 0.4051,
+      "step": 9790
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.143015308423779e-06,
+      "loss": 0.4014,
+      "step": 9791
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.141116623502771e-06,
+      "loss": 0.419,
+      "step": 9792
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.139218260138074e-06,
+      "loss": 0.3851,
+      "step": 9793
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.137320218433882e-06,
+      "loss": 0.4012,
+      "step": 9794
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.13542249849436e-06,
+      "loss": 0.435,
+      "step": 9795
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.13352510042367e-06,
+      "loss": 0.4329,
+      "step": 9796
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.131628024325942e-06,
+      "loss": 0.4067,
+      "step": 9797
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.129731270305296e-06,
+      "loss": 0.3936,
+      "step": 9798
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.127834838465832e-06,
+      "loss": 0.4285,
+      "step": 9799
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.125938728911634e-06,
+      "loss": 0.452,
+      "step": 9800
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.124042941746761e-06,
+      "loss": 0.3842,
+      "step": 9801
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.4247,
+      "step": 9802
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.120252335001182e-06,
+      "loss": 0.396,
+      "step": 9803
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.118357515628518e-06,
+      "loss": 0.4616,
+      "step": 9804
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1164630190612685e-06,
+      "loss": 0.4632,
+      "step": 9805
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1145688454034095e-06,
+      "loss": 0.3899,
+      "step": 9806
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.112674994758901e-06,
+      "loss": 0.4138,
+      "step": 9807
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1107814672316825e-06,
+      "loss": 0.4143,
+      "step": 9808
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.108888262925679e-06,
+      "loss": 0.4342,
+      "step": 9809
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1069953819447915e-06,
+      "loss": 0.47,
+      "step": 9810
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.105102824392912e-06,
+      "loss": 0.3977,
+      "step": 9811
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.103210590373916e-06,
+      "loss": 0.3686,
+      "step": 9812
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.101318679991651e-06,
+      "loss": 0.4375,
+      "step": 9813
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.099427093349952e-06,
+      "loss": 0.4225,
+      "step": 9814
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.097535830552635e-06,
+      "loss": 0.4345,
+      "step": 9815
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0956448917034995e-06,
+      "loss": 0.441,
+      "step": 9816
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.093754276906328e-06,
+      "loss": 0.4294,
+      "step": 9817
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.09186398626488e-06,
+      "loss": 0.3843,
+      "step": 9818
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0899740198829096e-06,
+      "loss": 0.4372,
+      "step": 9819
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.088084377864136e-06,
+      "loss": 0.3939,
+      "step": 9820
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.086195060312277e-06,
+      "loss": 0.4138,
+      "step": 9821
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.084306067331022e-06,
+      "loss": 0.4276,
+      "step": 9822
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.082417399024046e-06,
+      "loss": 0.4241,
+      "step": 9823
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.080529055495004e-06,
+      "loss": 0.3923,
+      "step": 9824
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.078641036847537e-06,
+      "loss": 0.3828,
+      "step": 9825
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.076753343185262e-06,
+      "loss": 0.4089,
+      "step": 9826
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.074865974611789e-06,
+      "loss": 0.4371,
+      "step": 9827
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.072978931230697e-06,
+      "loss": 0.4285,
+      "step": 9828
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0710922131455595e-06,
+      "loss": 0.4212,
+      "step": 9829
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.069205820459924e-06,
+      "loss": 0.4072,
+      "step": 9830
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.067319753277322e-06,
+      "loss": 0.4164,
+      "step": 9831
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.065434011701268e-06,
+      "loss": 0.4304,
+      "step": 9832
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.063548595835259e-06,
+      "loss": 0.396,
+      "step": 9833
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.06166350578277e-06,
+      "loss": 0.3949,
+      "step": 9834
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0597787416472614e-06,
+      "loss": 0.4223,
+      "step": 9835
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.057894303532177e-06,
+      "loss": 0.3909,
+      "step": 9836
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.056010191540946e-06,
+      "loss": 0.4029,
+      "step": 9837
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.054126405776971e-06,
+      "loss": 0.3793,
+      "step": 9838
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.052242946343643e-06,
+      "loss": 0.3848,
+      "step": 9839
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.05035981334433e-06,
+      "loss": 0.3852,
+      "step": 9840
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0484770068823865e-06,
+      "loss": 0.3969,
+      "step": 9841
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.046594527061149e-06,
+      "loss": 0.4138,
+      "step": 9842
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0447123739839265e-06,
+      "loss": 0.3725,
+      "step": 9843
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.042830547754031e-06,
+      "loss": 0.4153,
+      "step": 9844
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.040949048474733e-06,
+      "loss": 0.3833,
+      "step": 9845
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.039067876249304e-06,
+      "loss": 0.4024,
+      "step": 9846
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.037187031180986e-06,
+      "loss": 0.4041,
+      "step": 9847
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0353065133730066e-06,
+      "loss": 0.4317,
+      "step": 9848
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.033426322928574e-06,
+      "loss": 0.3978,
+      "step": 9849
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.031546459950882e-06,
+      "loss": 0.4146,
+      "step": 9850
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.029666924543102e-06,
+      "loss": 0.4219,
+      "step": 9851
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.027787716808388e-06,
+      "loss": 0.3965,
+      "step": 9852
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.025908836849879e-06,
+      "loss": 0.3729,
+      "step": 9853
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0240302847707e-06,
+      "loss": 0.3984,
+      "step": 9854
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.022152060673948e-06,
+      "loss": 0.4041,
+      "step": 9855
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.020274164662708e-06,
+      "loss": 0.3857,
+      "step": 9856
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.018396596840043e-06,
+      "loss": 0.3995,
+      "step": 9857
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.016519357309002e-06,
+      "loss": 0.4179,
+      "step": 9858
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.014642446172615e-06,
+      "loss": 0.3792,
+      "step": 9859
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.012765863533889e-06,
+      "loss": 0.4016,
+      "step": 9860
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0108896094958276e-06,
+      "loss": 0.3997,
+      "step": 9861
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.009013684161395e-06,
+      "loss": 0.3837,
+      "step": 9862
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.007138087633558e-06,
+      "loss": 0.3938,
+      "step": 9863
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.005262820015251e-06,
+      "loss": 0.3837,
+      "step": 9864
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.003387881409397e-06,
+      "loss": 0.4143,
+      "step": 9865
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0015132719188996e-06,
+      "loss": 0.4107,
+      "step": 9866
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.999638991646641e-06,
+      "loss": 0.4224,
+      "step": 9867
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.997765040695489e-06,
+      "loss": 0.3901,
+      "step": 9868
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9958914191682974e-06,
+      "loss": 0.3938,
+      "step": 9869
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.994018127167889e-06,
+      "loss": 0.3967,
+      "step": 9870
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.992145164797086e-06,
+      "loss": 0.4079,
+      "step": 9871
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9902725321586785e-06,
+      "loss": 0.4376,
+      "step": 9872
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.988400229355444e-06,
+      "loss": 0.3595,
+      "step": 9873
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9865282564901406e-06,
+      "loss": 0.3977,
+      "step": 9874
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.98465661366551e-06,
+      "loss": 0.4252,
+      "step": 9875
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.982785300984269e-06,
+      "loss": 0.3864,
+      "step": 9876
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.98091431854913e-06,
+      "loss": 0.4377,
+      "step": 9877
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.979043666462774e-06,
+      "loss": 0.4083,
+      "step": 9878
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9771733448278685e-06,
+      "loss": 0.3919,
+      "step": 9879
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.975303353747068e-06,
+      "loss": 0.4232,
+      "step": 9880
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.973433693323003e-06,
+      "loss": 0.4083,
+      "step": 9881
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.971564363658286e-06,
+      "loss": 0.4177,
+      "step": 9882
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.969695364855512e-06,
+      "loss": 0.4253,
+      "step": 9883
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.967826697017253e-06,
+      "loss": 0.3839,
+      "step": 9884
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9659583602460795e-06,
+      "loss": 0.3954,
+      "step": 9885
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.964090354644525e-06,
+      "loss": 0.3734,
+      "step": 9886
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.962222680315112e-06,
+      "loss": 0.3592,
+      "step": 9887
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.96035533736035e-06,
+      "loss": 0.4219,
+      "step": 9888
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.958488325882721e-06,
+      "loss": 0.4354,
+      "step": 9889
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.956621645984696e-06,
+      "loss": 0.4323,
+      "step": 9890
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.954755297768723e-06,
+      "loss": 0.4403,
+      "step": 9891
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9528892813372354e-06,
+      "loss": 0.4137,
+      "step": 9892
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.951023596792641e-06,
+      "loss": 0.4311,
+      "step": 9893
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9491582442373445e-06,
+      "loss": 0.399,
+      "step": 9894
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.947293223773715e-06,
+      "loss": 0.46,
+      "step": 9895
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.945428535504118e-06,
+      "loss": 0.3968,
+      "step": 9896
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.943564179530892e-06,
+      "loss": 0.4032,
+      "step": 9897
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.941700155956358e-06,
+      "loss": 0.4121,
+      "step": 9898
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.939836464882821e-06,
+      "loss": 0.3868,
+      "step": 9899
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.937973106412567e-06,
+      "loss": 0.4061,
+      "step": 9900
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.93611008064786e-06,
+      "loss": 0.4504,
+      "step": 9901
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.934247387690958e-06,
+      "loss": 0.4127,
+      "step": 9902
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.932385027644085e-06,
+      "loss": 0.4062,
+      "step": 9903
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.930523000609455e-06,
+      "loss": 0.3984,
+      "step": 9904
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9286613066892665e-06,
+      "loss": 0.4093,
+      "step": 9905
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9267999459856935e-06,
+      "loss": 0.4283,
+      "step": 9906
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.924938918600894e-06,
+      "loss": 0.4115,
+      "step": 9907
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.923078224637007e-06,
+      "loss": 0.3947,
+      "step": 9908
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.921217864196151e-06,
+      "loss": 0.3879,
+      "step": 9909
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.919357837380437e-06,
+      "loss": 0.4135,
+      "step": 9910
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.917498144291945e-06,
+      "loss": 0.452,
+      "step": 9911
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.915638785032739e-06,
+      "loss": 0.4138,
+      "step": 9912
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.913779759704874e-06,
+      "loss": 0.3983,
+      "step": 9913
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.911921068410377e-06,
+      "loss": 0.4307,
+      "step": 9914
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.910062711251259e-06,
+      "loss": 0.4124,
+      "step": 9915
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.908204688329512e-06,
+      "loss": 0.3845,
+      "step": 9916
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9063469997471094e-06,
+      "loss": 0.3916,
+      "step": 9917
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.904489645606014e-06,
+      "loss": 0.4021,
+      "step": 9918
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.90263262600816e-06,
+      "loss": 0.4392,
+      "step": 9919
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.900775941055468e-06,
+      "loss": 0.4056,
+      "step": 9920
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.898919590849836e-06,
+      "loss": 0.3955,
+      "step": 9921
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.897063575493153e-06,
+      "loss": 0.4423,
+      "step": 9922
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.895207895087282e-06,
+      "loss": 0.4159,
+      "step": 9923
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.893352549734066e-06,
+      "loss": 0.3822,
+      "step": 9924
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.891497539535334e-06,
+      "loss": 0.3776,
+      "step": 9925
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8896428645929e-06,
+      "loss": 0.4058,
+      "step": 9926
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.88778852500855e-06,
+      "loss": 0.4109,
+      "step": 9927
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.88593452088406e-06,
+      "loss": 0.3855,
+      "step": 9928
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.884080852321179e-06,
+      "loss": 0.3713,
+      "step": 9929
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.882227519421651e-06,
+      "loss": 0.4219,
+      "step": 9930
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.880374522287188e-06,
+      "loss": 0.3742,
+      "step": 9931
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.878521861019492e-06,
+      "loss": 0.3764,
+      "step": 9932
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.876669535720238e-06,
+      "loss": 0.4165,
+      "step": 9933
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.874817546491095e-06,
+      "loss": 0.3805,
+      "step": 9934
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8729658934337044e-06,
+      "loss": 0.4285,
+      "step": 9935
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.871114576649691e-06,
+      "loss": 0.4335,
+      "step": 9936
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.869263596240661e-06,
+      "loss": 0.4483,
+      "step": 9937
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8674129523082006e-06,
+      "loss": 0.3904,
+      "step": 9938
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8655626449538865e-06,
+      "loss": 0.4282,
+      "step": 9939
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.863712674279265e-06,
+      "loss": 0.4001,
+      "step": 9940
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.861863040385868e-06,
+      "loss": 0.4278,
+      "step": 9941
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.860013743375215e-06,
+      "loss": 0.4435,
+      "step": 9942
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.858164783348799e-06,
+      "loss": 0.4606,
+      "step": 9943
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.856316160408098e-06,
+      "loss": 0.4426,
+      "step": 9944
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.85446787465457e-06,
+      "loss": 0.3805,
+      "step": 9945
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.852619926189655e-06,
+      "loss": 0.4125,
+      "step": 9946
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.850772315114777e-06,
+      "loss": 0.3742,
+      "step": 9947
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.84892504153134e-06,
+      "loss": 0.3958,
+      "step": 9948
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8470781055407235e-06,
+      "loss": 0.3685,
+      "step": 9949
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.845231507244301e-06,
+      "loss": 0.4091,
+      "step": 9950
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8433852467434175e-06,
+      "loss": 0.4426,
+      "step": 9951
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8415393241394025e-06,
+      "loss": 0.4278,
+      "step": 9952
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8396937395335665e-06,
+      "loss": 0.4023,
+      "step": 9953
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.837848493027198e-06,
+      "loss": 0.4102,
+      "step": 9954
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.836003584721577e-06,
+      "loss": 0.398,
+      "step": 9955
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.834159014717956e-06,
+      "loss": 0.3958,
+      "step": 9956
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.832314783117572e-06,
+      "loss": 0.4035,
+      "step": 9957
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.830470890021639e-06,
+      "loss": 0.3981,
+      "step": 9958
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.828627335531364e-06,
+      "loss": 0.4143,
+      "step": 9959
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8267841197479225e-06,
+      "loss": 0.3997,
+      "step": 9960
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.824941242772479e-06,
+      "loss": 0.4051,
+      "step": 9961
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.823098704706174e-06,
+      "loss": 0.4127,
+      "step": 9962
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.821256505650133e-06,
+      "loss": 0.3812,
+      "step": 9963
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.819414645705466e-06,
+      "loss": 0.3941,
+      "step": 9964
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8175731249732595e-06,
+      "loss": 0.4332,
+      "step": 9965
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.815731943554579e-06,
+      "loss": 0.4364,
+      "step": 9966
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.813891101550481e-06,
+      "loss": 0.3594,
+      "step": 9967
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.812050599061994e-06,
+      "loss": 0.4332,
+      "step": 9968
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8102104361901328e-06,
+      "loss": 0.3782,
+      "step": 9969
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.80837061303589e-06,
+      "loss": 0.4576,
+      "step": 9970
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8065311297002392e-06,
+      "loss": 0.4269,
+      "step": 9971
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.804691986284146e-06,
+      "loss": 0.4212,
+      "step": 9972
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.802853182888543e-06,
+      "loss": 0.3993,
+      "step": 9973
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8010147196143486e-06,
+      "loss": 0.4118,
+      "step": 9974
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7991765965624703e-06,
+      "loss": 0.3915,
+      "step": 9975
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7973388138337884e-06,
+      "loss": 0.3822,
+      "step": 9976
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.795501371529167e-06,
+      "loss": 0.407,
+      "step": 9977
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7936642697494497e-06,
+      "loss": 0.404,
+      "step": 9978
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.791827508595464e-06,
+      "loss": 0.4378,
+      "step": 9979
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7899910881680145e-06,
+      "loss": 0.4037,
+      "step": 9980
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7881550085678973e-06,
+      "loss": 0.4318,
+      "step": 9981
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.786319269895877e-06,
+      "loss": 0.4036,
+      "step": 9982
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.784483872252711e-06,
+      "loss": 0.4216,
+      "step": 9983
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7826488157391304e-06,
+      "loss": 0.3945,
+      "step": 9984
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7808141004558483e-06,
+      "loss": 0.3866,
+      "step": 9985
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.778979726503561e-06,
+      "loss": 0.4107,
+      "step": 9986
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7771456939829456e-06,
+      "loss": 0.4262,
+      "step": 9987
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7753120029946553e-06,
+      "loss": 0.4057,
+      "step": 9988
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7734786536393395e-06,
+      "loss": 0.3874,
+      "step": 9989
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7716456460176088e-06,
+      "loss": 0.4242,
+      "step": 9990
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7698129802300744e-06,
+      "loss": 0.4059,
+      "step": 9991
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.767980656377316e-06,
+      "loss": 0.3987,
+      "step": 9992
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.766148674559895e-06,
+      "loss": 0.371,
+      "step": 9993
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7643170348783607e-06,
+      "loss": 0.4238,
+      "step": 9994
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.762485737433239e-06,
+      "loss": 0.4149,
+      "step": 9995
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7606547823250374e-06,
+      "loss": 0.4092,
+      "step": 9996
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7588241696542417e-06,
+      "loss": 0.3878,
+      "step": 9997
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7569938995213253e-06,
+      "loss": 0.414,
+      "step": 9998
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7551639720267453e-06,
+      "loss": 0.4189,
+      "step": 9999
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7533343872709295e-06,
+      "loss": 0.3867,
+      "step": 10000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7515051453542926e-06,
+      "loss": 0.4474,
+      "step": 10001
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7496762463772307e-06,
+      "loss": 0.3891,
+      "step": 10002
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7478476904401182e-06,
+      "loss": 0.4288,
+      "step": 10003
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7460194776433144e-06,
+      "loss": 0.4107,
+      "step": 10004
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7441916080871543e-06,
+      "loss": 0.4157,
+      "step": 10005
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.742364081871962e-06,
+      "loss": 0.3808,
+      "step": 10006
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7405368990980395e-06,
+      "loss": 0.399,
+      "step": 10007
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.738710059865669e-06,
+      "loss": 0.376,
+      "step": 10008
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7368835642751123e-06,
+      "loss": 0.3968,
+      "step": 10009
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7350574124266138e-06,
+      "loss": 0.4219,
+      "step": 10010
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7332316044203985e-06,
+      "loss": 0.4115,
+      "step": 10011
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7314061403566748e-06,
+      "loss": 0.3787,
+      "step": 10012
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7295810203356262e-06,
+      "loss": 0.3961,
+      "step": 10013
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.72775624445743e-06,
+      "loss": 0.4205,
+      "step": 10014
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7259318128222276e-06,
+      "loss": 0.421,
+      "step": 10015
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7241077255301584e-06,
+      "loss": 0.3919,
+      "step": 10016
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7222839826813318e-06,
+      "loss": 0.4461,
+      "step": 10017
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.720460584375839e-06,
+      "loss": 0.3765,
+      "step": 10018
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.718637530713757e-06,
+      "loss": 0.3745,
+      "step": 10019
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7168148217951395e-06,
+      "loss": 0.3662,
+      "step": 10020
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7149924577200247e-06,
+      "loss": 0.3953,
+      "step": 10021
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.713170438588427e-06,
+      "loss": 0.4069,
+      "step": 10022
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7113487645003477e-06,
+      "loss": 0.4425,
+      "step": 10023
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.709527435555771e-06,
+      "loss": 0.4299,
+      "step": 10024
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7077064518546544e-06,
+      "loss": 0.4047,
+      "step": 10025
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.705885813496939e-06,
+      "loss": 0.4701,
+      "step": 10026
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.704065520582549e-06,
+      "loss": 0.396,
+      "step": 10027
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.702245573211388e-06,
+      "loss": 0.4096,
+      "step": 10028
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.700425971483341e-06,
+      "loss": 0.4303,
+      "step": 10029
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6986067154982717e-06,
+      "loss": 0.3891,
+      "step": 10030
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6967878053560334e-06,
+      "loss": 0.3874,
+      "step": 10031
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6949692411564473e-06,
+      "loss": 0.4012,
+      "step": 10032
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.693151022999331e-06,
+      "loss": 0.4141,
+      "step": 10033
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6913331509844686e-06,
+      "loss": 0.3647,
+      "step": 10034
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.689515625211634e-06,
+      "loss": 0.3931,
+      "step": 10035
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6876984457805786e-06,
+      "loss": 0.3949,
+      "step": 10036
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6858816127910347e-06,
+      "loss": 0.3783,
+      "step": 10037
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6840651263427175e-06,
+      "loss": 0.3897,
+      "step": 10038
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.682248986535318e-06,
+      "loss": 0.423,
+      "step": 10039
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6804331934685176e-06,
+      "loss": 0.4309,
+      "step": 10040
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.678617747241976e-06,
+      "loss": 0.3931,
+      "step": 10041
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6768026479553266e-06,
+      "loss": 0.4048,
+      "step": 10042
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6749878957081896e-06,
+      "loss": 0.4538,
+      "step": 10043
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.673173490600166e-06,
+      "loss": 0.4001,
+      "step": 10044
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6713594327308343e-06,
+      "loss": 0.3967,
+      "step": 10045
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.669545722199759e-06,
+      "loss": 0.4098,
+      "step": 10046
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.667732359106478e-06,
+      "loss": 0.3975,
+      "step": 10047
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6659193435505234e-06,
+      "loss": 0.4279,
+      "step": 10048
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6641066756313914e-06,
+      "loss": 0.3764,
+      "step": 10049
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6622943554485747e-06,
+      "loss": 0.4065,
+      "step": 10050
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.660482383101537e-06,
+      "loss": 0.4274,
+      "step": 10051
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6586707586897265e-06,
+      "loss": 0.3881,
+      "step": 10052
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6568594823125703e-06,
+      "loss": 0.3829,
+      "step": 10053
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.655048554069478e-06,
+      "loss": 0.3969,
+      "step": 10054
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.653237974059837e-06,
+      "loss": 0.4117,
+      "step": 10055
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6514277423830236e-06,
+      "loss": 0.4166,
+      "step": 10056
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6496178591383847e-06,
+      "loss": 0.4072,
+      "step": 10057
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6478083244252594e-06,
+      "loss": 0.4371,
+      "step": 10058
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6459991383429572e-06,
+      "loss": 0.446,
+      "step": 10059
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.644190300990774e-06,
+      "loss": 0.424,
+      "step": 10060
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6423818124679846e-06,
+      "loss": 0.4546,
+      "step": 10061
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.640573672873845e-06,
+      "loss": 0.4274,
+      "step": 10062
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.638765882307589e-06,
+      "loss": 0.3861,
+      "step": 10063
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.636958440868441e-06,
+      "loss": 0.4274,
+      "step": 10064
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6351513486555934e-06,
+      "loss": 0.4317,
+      "step": 10065
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.633344605768233e-06,
+      "loss": 0.4181,
+      "step": 10066
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6315382123055163e-06,
+      "loss": 0.3964,
+      "step": 10067
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6297321683665852e-06,
+      "loss": 0.3949,
+      "step": 10068
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6279264740505616e-06,
+      "loss": 0.4312,
+      "step": 10069
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.626121129456548e-06,
+      "loss": 0.4261,
+      "step": 10070
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6243161346836242e-06,
+      "loss": 0.4064,
+      "step": 10071
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.622511489830863e-06,
+      "loss": 0.4036,
+      "step": 10072
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.620707194997306e-06,
+      "loss": 0.4086,
+      "step": 10073
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6189032502819742e-06,
+      "loss": 0.3915,
+      "step": 10074
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6170996557838843e-06,
+      "loss": 0.4286,
+      "step": 10075
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6152964116020174e-06,
+      "loss": 0.4183,
+      "step": 10076
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6134935178353438e-06,
+      "loss": 0.3814,
+      "step": 10077
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6116909745828135e-06,
+      "loss": 0.4114,
+      "step": 10078
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6098887819433506e-06,
+      "loss": 0.3914,
+      "step": 10079
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6080869400158737e-06,
+      "loss": 0.4042,
+      "step": 10080
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6062854488992714e-06,
+      "loss": 0.4199,
+      "step": 10081
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.604484308692412e-06,
+      "loss": 0.4035,
+      "step": 10082
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6026835194941555e-06,
+      "loss": 0.4315,
+      "step": 10083
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6008830814033326e-06,
+      "loss": 0.4073,
+      "step": 10084
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5990829945187568e-06,
+      "loss": 0.3977,
+      "step": 10085
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.597283258939224e-06,
+      "loss": 0.4264,
+      "step": 10086
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.595483874763507e-06,
+      "loss": 0.3726,
+      "step": 10087
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.593684842090368e-06,
+      "loss": 0.4593,
+      "step": 10088
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.591886161018542e-06,
+      "loss": 0.3883,
+      "step": 10089
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.590087831646746e-06,
+      "loss": 0.4024,
+      "step": 10090
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5882898540736756e-06,
+      "loss": 0.4031,
+      "step": 10091
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5864922283980175e-06,
+      "loss": 0.3853,
+      "step": 10092
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5846949547184282e-06,
+      "loss": 0.3786,
+      "step": 10093
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5828980331335483e-06,
+      "loss": 0.42,
+      "step": 10094
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5811014637419994e-06,
+      "loss": 0.4023,
+      "step": 10095
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5793052466423796e-06,
+      "loss": 0.4369,
+      "step": 10096
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5775093819332796e-06,
+      "loss": 0.404,
+      "step": 10097
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.575713869713258e-06,
+      "loss": 0.3895,
+      "step": 10098
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.573918710080857e-06,
+      "loss": 0.4342,
+      "step": 10099
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 0.3802,
+      "step": 10100
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.57032944897301e-06,
+      "loss": 0.3624,
+      "step": 10101
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5685353476945537e-06,
+      "loss": 0.395,
+      "step": 10102
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5667415993977026e-06,
+      "loss": 0.3862,
+      "step": 10103
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5649482041809016e-06,
+      "loss": 0.4227,
+      "step": 10104
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5631551621425843e-06,
+      "loss": 0.4084,
+      "step": 10105
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.561362473381158e-06,
+      "loss": 0.3906,
+      "step": 10106
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5595701379950098e-06,
+      "loss": 0.4246,
+      "step": 10107
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5577781560825065e-06,
+      "loss": 0.3857,
+      "step": 10108
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.555986527742006e-06,
+      "loss": 0.4255,
+      "step": 10109
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.554195253071835e-06,
+      "loss": 0.3498,
+      "step": 10110
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5524043321703063e-06,
+      "loss": 0.3985,
+      "step": 10111
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5506137651357063e-06,
+      "loss": 0.4073,
+      "step": 10112
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.548823552066316e-06,
+      "loss": 0.4126,
+      "step": 10113
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5470336930603856e-06,
+      "loss": 0.4165,
+      "step": 10114
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.545244188216148e-06,
+      "loss": 0.4166,
+      "step": 10115
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5434550376318157e-06,
+      "loss": 0.4052,
+      "step": 10116
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.541666241405588e-06,
+      "loss": 0.4227,
+      "step": 10117
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5398777996356383e-06,
+      "loss": 0.4313,
+      "step": 10118
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.538089712420123e-06,
+      "loss": 0.3975,
+      "step": 10119
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.536301979857174e-06,
+      "loss": 0.4084,
+      "step": 10120
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.534514602044916e-06,
+      "loss": 0.4035,
+      "step": 10121
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.532727579081444e-06,
+      "loss": 0.3911,
+      "step": 10122
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5309409110648364e-06,
+      "loss": 0.4103,
+      "step": 10123
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5291545980931464e-06,
+      "loss": 0.3935,
+      "step": 10124
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5273686402644213e-06,
+      "loss": 0.3987,
+      "step": 10125
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5255830376766764e-06,
+      "loss": 0.4145,
+      "step": 10126
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.523797790427914e-06,
+      "loss": 0.4147,
+      "step": 10127
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5220128986161096e-06,
+      "loss": 0.4347,
+      "step": 10128
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.520228362339232e-06,
+      "loss": 0.3941,
+      "step": 10129
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5184441816952185e-06,
+      "loss": 0.4104,
+      "step": 10130
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5166603567819934e-06,
+      "loss": 0.4051,
+      "step": 10131
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5148768876974572e-06,
+      "loss": 0.4264,
+      "step": 10132
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.51309377453949e-06,
+      "loss": 0.4256,
+      "step": 10133
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5113110174059628e-06,
+      "loss": 0.3446,
+      "step": 10134
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.509528616394716e-06,
+      "loss": 0.4494,
+      "step": 10135
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5077465716035706e-06,
+      "loss": 0.3954,
+      "step": 10136
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5059648831303382e-06,
+      "loss": 0.3824,
+      "step": 10137
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5041835510728016e-06,
+      "loss": 0.4489,
+      "step": 10138
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5024025755287263e-06,
+      "loss": 0.4019,
+      "step": 10139
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5006219565958568e-06,
+      "loss": 0.4065,
+      "step": 10140
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4988416943719196e-06,
+      "loss": 0.3699,
+      "step": 10141
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4970617889546254e-06,
+      "loss": 0.3978,
+      "step": 10142
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.495282240441661e-06,
+      "loss": 0.372,
+      "step": 10143
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4935030489306886e-06,
+      "loss": 0.4127,
+      "step": 10144
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4917242145193654e-06,
+      "loss": 0.4261,
+      "step": 10145
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4899457373053148e-06,
+      "loss": 0.4227,
+      "step": 10146
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.488167617386148e-06,
+      "loss": 0.4761,
+      "step": 10147
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4863898548594534e-06,
+      "loss": 0.4345,
+      "step": 10148
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.484612449822801e-06,
+      "loss": 0.4163,
+      "step": 10149
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4828354023737366e-06,
+      "loss": 0.4244,
+      "step": 10150
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4810587126097995e-06,
+      "loss": 0.4003,
+      "step": 10151
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4792823806284937e-06,
+      "loss": 0.3825,
+      "step": 10152
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.477506406527317e-06,
+      "loss": 0.3559,
+      "step": 10153
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4757307904037375e-06,
+      "loss": 0.4495,
+      "step": 10154
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4739555323552064e-06,
+      "loss": 0.4377,
+      "step": 10155
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4721806324791585e-06,
+      "loss": 0.4388,
+      "step": 10156
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4704060908730052e-06,
+      "loss": 0.4254,
+      "step": 10157
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.468631907634136e-06,
+      "loss": 0.4379,
+      "step": 10158
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.466858082859932e-06,
+      "loss": 0.3941,
+      "step": 10159
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4650846166477394e-06,
+      "loss": 0.4291,
+      "step": 10160
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4633115090949e-06,
+      "loss": 0.4138,
+      "step": 10161
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4615387602987238e-06,
+      "loss": 0.4272,
+      "step": 10162
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4597663703565065e-06,
+      "loss": 0.3812,
+      "step": 10163
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4579943393655223e-06,
+      "loss": 0.4291,
+      "step": 10164
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.456222667423028e-06,
+      "loss": 0.3945,
+      "step": 10165
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.454451354626257e-06,
+      "loss": 0.4122,
+      "step": 10166
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4526804010724246e-06,
+      "loss": 0.4157,
+      "step": 10167
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.450909806858731e-06,
+      "loss": 0.425,
+      "step": 10168
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4491395720823474e-06,
+      "loss": 0.4191,
+      "step": 10169
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4473696968404367e-06,
+      "loss": 0.4414,
+      "step": 10170
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4456001812301344e-06,
+      "loss": 0.384,
+      "step": 10171
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.443831025348555e-06,
+      "loss": 0.4411,
+      "step": 10172
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4420622292927974e-06,
+      "loss": 0.4453,
+      "step": 10173
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.44029379315994e-06,
+      "loss": 0.4208,
+      "step": 10174
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4385257170470376e-06,
+      "loss": 0.4334,
+      "step": 10175
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4367580010511336e-06,
+      "loss": 0.3786,
+      "step": 10176
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4349906452692404e-06,
+      "loss": 0.4454,
+      "step": 10177
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.433223649798365e-06,
+      "loss": 0.4195,
+      "step": 10178
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.431457014735481e-06,
+      "loss": 0.4342,
+      "step": 10179
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4296907401775494e-06,
+      "loss": 0.4121,
+      "step": 10180
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.427924826221507e-06,
+      "loss": 0.3907,
+      "step": 10181
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4261592729642767e-06,
+      "loss": 0.4035,
+      "step": 10182
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.424394080502752e-06,
+      "loss": 0.4094,
+      "step": 10183
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.422629248933822e-06,
+      "loss": 0.4137,
+      "step": 10184
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4208647783543394e-06,
+      "loss": 0.4576,
+      "step": 10185
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4191006688611508e-06,
+      "loss": 0.3952,
+      "step": 10186
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4173369205510733e-06,
+      "loss": 0.391,
+      "step": 10187
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4155735335209083e-06,
+      "loss": 0.407,
+      "step": 10188
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4138105078674366e-06,
+      "loss": 0.4073,
+      "step": 10189
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4120478436874195e-06,
+      "loss": 0.4084,
+      "step": 10190
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.410285541077598e-06,
+      "loss": 0.4122,
+      "step": 10191
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.40852360013469e-06,
+      "loss": 0.4027,
+      "step": 10192
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4067620209554007e-06,
+      "loss": 0.4359,
+      "step": 10193
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4050008036364145e-06,
+      "loss": 0.3808,
+      "step": 10194
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4032399482743916e-06,
+      "loss": 0.4271,
+      "step": 10195
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4014794549659722e-06,
+      "loss": 0.4323,
+      "step": 10196
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.39971932380778e-06,
+      "loss": 0.3941,
+      "step": 10197
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.397959554896415e-06,
+      "loss": 0.4221,
+      "step": 10198
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3962001483284623e-06,
+      "loss": 0.4279,
+      "step": 10199
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.394441104200479e-06,
+      "loss": 0.4203,
+      "step": 10200
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.392682422609015e-06,
+      "loss": 0.4375,
+      "step": 10201
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.390924103650587e-06,
+      "loss": 0.3991,
+      "step": 10202
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.389166147421703e-06,
+      "loss": 0.3871,
+      "step": 10203
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.387408554018843e-06,
+      "loss": 0.3756,
+      "step": 10204
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3856513235384713e-06,
+      "loss": 0.4239,
+      "step": 10205
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3838944560770293e-06,
+      "loss": 0.3979,
+      "step": 10206
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3821379517309406e-06,
+      "loss": 0.3783,
+      "step": 10207
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.380381810596609e-06,
+      "loss": 0.3836,
+      "step": 10208
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3786260327704147e-06,
+      "loss": 0.4173,
+      "step": 10209
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.376870618348722e-06,
+      "loss": 0.4178,
+      "step": 10210
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3751155674278803e-06,
+      "loss": 0.4145,
+      "step": 10211
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3733608801042085e-06,
+      "loss": 0.4033,
+      "step": 10212
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3716065564740108e-06,
+      "loss": 0.4407,
+      "step": 10213
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3698525966335693e-06,
+      "loss": 0.4009,
+      "step": 10214
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3680990006791482e-06,
+      "loss": 0.3979,
+      "step": 10215
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3663457687069924e-06,
+      "loss": 0.397,
+      "step": 10216
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3645929008133194e-06,
+      "loss": 0.4326,
+      "step": 10217
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3628403970943422e-06,
+      "loss": 0.4074,
+      "step": 10218
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3610882576462357e-06,
+      "loss": 0.3909,
+      "step": 10219
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.359336482565172e-06,
+      "loss": 0.3736,
+      "step": 10220
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3575850719472903e-06,
+      "loss": 0.4102,
+      "step": 10221
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3558340258887144e-06,
+      "loss": 0.4006,
+      "step": 10222
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.354083344485548e-06,
+      "loss": 0.3872,
+      "step": 10223
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.352333027833874e-06,
+      "loss": 0.4029,
+      "step": 10224
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3505830760297543e-06,
+      "loss": 0.4463,
+      "step": 10225
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3488334891692375e-06,
+      "loss": 0.4148,
+      "step": 10226
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3470842673483418e-06,
+      "loss": 0.426,
+      "step": 10227
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3453354106630763e-06,
+      "loss": 0.3723,
+      "step": 10228
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.343586919209422e-06,
+      "loss": 0.4121,
+      "step": 10229
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3418387930833408e-06,
+      "loss": 0.397,
+      "step": 10230
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.340091032380778e-06,
+      "loss": 0.4035,
+      "step": 10231
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3383436371976552e-06,
+      "loss": 0.4438,
+      "step": 10232
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3365966076298774e-06,
+      "loss": 0.3834,
+      "step": 10233
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3348499437733227e-06,
+      "loss": 0.4008,
+      "step": 10234
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3331036457238632e-06,
+      "loss": 0.4178,
+      "step": 10235
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3313577135773324e-06,
+      "loss": 0.4261,
+      "step": 10236
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3296121474295617e-06,
+      "loss": 0.3921,
+      "step": 10237
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.327866947376349e-06,
+      "loss": 0.4234,
+      "step": 10238
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.32612211351348e-06,
+      "loss": 0.3875,
+      "step": 10239
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.324377645936715e-06,
+      "loss": 0.4048,
+      "step": 10240
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.322633544741798e-06,
+      "loss": 0.386,
+      "step": 10241
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.320889810024447e-06,
+      "loss": 0.4067,
+      "step": 10242
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.319146441880371e-06,
+      "loss": 0.3958,
+      "step": 10243
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.317403440405246e-06,
+      "loss": 0.3965,
+      "step": 10244
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3156608056947405e-06,
+      "loss": 0.4192,
+      "step": 10245
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.313918537844494e-06,
+      "loss": 0.4141,
+      "step": 10246
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.312176636950127e-06,
+      "loss": 0.4315,
+      "step": 10247
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3104351031072425e-06,
+      "loss": 0.3842,
+      "step": 10248
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 0.4213,
+      "step": 10249
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.306953136958221e-06,
+      "loss": 0.4332,
+      "step": 10250
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3052127048431913e-06,
+      "loss": 0.3883,
+      "step": 10251
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3034726401618444e-06,
+      "loss": 0.4063,
+      "step": 10252
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.30173294300969e-06,
+      "loss": 0.4113,
+      "step": 10253
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.299993613482204e-06,
+      "loss": 0.3861,
+      "step": 10254
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2982546516748483e-06,
+      "loss": 0.4224,
+      "step": 10255
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.296516057683061e-06,
+      "loss": 0.4074,
+      "step": 10256
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.294777831602265e-06,
+      "loss": 0.4056,
+      "step": 10257
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.293039973527855e-06,
+      "loss": 0.3795,
+      "step": 10258
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2913024835552186e-06,
+      "loss": 0.4166,
+      "step": 10259
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.289565361779712e-06,
+      "loss": 0.4094,
+      "step": 10260
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2878286082966704e-06,
+      "loss": 0.4046,
+      "step": 10261
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2860922232014215e-06,
+      "loss": 0.3849,
+      "step": 10262
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.284356206589259e-06,
+      "loss": 0.4234,
+      "step": 10263
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2826205585554617e-06,
+      "loss": 0.412,
+      "step": 10264
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2808852791952892e-06,
+      "loss": 0.4292,
+      "step": 10265
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2791503686039763e-06,
+      "loss": 0.372,
+      "step": 10266
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2774158268767475e-06,
+      "loss": 0.3974,
+      "step": 10267
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2756816541087967e-06,
+      "loss": 0.3751,
+      "step": 10268
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.273947850395299e-06,
+      "loss": 0.3955,
+      "step": 10269
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2722144158314184e-06,
+      "loss": 0.3969,
+      "step": 10270
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.270481350512288e-06,
+      "loss": 0.3888,
+      "step": 10271
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.268748654533025e-06,
+      "loss": 0.4214,
+      "step": 10272
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2670163279887257e-06,
+      "loss": 0.3953,
+      "step": 10273
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.265284370974462e-06,
+      "loss": 0.3865,
+      "step": 10274
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.263552783585299e-06,
+      "loss": 0.4271,
+      "step": 10275
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2618215659162665e-06,
+      "loss": 0.4038,
+      "step": 10276
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2600907180623818e-06,
+      "loss": 0.4285,
+      "step": 10277
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.258360240118635e-06,
+      "loss": 0.3936,
+      "step": 10278
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.256630132180009e-06,
+      "loss": 0.3539,
+      "step": 10279
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2549003943414537e-06,
+      "loss": 0.3793,
+      "step": 10280
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2531710266979032e-06,
+      "loss": 0.4414,
+      "step": 10281
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2514420293442704e-06,
+      "loss": 0.4287,
+      "step": 10282
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2497134023754517e-06,
+      "loss": 0.43,
+      "step": 10283
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.24798514588632e-06,
+      "loss": 0.3811,
+      "step": 10284
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.246257259971727e-06,
+      "loss": 0.4161,
+      "step": 10285
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.244529744726502e-06,
+      "loss": 0.4335,
+      "step": 10286
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2428026002454636e-06,
+      "loss": 0.3969,
+      "step": 10287
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.241075826623401e-06,
+      "loss": 0.3949,
+      "step": 10288
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.239349423955085e-06,
+      "loss": 0.4229,
+      "step": 10289
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2376233923352627e-06,
+      "loss": 0.3987,
+      "step": 10290
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2358977318586726e-06,
+      "loss": 0.4456,
+      "step": 10291
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2341724426200214e-06,
+      "loss": 0.4208,
+      "step": 10292
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.232447524714e-06,
+      "loss": 0.422,
+      "step": 10293
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2307229782352765e-06,
+      "loss": 0.4242,
+      "step": 10294
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2289988032784967e-06,
+      "loss": 0.4375,
+      "step": 10295
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.227274999938298e-06,
+      "loss": 0.4144,
+      "step": 10296
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.225551568309284e-06,
+      "loss": 0.4021,
+      "step": 10297
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.22382850848604e-06,
+      "loss": 0.4354,
+      "step": 10298
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2221058205631405e-06,
+      "loss": 0.4045,
+      "step": 10299
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2203835046351286e-06,
+      "loss": 0.3862,
+      "step": 10300
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.218661560796531e-06,
+      "loss": 0.4336,
+      "step": 10301
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2169399891418553e-06,
+      "loss": 0.4343,
+      "step": 10302
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2152187897655827e-06,
+      "loss": 0.392,
+      "step": 10303
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2134979627621863e-06,
+      "loss": 0.4108,
+      "step": 10304
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2117775082261083e-06,
+      "loss": 0.4042,
+      "step": 10305
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2100574262517724e-06,
+      "loss": 0.3955,
+      "step": 10306
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2083377169335792e-06,
+      "loss": 0.4183,
+      "step": 10307
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.20661838036592e-06,
+      "loss": 0.3898,
+      "step": 10308
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.204899416643156e-06,
+      "loss": 0.384,
+      "step": 10309
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2031808258596265e-06,
+      "loss": 0.3972,
+      "step": 10310
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2014626081096536e-06,
+      "loss": 0.4216,
+      "step": 10311
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1997447634875444e-06,
+      "loss": 0.3887,
+      "step": 10312
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1980272920875765e-06,
+      "loss": 0.3746,
+      "step": 10313
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.196310194004013e-06,
+      "loss": 0.4336,
+      "step": 10314
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1945934693310897e-06,
+      "loss": 0.4203,
+      "step": 10315
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1928771181630324e-06,
+      "loss": 0.4375,
+      "step": 10316
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.191161140594039e-06,
+      "loss": 0.401,
+      "step": 10317
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1894455367182864e-06,
+      "loss": 0.427,
+      "step": 10318
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1877303066299346e-06,
+      "loss": 0.403,
+      "step": 10319
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.186015450423118e-06,
+      "loss": 0.4412,
+      "step": 10320
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.18430096819196e-06,
+      "loss": 0.4175,
+      "step": 10321
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.182586860030554e-06,
+      "loss": 0.4105,
+      "step": 10322
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1808731260329738e-06,
+      "loss": 0.3951,
+      "step": 10323
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1791597662932817e-06,
+      "loss": 0.4256,
+      "step": 10324
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1774467809055097e-06,
+      "loss": 0.379,
+      "step": 10325
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1757341699636725e-06,
+      "loss": 0.3633,
+      "step": 10326
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1740219335617627e-06,
+      "loss": 0.3739,
+      "step": 10327
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1723100717937528e-06,
+      "loss": 0.4525,
+      "step": 10328
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1705985847536015e-06,
+      "loss": 0.4138,
+      "step": 10329
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.168887472535238e-06,
+      "loss": 0.4308,
+      "step": 10330
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1671767352325723e-06,
+      "loss": 0.4001,
+      "step": 10331
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1654663729394997e-06,
+      "loss": 0.3763,
+      "step": 10332
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1637563857498886e-06,
+      "loss": 0.4384,
+      "step": 10333
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.162046773757591e-06,
+      "loss": 0.4296,
+      "step": 10334
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1603375370564337e-06,
+      "loss": 0.4036,
+      "step": 10335
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1586286757402272e-06,
+      "loss": 0.3817,
+      "step": 10336
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.156920189902758e-06,
+      "loss": 0.4258,
+      "step": 10337
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1552120796377974e-06,
+      "loss": 0.4373,
+      "step": 10338
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.153504345039088e-06,
+      "loss": 0.3791,
+      "step": 10339
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1517969862003617e-06,
+      "loss": 0.3558,
+      "step": 10340
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1500900032153213e-06,
+      "loss": 0.3835,
+      "step": 10341
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1483833961776534e-06,
+      "loss": 0.4181,
+      "step": 10342
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1466771651810223e-06,
+      "loss": 0.3991,
+      "step": 10343
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1449713103190704e-06,
+      "loss": 0.4156,
+      "step": 10344
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1432658316854194e-06,
+      "loss": 0.4206,
+      "step": 10345
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.141560729373678e-06,
+      "loss": 0.4198,
+      "step": 10346
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1398560034774216e-06,
+      "loss": 0.4315,
+      "step": 10347
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1381516540902192e-06,
+      "loss": 0.432,
+      "step": 10348
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1364476813056067e-06,
+      "loss": 0.3826,
+      "step": 10349
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134744085217105e-06,
+      "loss": 0.4372,
+      "step": 10350
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.133040865918213e-06,
+      "loss": 0.4323,
+      "step": 10351
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1313380235024103e-06,
+      "loss": 0.4229,
+      "step": 10352
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1296355580631545e-06,
+      "loss": 0.4062,
+      "step": 10353
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.127933469693879e-06,
+      "loss": 0.387,
+      "step": 10354
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1262317584880054e-06,
+      "loss": 0.4508,
+      "step": 10355
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.124530424538931e-06,
+      "loss": 0.4351,
+      "step": 10356
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1228294679400285e-06,
+      "loss": 0.3888,
+      "step": 10357
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1211288887846526e-06,
+      "loss": 0.3692,
+      "step": 10358
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.119428687166137e-06,
+      "loss": 0.4158,
+      "step": 10359
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1177288631777956e-06,
+      "loss": 0.3733,
+      "step": 10360
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.11602941691292e-06,
+      "loss": 0.4066,
+      "step": 10361
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.114330348464778e-06,
+      "loss": 0.4295,
+      "step": 10362
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1126316579266257e-06,
+      "loss": 0.4123,
+      "step": 10363
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1109333453916933e-06,
+      "loss": 0.4032,
+      "step": 10364
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.109235410953191e-06,
+      "loss": 0.3732,
+      "step": 10365
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1075378547043034e-06,
+      "loss": 0.4087,
+      "step": 10366
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.105840676738201e-06,
+      "loss": 0.3688,
+      "step": 10367
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1041438771480316e-06,
+      "loss": 0.4476,
+      "step": 10368
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.102447456026919e-06,
+      "loss": 0.4331,
+      "step": 10369
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1007514134679673e-06,
+      "loss": 0.3994,
+      "step": 10370
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.099055749564267e-06,
+      "loss": 0.4206,
+      "step": 10371
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0973604644088763e-06,
+      "loss": 0.4399,
+      "step": 10372
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.095665558094845e-06,
+      "loss": 0.3696,
+      "step": 10373
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0939710307151915e-06,
+      "loss": 0.4149,
+      "step": 10374
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.092276882362918e-06,
+      "loss": 0.3935,
+      "step": 10375
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.090583113131006e-06,
+      "loss": 0.3801,
+      "step": 10376
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0888897231124147e-06,
+      "loss": 0.4241,
+      "step": 10377
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0871967124000835e-06,
+      "loss": 0.4249,
+      "step": 10378
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0855040810869274e-06,
+      "loss": 0.4158,
+      "step": 10379
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.083811829265848e-06,
+      "loss": 0.399,
+      "step": 10380
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0821199570297245e-06,
+      "loss": 0.4197,
+      "step": 10381
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.080428464471411e-06,
+      "loss": 0.3828,
+      "step": 10382
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0787373516837406e-06,
+      "loss": 0.4254,
+      "step": 10383
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.077046618759529e-06,
+      "loss": 0.4059,
+      "step": 10384
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.075356265791569e-06,
+      "loss": 0.4239,
+      "step": 10385
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0736662928726337e-06,
+      "loss": 0.4193,
+      "step": 10386
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0719767000954714e-06,
+      "loss": 0.3886,
+      "step": 10387
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0702874875528187e-06,
+      "loss": 0.3995,
+      "step": 10388
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0685986553373805e-06,
+      "loss": 0.3926,
+      "step": 10389
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.066910203541851e-06,
+      "loss": 0.3934,
+      "step": 10390
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.065222132258896e-06,
+      "loss": 0.3662,
+      "step": 10391
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.063534441581162e-06,
+      "loss": 0.4037,
+      "step": 10392
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0618471316012765e-06,
+      "loss": 0.3975,
+      "step": 10393
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.060160202411845e-06,
+      "loss": 0.4301,
+      "step": 10394
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0584736541054517e-06,
+      "loss": 0.4109,
+      "step": 10395
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.056787486774656e-06,
+      "loss": 0.4162,
+      "step": 10396
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0551017005120064e-06,
+      "loss": 0.4251,
+      "step": 10397
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0534162954100264e-06,
+      "loss": 0.4055,
+      "step": 10398
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0517312715612148e-06,
+      "loss": 0.4182,
+      "step": 10399
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0500466290580498e-06,
+      "loss": 0.4088,
+      "step": 10400
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.048362367992993e-06,
+      "loss": 0.4013,
+      "step": 10401
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0466784884584823e-06,
+      "loss": 0.404,
+      "step": 10402
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0449949905469325e-06,
+      "loss": 0.3865,
+      "step": 10403
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0433118743507396e-06,
+      "loss": 0.4025,
+      "step": 10404
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0416291399622834e-06,
+      "loss": 0.3929,
+      "step": 10405
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.039946787473913e-06,
+      "loss": 0.4044,
+      "step": 10406
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.038264816977968e-06,
+      "loss": 0.3856,
+      "step": 10407
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0365832285667574e-06,
+      "loss": 0.4063,
+      "step": 10408
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0349020223325742e-06,
+      "loss": 0.4163,
+      "step": 10409
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.033221198367686e-06,
+      "loss": 0.43,
+      "step": 10410
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.031540756764346e-06,
+      "loss": 0.4016,
+      "step": 10411
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.029860697614777e-06,
+      "loss": 0.4126,
+      "step": 10412
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.028181021011194e-06,
+      "loss": 0.3958,
+      "step": 10413
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0265017270457777e-06,
+      "loss": 0.3765,
+      "step": 10414
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0248228158106983e-06,
+      "loss": 0.4121,
+      "step": 10415
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.023144287398099e-06,
+      "loss": 0.4064,
+      "step": 10416
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.021466141900102e-06,
+      "loss": 0.438,
+      "step": 10417
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.019788379408812e-06,
+      "loss": 0.4348,
+      "step": 10418
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0181110000163084e-06,
+      "loss": 0.3946,
+      "step": 10419
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.01643400381465e-06,
+      "loss": 0.4532,
+      "step": 10420
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0147573908958815e-06,
+      "loss": 0.4655,
+      "step": 10421
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.013081161352015e-06,
+      "loss": 0.3696,
+      "step": 10422
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0114053152750557e-06,
+      "loss": 0.3974,
+      "step": 10423
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0097298527569764e-06,
+      "loss": 0.4007,
+      "step": 10424
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0080547738897304e-06,
+      "loss": 0.3952,
+      "step": 10425
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.006380078765254e-06,
+      "loss": 0.4048,
+      "step": 10426
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.004705767475461e-06,
+      "loss": 0.3845,
+      "step": 10427
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0030318401122384e-06,
+      "loss": 0.419,
+      "step": 10428
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.001358296767465e-06,
+      "loss": 0.4354,
+      "step": 10429
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.999685137532986e-06,
+      "loss": 0.4189,
+      "step": 10430
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9980123625006296e-06,
+      "loss": 0.3808,
+      "step": 10431
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9963399717622075e-06,
+      "loss": 0.4046,
+      "step": 10432
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9946679654095057e-06,
+      "loss": 0.3898,
+      "step": 10433
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.992996343534288e-06,
+      "loss": 0.3556,
+      "step": 10434
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.991325106228299e-06,
+      "loss": 0.4222,
+      "step": 10435
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.989654253583261e-06,
+      "loss": 0.4298,
+      "step": 10436
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.98798378569088e-06,
+      "loss": 0.386,
+      "step": 10437
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9863137026428345e-06,
+      "loss": 0.3892,
+      "step": 10438
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9846440045307836e-06,
+      "loss": 0.4299,
+      "step": 10439
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9829746914463697e-06,
+      "loss": 0.4213,
+      "step": 10440
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9813057634812104e-06,
+      "loss": 0.4429,
+      "step": 10441
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9796372207268997e-06,
+      "loss": 0.414,
+      "step": 10442
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9779690632750145e-06,
+      "loss": 0.3738,
+      "step": 10443
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9763012912171097e-06,
+      "loss": 0.4098,
+      "step": 10444
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9746339046447137e-06,
+      "loss": 0.413,
+      "step": 10445
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.972966903649347e-06,
+      "loss": 0.4271,
+      "step": 10446
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.971300288322495e-06,
+      "loss": 0.4082,
+      "step": 10447
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9696340587556262e-06,
+      "loss": 0.3982,
+      "step": 10448
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9679682150401957e-06,
+      "loss": 0.4089,
+      "step": 10449
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.966302757267625e-06,
+      "loss": 0.3949,
+      "step": 10450
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9646376855293234e-06,
+      "loss": 0.4073,
+      "step": 10451
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.962972999916676e-06,
+      "loss": 0.4093,
+      "step": 10452
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9613087005210406e-06,
+      "loss": 0.3984,
+      "step": 10453
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9596447874337685e-06,
+      "loss": 0.401,
+      "step": 10454
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9579812607461777e-06,
+      "loss": 0.4325,
+      "step": 10455
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.956318120549564e-06,
+      "loss": 0.4274,
+      "step": 10456
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.954655366935214e-06,
+      "loss": 0.4232,
+      "step": 10457
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.952992999994383e-06,
+      "loss": 0.3675,
+      "step": 10458
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9513310198183067e-06,
+      "loss": 0.4312,
+      "step": 10459
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9496694264981995e-06,
+      "loss": 0.4239,
+      "step": 10460
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9480082201252535e-06,
+      "loss": 0.4594,
+      "step": 10461
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9463474007906477e-06,
+      "loss": 0.4187,
+      "step": 10462
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9446869685855315e-06,
+      "loss": 0.4093,
+      "step": 10463
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9430269236010346e-06,
+      "loss": 0.4328,
+      "step": 10464
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.941367265928262e-06,
+      "loss": 0.4461,
+      "step": 10465
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9397079956583084e-06,
+      "loss": 0.4316,
+      "step": 10466
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9380491128822385e-06,
+      "loss": 0.4134,
+      "step": 10467
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9363906176910973e-06,
+      "loss": 0.3951,
+      "step": 10468
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9347325101759038e-06,
+      "loss": 0.4252,
+      "step": 10469
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.93307479042767e-06,
+      "loss": 0.3895,
+      "step": 10470
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.931417458537372e-06,
+      "loss": 0.4027,
+      "step": 10471
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.929760514595972e-06,
+      "loss": 0.4539,
+      "step": 10472
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9281039586944037e-06,
+      "loss": 0.4237,
+      "step": 10473
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9264477909235933e-06,
+      "loss": 0.4249,
+      "step": 10474
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.924792011374432e-06,
+      "loss": 0.3884,
+      "step": 10475
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9231366201377965e-06,
+      "loss": 0.4158,
+      "step": 10476
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.921481617304536e-06,
+      "loss": 0.4059,
+      "step": 10477
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9198270029654895e-06,
+      "loss": 0.3981,
+      "step": 10478
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9181727772114665e-06,
+      "loss": 0.4342,
+      "step": 10479
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.916518940133255e-06,
+      "loss": 0.4684,
+      "step": 10480
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.91486549182162e-06,
+      "loss": 0.3962,
+      "step": 10481
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9132124323673172e-06,
+      "loss": 0.4058,
+      "step": 10482
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.911559761861067e-06,
+      "loss": 0.4343,
+      "step": 10483
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9099074803935755e-06,
+      "loss": 0.4191,
+      "step": 10484
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.908255588055521e-06,
+      "loss": 0.418,
+      "step": 10485
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.906604084937572e-06,
+      "loss": 0.3833,
+      "step": 10486
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.904952971130367e-06,
+      "loss": 0.4021,
+      "step": 10487
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9033022467245233e-06,
+      "loss": 0.4204,
+      "step": 10488
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.901651911810639e-06,
+      "loss": 0.3686,
+      "step": 10489
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9000019664792876e-06,
+      "loss": 0.4015,
+      "step": 10490
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8983524108210303e-06,
+      "loss": 0.36,
+      "step": 10491
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8967032449263975e-06,
+      "loss": 0.4173,
+      "step": 10492
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8950544688858974e-06,
+      "loss": 0.403,
+      "step": 10493
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8934060827900267e-06,
+      "loss": 0.3958,
+      "step": 10494
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.891758086729253e-06,
+      "loss": 0.4227,
+      "step": 10495
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8901104807940226e-06,
+      "loss": 0.3962,
+      "step": 10496
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8884632650747634e-06,
+      "loss": 0.3778,
+      "step": 10497
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.886816439661877e-06,
+      "loss": 0.3943,
+      "step": 10498
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.885170004645752e-06,
+      "loss": 0.4178,
+      "step": 10499
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8835239601167496e-06,
+      "loss": 0.3937,
+      "step": 10500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8818783061652046e-06,
+      "loss": 0.428,
+      "step": 10501
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.880233042881445e-06,
+      "loss": 0.3905,
+      "step": 10502
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.878588170355765e-06,
+      "loss": 0.4099,
+      "step": 10503
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.876943688678441e-06,
+      "loss": 0.37,
+      "step": 10504
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.875299597939727e-06,
+      "loss": 0.4142,
+      "step": 10505
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.873655898229858e-06,
+      "loss": 0.4109,
+      "step": 10506
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8720125896390415e-06,
+      "loss": 0.4026,
+      "step": 10507
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8703696722574757e-06,
+      "loss": 0.4621,
+      "step": 10508
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8687271461753262e-06,
+      "loss": 0.3588,
+      "step": 10509
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8670850114827376e-06,
+      "loss": 0.3879,
+      "step": 10510
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8654432682698418e-06,
+      "loss": 0.3878,
+      "step": 10511
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8638019166267406e-06,
+      "loss": 0.3727,
+      "step": 10512
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8621609566435173e-06,
+      "loss": 0.3938,
+      "step": 10513
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.860520388410234e-06,
+      "loss": 0.4091,
+      "step": 10514
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8588802120169268e-06,
+      "loss": 0.4462,
+      "step": 10515
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8572404275536213e-06,
+      "loss": 0.4051,
+      "step": 10516
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8556010351103124e-06,
+      "loss": 0.3944,
+      "step": 10517
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8539620347769713e-06,
+      "loss": 0.4138,
+      "step": 10518
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8523234266435585e-06,
+      "loss": 0.3936,
+      "step": 10519
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.850685210800004e-06,
+      "loss": 0.3854,
+      "step": 10520
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.849047387336219e-06,
+      "loss": 0.4111,
+      "step": 10521
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.847409956342092e-06,
+      "loss": 0.3781,
+      "step": 10522
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8457729179074923e-06,
+      "loss": 0.417,
+      "step": 10523
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.844136272122263e-06,
+      "loss": 0.433,
+      "step": 10524
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.842500019076235e-06,
+      "loss": 0.4097,
+      "step": 10525
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8408641588592056e-06,
+      "loss": 0.3851,
+      "step": 10526
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8392286915609635e-06,
+      "loss": 0.4134,
+      "step": 10527
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8375936172712638e-06,
+      "loss": 0.4143,
+      "step": 10528
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8359589360798466e-06,
+      "loss": 0.4456,
+      "step": 10529
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8343246480764296e-06,
+      "loss": 0.434,
+      "step": 10530
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8326907533507074e-06,
+      "loss": 0.4463,
+      "step": 10531
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8310572519923507e-06,
+      "loss": 0.3837,
+      "step": 10532
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8294241440910187e-06,
+      "loss": 0.3781,
+      "step": 10533
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8277914297363363e-06,
+      "loss": 0.3833,
+      "step": 10534
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8261591090179177e-06,
+      "loss": 0.4279,
+      "step": 10535
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8245271820253474e-06,
+      "loss": 0.4222,
+      "step": 10536
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8228956488481917e-06,
+      "loss": 0.3994,
+      "step": 10537
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8212645095759963e-06,
+      "loss": 0.4358,
+      "step": 10538
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.819633764298282e-06,
+      "loss": 0.4206,
+      "step": 10539
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8180034131045463e-06,
+      "loss": 0.3744,
+      "step": 10540
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.816373456084277e-06,
+      "loss": 0.3937,
+      "step": 10541
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8147438933269257e-06,
+      "loss": 0.427,
+      "step": 10542
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8131147249219325e-06,
+      "loss": 0.4061,
+      "step": 10543
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.811485950958711e-06,
+      "loss": 0.4432,
+      "step": 10544
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.809857571526653e-06,
+      "loss": 0.4501,
+      "step": 10545
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.80822958671513e-06,
+      "loss": 0.3887,
+      "step": 10546
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8066019966134907e-06,
+      "loss": 0.4004,
+      "step": 10547
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8049748013110646e-06,
+      "loss": 0.4291,
+      "step": 10548
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.803348000897155e-06,
+      "loss": 0.405,
+      "step": 10549
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8017215954610475e-06,
+      "loss": 0.3907,
+      "step": 10550
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.80009558509201e-06,
+      "loss": 0.4594,
+      "step": 10551
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7984699698792793e-06,
+      "loss": 0.4247,
+      "step": 10552
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7968447499120767e-06,
+      "loss": 0.441,
+      "step": 10553
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7952199252795985e-06,
+      "loss": 0.4156,
+      "step": 10554
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.793595496071021e-06,
+      "loss": 0.4104,
+      "step": 10555
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.791971462375499e-06,
+      "loss": 0.3979,
+      "step": 10556
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7903478242821624e-06,
+      "loss": 0.4104,
+      "step": 10557
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.788724581880128e-06,
+      "loss": 0.3873,
+      "step": 10558
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7871017352584785e-06,
+      "loss": 0.3829,
+      "step": 10559
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.785479284506288e-06,
+      "loss": 0.453,
+      "step": 10560
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7838572297126e-06,
+      "loss": 0.4172,
+      "step": 10561
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.782235570966438e-06,
+      "loss": 0.4464,
+      "step": 10562
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.780614308356805e-06,
+      "loss": 0.4292,
+      "step": 10563
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7789934419726804e-06,
+      "loss": 0.4091,
+      "step": 10564
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.777372971903024e-06,
+      "loss": 0.436,
+      "step": 10565
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7757528982367688e-06,
+      "loss": 0.4344,
+      "step": 10566
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7741332210628346e-06,
+      "loss": 0.3923,
+      "step": 10567
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7725139404701184e-06,
+      "loss": 0.3826,
+      "step": 10568
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.770895056547488e-06,
+      "loss": 0.3853,
+      "step": 10569
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.769276569383793e-06,
+      "loss": 0.3997,
+      "step": 10570
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.767658479067863e-06,
+      "loss": 0.3674,
+      "step": 10571
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7660407856885043e-06,
+      "loss": 0.4207,
+      "step": 10572
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7644234893345002e-06,
+      "loss": 0.4086,
+      "step": 10573
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7628065900946123e-06,
+      "loss": 0.449,
+      "step": 10574
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.761190088057587e-06,
+      "loss": 0.4128,
+      "step": 10575
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7595739833121384e-06,
+      "loss": 0.3877,
+      "step": 10576
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.75795827594697e-06,
+      "loss": 0.4178,
+      "step": 10577
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7563429660507534e-06,
+      "loss": 0.4073,
+      "step": 10578
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7547280537121435e-06,
+      "loss": 0.4471,
+      "step": 10579
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7531135390197728e-06,
+      "loss": 0.4072,
+      "step": 10580
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.75149942206225e-06,
+      "loss": 0.4441,
+      "step": 10581
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7498857029281645e-06,
+      "loss": 0.4399,
+      "step": 10582
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.748272381706081e-06,
+      "loss": 0.4206,
+      "step": 10583
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.746659458484545e-06,
+      "loss": 0.4341,
+      "step": 10584
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7450469333520856e-06,
+      "loss": 0.39,
+      "step": 10585
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7434348063971984e-06,
+      "loss": 0.3913,
+      "step": 10586
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.741823077708362e-06,
+      "loss": 0.4033,
+      "step": 10587
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.740211747374036e-06,
+      "loss": 0.3852,
+      "step": 10588
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.738600815482655e-06,
+      "loss": 0.4446,
+      "step": 10589
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7369902821226333e-06,
+      "loss": 0.409,
+      "step": 10590
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.735380147382358e-06,
+      "loss": 0.4302,
+      "step": 10591
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.733770411350204e-06,
+      "loss": 0.401,
+      "step": 10592
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.732161074114521e-06,
+      "loss": 0.4407,
+      "step": 10593
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7305521357636322e-06,
+      "loss": 0.4244,
+      "step": 10594
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7289435963858426e-06,
+      "loss": 0.4048,
+      "step": 10595
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7273354560694343e-06,
+      "loss": 0.4117,
+      "step": 10596
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7257277149026673e-06,
+      "loss": 0.4093,
+      "step": 10597
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7241203729737797e-06,
+      "loss": 0.3697,
+      "step": 10598
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7225134303709864e-06,
+      "loss": 0.4329,
+      "step": 10599
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7209068871824883e-06,
+      "loss": 0.3838,
+      "step": 10600
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7193007434964513e-06,
+      "loss": 0.3864,
+      "step": 10601
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7176949994010316e-06,
+      "loss": 0.38,
+      "step": 10602
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.716089654984356e-06,
+      "loss": 0.429,
+      "step": 10603
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.714484710334533e-06,
+      "loss": 0.4192,
+      "step": 10604
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.712880165539644e-06,
+      "loss": 0.4273,
+      "step": 10605
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.711276020687755e-06,
+      "loss": 0.4235,
+      "step": 10606
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7096722758669047e-06,
+      "loss": 0.4181,
+      "step": 10607
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7080689311651155e-06,
+      "loss": 0.4305,
+      "step": 10608
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7064659866703803e-06,
+      "loss": 0.435,
+      "step": 10609
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.704863442470681e-06,
+      "loss": 0.3731,
+      "step": 10610
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.703261298653966e-06,
+      "loss": 0.4186,
+      "step": 10611
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.701659555308169e-06,
+      "loss": 0.4071,
+      "step": 10612
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.700058212521197e-06,
+      "loss": 0.3897,
+      "step": 10613
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.69845727038094e-06,
+      "loss": 0.3923,
+      "step": 10614
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6968567289752578e-06,
+      "loss": 0.4156,
+      "step": 10615
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6952565883920013e-06,
+      "loss": 0.4323,
+      "step": 10616
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6936568487189886e-06,
+      "loss": 0.4152,
+      "step": 10617
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.692057510044015e-06,
+      "loss": 0.3784,
+      "step": 10618
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.690458572454866e-06,
+      "loss": 0.377,
+      "step": 10619
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6888600360392927e-06,
+      "loss": 0.4369,
+      "step": 10620
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6872619008850274e-06,
+      "loss": 0.3829,
+      "step": 10621
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6856641670797834e-06,
+      "loss": 0.4364,
+      "step": 10622
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6840668347112464e-06,
+      "loss": 0.3912,
+      "step": 10623
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6824699038670898e-06,
+      "loss": 0.3861,
+      "step": 10624
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6808733746349556e-06,
+      "loss": 0.4083,
+      "step": 10625
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.679277247102463e-06,
+      "loss": 0.413,
+      "step": 10626
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6776815213572205e-06,
+      "loss": 0.3659,
+      "step": 10627
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6760861974868037e-06,
+      "loss": 0.4019,
+      "step": 10628
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.67449127557877e-06,
+      "loss": 0.4003,
+      "step": 10629
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6728967557206544e-06,
+      "loss": 0.4071,
+      "step": 10630
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.671302637999965e-06,
+      "loss": 0.3825,
+      "step": 10631
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.669708922504202e-06,
+      "loss": 0.4191,
+      "step": 10632
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6681156093208282e-06,
+      "loss": 0.416,
+      "step": 10633
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6665226985372905e-06,
+      "loss": 0.4382,
+      "step": 10634
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6649301902410128e-06,
+      "loss": 0.423,
+      "step": 10635
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.663338084519401e-06,
+      "loss": 0.375,
+      "step": 10636
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6617463814598354e-06,
+      "loss": 0.4148,
+      "step": 10637
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.660155081149671e-06,
+      "loss": 0.4097,
+      "step": 10638
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.658564183676243e-06,
+      "loss": 0.3581,
+      "step": 10639
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6569736891268714e-06,
+      "loss": 0.3788,
+      "step": 10640
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6553835975888443e-06,
+      "loss": 0.4162,
+      "step": 10641
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6537939091494324e-06,
+      "loss": 0.4176,
+      "step": 10642
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.652204623895881e-06,
+      "loss": 0.3808,
+      "step": 10643
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6506157419154198e-06,
+      "loss": 0.4333,
+      "step": 10644
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6490272632952506e-06,
+      "loss": 0.4157,
+      "step": 10645
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6474391881225548e-06,
+      "loss": 0.4054,
+      "step": 10646
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6458515164844912e-06,
+      "loss": 0.4053,
+      "step": 10647
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6442642484681946e-06,
+      "loss": 0.4234,
+      "step": 10648
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6426773841607845e-06,
+      "loss": 0.3597,
+      "step": 10649
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.641090923649352e-06,
+      "loss": 0.4157,
+      "step": 10650
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6395048670209635e-06,
+      "loss": 0.3916,
+      "step": 10651
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6379192143626743e-06,
+      "loss": 0.3825,
+      "step": 10652
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.636333965761507e-06,
+      "loss": 0.4219,
+      "step": 10653
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.634749121304466e-06,
+      "loss": 0.4265,
+      "step": 10654
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6331646810785337e-06,
+      "loss": 0.4003,
+      "step": 10655
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6315806451706673e-06,
+      "loss": 0.3744,
+      "step": 10656
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.629997013667808e-06,
+      "loss": 0.4588,
+      "step": 10657
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6284137866568693e-06,
+      "loss": 0.4185,
+      "step": 10658
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.626830964224746e-06,
+      "loss": 0.3953,
+      "step": 10659
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6252485464583033e-06,
+      "loss": 0.3997,
+      "step": 10660
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6236665334443967e-06,
+      "loss": 0.3902,
+      "step": 10661
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.622084925269851e-06,
+      "loss": 0.4057,
+      "step": 10662
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.620503722021469e-06,
+      "loss": 0.4144,
+      "step": 10663
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6189229237860303e-06,
+      "loss": 0.4433,
+      "step": 10664
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.617342530650301e-06,
+      "loss": 0.4133,
+      "step": 10665
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6157625427010158e-06,
+      "loss": 0.4271,
+      "step": 10666
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.614182960024889e-06,
+      "loss": 0.4083,
+      "step": 10667
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6126037827086114e-06,
+      "loss": 0.3859,
+      "step": 10668
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6110250108388603e-06,
+      "loss": 0.4341,
+      "step": 10669
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6094466445022805e-06,
+      "loss": 0.3989,
+      "step": 10670
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6078686837854995e-06,
+      "loss": 0.4152,
+      "step": 10671
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.606291128775116e-06,
+      "loss": 0.4371,
+      "step": 10672
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.60471397955772e-06,
+      "loss": 0.3944,
+      "step": 10673
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6031372362198682e-06,
+      "loss": 0.4221,
+      "step": 10674
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6015608988480956e-06,
+      "loss": 0.4378,
+      "step": 10675
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5999849675289203e-06,
+      "loss": 0.3708,
+      "step": 10676
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5984094423488295e-06,
+      "loss": 0.4117,
+      "step": 10677
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5968343233943006e-06,
+      "loss": 0.42,
+      "step": 10678
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.595259610751778e-06,
+      "loss": 0.4459,
+      "step": 10679
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.593685304507686e-06,
+      "loss": 0.4291,
+      "step": 10680
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5921114047484332e-06,
+      "loss": 0.39,
+      "step": 10681
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.590537911560397e-06,
+      "loss": 0.4052,
+      "step": 10682
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5889648250299383e-06,
+      "loss": 0.3824,
+      "step": 10683
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5873921452433913e-06,
+      "loss": 0.4039,
+      "step": 10684
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5858198722870696e-06,
+      "loss": 0.3952,
+      "step": 10685
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5842480062472697e-06,
+      "loss": 0.4225,
+      "step": 10686
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5826765472102577e-06,
+      "loss": 0.4435,
+      "step": 10687
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5811054952622794e-06,
+      "loss": 0.4136,
+      "step": 10688
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5795348504895643e-06,
+      "loss": 0.4151,
+      "step": 10689
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.577964612978313e-06,
+      "loss": 0.4121,
+      "step": 10690
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.576394782814705e-06,
+      "loss": 0.3909,
+      "step": 10691
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5748253600848994e-06,
+      "loss": 0.3891,
+      "step": 10692
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5732563448750304e-06,
+      "loss": 0.3824,
+      "step": 10693
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.571687737271208e-06,
+      "loss": 0.4367,
+      "step": 10694
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.570119537359529e-06,
+      "loss": 0.4277,
+      "step": 10695
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5685517452260566e-06,
+      "loss": 0.4304,
+      "step": 10696
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5669843609568425e-06,
+      "loss": 0.3892,
+      "step": 10697
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5654173846379073e-06,
+      "loss": 0.4182,
+      "step": 10698
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.563850816355251e-06,
+      "loss": 0.3901,
+      "step": 10699
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5622846561948543e-06,
+      "loss": 0.422,
+      "step": 10700
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.560718904242674e-06,
+      "loss": 0.4103,
+      "step": 10701
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5591535605846383e-06,
+      "loss": 0.4358,
+      "step": 10702
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5575886253066674e-06,
+      "loss": 0.4322,
+      "step": 10703
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5560240984946437e-06,
+      "loss": 0.4595,
+      "step": 10704
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5544599802344396e-06,
+      "loss": 0.4133,
+      "step": 10705
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5528962706118966e-06,
+      "loss": 0.4115,
+      "step": 10706
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5513329697128376e-06,
+      "loss": 0.3883,
+      "step": 10707
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5497700776230604e-06,
+      "loss": 0.4226,
+      "step": 10708
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5482075944283423e-06,
+      "loss": 0.406,
+      "step": 10709
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.546645520214437e-06,
+      "loss": 0.4034,
+      "step": 10710
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5450838550670808e-06,
+      "loss": 0.3881,
+      "step": 10711
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.543522599071978e-06,
+      "loss": 0.4463,
+      "step": 10712
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.541961752314821e-06,
+      "loss": 0.4299,
+      "step": 10713
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5404013148812736e-06,
+      "loss": 0.3957,
+      "step": 10714
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5388412868569756e-06,
+      "loss": 0.4187,
+      "step": 10715
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.537281668327549e-06,
+      "loss": 0.4329,
+      "step": 10716
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5357224593785902e-06,
+      "loss": 0.4175,
+      "step": 10717
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.534163660095674e-06,
+      "loss": 0.403,
+      "step": 10718
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.53260527056435e-06,
+      "loss": 0.4312,
+      "step": 10719
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5310472908701555e-06,
+      "loss": 0.4162,
+      "step": 10720
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.529489721098589e-06,
+      "loss": 0.3847,
+      "step": 10721
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5279325613351445e-06,
+      "loss": 0.4106,
+      "step": 10722
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5263758116652805e-06,
+      "loss": 0.3996,
+      "step": 10723
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5248194721744357e-06,
+      "loss": 0.4347,
+      "step": 10724
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.523263542948028e-06,
+      "loss": 0.4015,
+      "step": 10725
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.521708024071453e-06,
+      "loss": 0.3902,
+      "step": 10726
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5201529156300807e-06,
+      "loss": 0.3714,
+      "step": 10727
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.518598217709266e-06,
+      "loss": 0.4488,
+      "step": 10728
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5170439303943295e-06,
+      "loss": 0.4254,
+      "step": 10729
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.515490053770583e-06,
+      "loss": 0.3827,
+      "step": 10730
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5139365879233058e-06,
+      "loss": 0.403,
+      "step": 10731
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5123835329377576e-06,
+      "loss": 0.4351,
+      "step": 10732
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5108308888991763e-06,
+      "loss": 0.4273,
+      "step": 10733
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.509278655892774e-06,
+      "loss": 0.4317,
+      "step": 10734
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5077268340037454e-06,
+      "loss": 0.4077,
+      "step": 10735
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5061754233172555e-06,
+      "loss": 0.4288,
+      "step": 10736
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5046244239184547e-06,
+      "loss": 0.393,
+      "step": 10737
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.503073835892471e-06,
+      "loss": 0.3897,
+      "step": 10738
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.501523659324402e-06,
+      "loss": 0.3729,
+      "step": 10739
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.499973894299328e-06,
+      "loss": 0.3983,
+      "step": 10740
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4984245409023046e-06,
+      "loss": 0.3967,
+      "step": 10741
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.496875599218366e-06,
+      "loss": 0.4291,
+      "step": 10742
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.495327069332524e-06,
+      "loss": 0.4223,
+      "step": 10743
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.493778951329764e-06,
+      "loss": 0.4035,
+      "step": 10744
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.492231245295058e-06,
+      "loss": 0.4021,
+      "step": 10745
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.490683951313345e-06,
+      "loss": 0.4034,
+      "step": 10746
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4891370694695516e-06,
+      "loss": 0.4146,
+      "step": 10747
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4875905998485716e-06,
+      "loss": 0.4188,
+      "step": 10748
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.486044542535282e-06,
+      "loss": 0.4136,
+      "step": 10749
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4844988976145358e-06,
+      "loss": 0.4181,
+      "step": 10750
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.482953665171164e-06,
+      "loss": 0.4278,
+      "step": 10751
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.481408845289973e-06,
+      "loss": 0.3972,
+      "step": 10752
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.479864438055746e-06,
+      "loss": 0.4312,
+      "step": 10753
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4783204435532482e-06,
+      "loss": 0.4322,
+      "step": 10754
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.476776861867223e-06,
+      "loss": 0.4536,
+      "step": 10755
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.475233693082384e-06,
+      "loss": 0.3726,
+      "step": 10756
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4736909372834263e-06,
+      "loss": 0.395,
+      "step": 10757
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4721485945550216e-06,
+      "loss": 0.3754,
+      "step": 10758
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4706066649818185e-06,
+      "loss": 0.3941,
+      "step": 10759
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4690651486484452e-06,
+      "loss": 0.4154,
+      "step": 10760
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.467524045639501e-06,
+      "loss": 0.4354,
+      "step": 10761
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.465983356039573e-06,
+      "loss": 0.4068,
+      "step": 10762
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4644430799332153e-06,
+      "loss": 0.4246,
+      "step": 10763
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4629032174049693e-06,
+      "loss": 0.3825,
+      "step": 10764
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4613637685393433e-06,
+      "loss": 0.4181,
+      "step": 10765
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.45982473342083e-06,
+      "loss": 0.4335,
+      "step": 10766
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4582861121338963e-06,
+      "loss": 0.3855,
+      "step": 10767
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.456747904762986e-06,
+      "loss": 0.4379,
+      "step": 10768
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4552101113925207e-06,
+      "loss": 0.4184,
+      "step": 10769
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.453672732106904e-06,
+      "loss": 0.4064,
+      "step": 10770
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4521357669905078e-06,
+      "loss": 0.3851,
+      "step": 10771
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4505992161276915e-06,
+      "loss": 0.3823,
+      "step": 10772
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4490630796027838e-06,
+      "loss": 0.3928,
+      "step": 10773
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4475273575000936e-06,
+      "loss": 0.4582,
+      "step": 10774
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.445992049903906e-06,
+      "loss": 0.3919,
+      "step": 10775
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.444457156898485e-06,
+      "loss": 0.4179,
+      "step": 10776
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4429226785680684e-06,
+      "loss": 0.4079,
+      "step": 10777
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4413886149968778e-06,
+      "loss": 0.3976,
+      "step": 10778
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.439854966269103e-06,
+      "loss": 0.401,
+      "step": 10779
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4383217324689234e-06,
+      "loss": 0.4114,
+      "step": 10780
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4367889136804834e-06,
+      "loss": 0.4404,
+      "step": 10781
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.435256509987911e-06,
+      "loss": 0.3721,
+      "step": 10782
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4337245214753104e-06,
+      "loss": 0.3833,
+      "step": 10783
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4321929482267592e-06,
+      "loss": 0.4078,
+      "step": 10784
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.43066179032632e-06,
+      "loss": 0.4047,
+      "step": 10785
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.429131047858022e-06,
+      "loss": 0.4097,
+      "step": 10786
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.427600720905885e-06,
+      "loss": 0.3998,
+      "step": 10787
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4260708095538933e-06,
+      "loss": 0.4191,
+      "step": 10788
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4245413138860186e-06,
+      "loss": 0.4241,
+      "step": 10789
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4230122339862015e-06,
+      "loss": 0.3956,
+      "step": 10790
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4214835699383664e-06,
+      "loss": 0.3934,
+      "step": 10791
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4199553218264095e-06,
+      "loss": 0.4136,
+      "step": 10792
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4184274897342066e-06,
+      "loss": 0.4132,
+      "step": 10793
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4169000737456074e-06,
+      "loss": 0.4529,
+      "step": 10794
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.415373073944449e-06,
+      "loss": 0.4456,
+      "step": 10795
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.413846490414531e-06,
+      "loss": 0.4051,
+      "step": 10796
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4123203232396454e-06,
+      "loss": 0.3945,
+      "step": 10797
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4107945725035497e-06,
+      "loss": 0.409,
+      "step": 10798
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.409269238289983e-06,
+      "loss": 0.4495,
+      "step": 10799
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.40774432068266e-06,
+      "loss": 0.4228,
+      "step": 10800
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.406219819765275e-06,
+      "loss": 0.4046,
+      "step": 10801
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.404695735621495e-06,
+      "loss": 0.417,
+      "step": 10802
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4031720683349715e-06,
+      "loss": 0.3684,
+      "step": 10803
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4016488179893272e-06,
+      "loss": 0.3991,
+      "step": 10804
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4001259846681603e-06,
+      "loss": 0.4197,
+      "step": 10805
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3986035684550545e-06,
+      "loss": 0.4033,
+      "step": 10806
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3970815694335625e-06,
+      "loss": 0.4227,
+      "step": 10807
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3955599876872184e-06,
+      "loss": 0.3936,
+      "step": 10808
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3940388232995305e-06,
+      "loss": 0.3828,
+      "step": 10809
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3925180763539847e-06,
+      "loss": 0.3785,
+      "step": 10810
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3909977469340484e-06,
+      "loss": 0.44,
+      "step": 10811
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3894778351231606e-06,
+      "loss": 0.407,
+      "step": 10812
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.387958341004737e-06,
+      "loss": 0.3867,
+      "step": 10813
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3864392646621793e-06,
+      "loss": 0.3856,
+      "step": 10814
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3849206061788555e-06,
+      "loss": 0.3631,
+      "step": 10815
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3834023656381143e-06,
+      "loss": 0.4193,
+      "step": 10816
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.381884543123285e-06,
+      "loss": 0.4529,
+      "step": 10817
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.380367138717666e-06,
+      "loss": 0.3958,
+      "step": 10818
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.378850152504544e-06,
+      "loss": 0.4323,
+      "step": 10819
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.377333584567174e-06,
+      "loss": 0.4434,
+      "step": 10820
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.375817434988791e-06,
+      "loss": 0.3875,
+      "step": 10821
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.374301703852603e-06,
+      "loss": 0.3777,
+      "step": 10822
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3727863912418036e-06,
+      "loss": 0.4359,
+      "step": 10823
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3712714972395578e-06,
+      "loss": 0.376,
+      "step": 10824
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.369757021929008e-06,
+      "loss": 0.402,
+      "step": 10825
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3682429653932693e-06,
+      "loss": 0.3755,
+      "step": 10826
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.366729327715446e-06,
+      "loss": 0.4177,
+      "step": 10827
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3652161089786087e-06,
+      "loss": 0.3905,
+      "step": 10828
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.363703309265807e-06,
+      "loss": 0.4125,
+      "step": 10829
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3621909286600675e-06,
+      "loss": 0.4138,
+      "step": 10830
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3606789672443997e-06,
+      "loss": 0.4009,
+      "step": 10831
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3591674251017825e-06,
+      "loss": 0.4027,
+      "step": 10832
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3576563023151767e-06,
+      "loss": 0.4213,
+      "step": 10833
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3561455989675117e-06,
+      "loss": 0.4243,
+      "step": 10834
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.354635315141709e-06,
+      "loss": 0.4092,
+      "step": 10835
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3531254509206534e-06,
+      "loss": 0.3965,
+      "step": 10836
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.351616006387214e-06,
+      "loss": 0.4095,
+      "step": 10837
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3501069816242294e-06,
+      "loss": 0.3892,
+      "step": 10838
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3485983767145272e-06,
+      "loss": 0.3964,
+      "step": 10839
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3470901917409006e-06,
+      "loss": 0.4154,
+      "step": 10840
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3455824267861262e-06,
+      "loss": 0.3995,
+      "step": 10841
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3440750819329516e-06,
+      "loss": 0.4175,
+      "step": 10842
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.342568157264111e-06,
+      "loss": 0.3702,
+      "step": 10843
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3410616528623064e-06,
+      "loss": 0.4248,
+      "step": 10844
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.4142,
+      "step": 10845
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3380499051905138e-06,
+      "loss": 0.3877,
+      "step": 10846
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.336544662085818e-06,
+      "loss": 0.4314,
+      "step": 10847
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.335039839578752e-06,
+      "loss": 0.4221,
+      "step": 10848
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.333535437751904e-06,
+      "loss": 0.3842,
+      "step": 10849
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3320314566878365e-06,
+      "loss": 0.4168,
+      "step": 10850
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3305278964691003e-06,
+      "loss": 0.3952,
+      "step": 10851
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.329024757178212e-06,
+      "loss": 0.3889,
+      "step": 10852
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3275220388976714e-06,
+      "loss": 0.413,
+      "step": 10853
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.326019741709952e-06,
+      "loss": 0.3805,
+      "step": 10854
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.324517865697501e-06,
+      "loss": 0.4243,
+      "step": 10855
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3230164109427532e-06,
+      "loss": 0.412,
+      "step": 10856
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.321515377528112e-06,
+      "loss": 0.4105,
+      "step": 10857
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3200147655359575e-06,
+      "loss": 0.4212,
+      "step": 10858
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.318514575048648e-06,
+      "loss": 0.3957,
+      "step": 10859
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.317014806148523e-06,
+      "loss": 0.426,
+      "step": 10860
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.315515458917894e-06,
+      "loss": 0.4182,
+      "step": 10861
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3140165334390485e-06,
+      "loss": 0.4374,
+      "step": 10862
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.312518029794255e-06,
+      "loss": 0.404,
+      "step": 10863
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3110199480657525e-06,
+      "loss": 0.4339,
+      "step": 10864
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3095222883357673e-06,
+      "loss": 0.4014,
+      "step": 10865
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3080250506864933e-06,
+      "loss": 0.4191,
+      "step": 10866
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3065282352001004e-06,
+      "loss": 0.4216,
+      "step": 10867
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3050318419587457e-06,
+      "loss": 0.4067,
+      "step": 10868
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.303535871044554e-06,
+      "loss": 0.3956,
+      "step": 10869
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3020403225396304e-06,
+      "loss": 0.3996,
+      "step": 10870
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.300545196526054e-06,
+      "loss": 0.4242,
+      "step": 10871
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2990504930858794e-06,
+      "loss": 0.4154,
+      "step": 10872
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2975562123011497e-06,
+      "loss": 0.3847,
+      "step": 10873
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2960623542538705e-06,
+      "loss": 0.432,
+      "step": 10874
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.294568919026029e-06,
+      "loss": 0.3824,
+      "step": 10875
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2930759066995956e-06,
+      "loss": 0.3975,
+      "step": 10876
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2915833173565083e-06,
+      "loss": 0.4057,
+      "step": 10877
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2900911510786873e-06,
+      "loss": 0.3869,
+      "step": 10878
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.288599407948027e-06,
+      "loss": 0.4227,
+      "step": 10879
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.287108088046399e-06,
+      "loss": 0.4078,
+      "step": 10880
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.285617191455651e-06,
+      "loss": 0.4228,
+      "step": 10881
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2841267182576145e-06,
+      "loss": 0.4014,
+      "step": 10882
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2826366685340826e-06,
+      "loss": 0.4282,
+      "step": 10883
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2811470423668447e-06,
+      "loss": 0.4344,
+      "step": 10884
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2796578398376525e-06,
+      "loss": 0.437,
+      "step": 10885
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.278169061028237e-06,
+      "loss": 0.4387,
+      "step": 10886
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.27668070602031e-06,
+      "loss": 0.4044,
+      "step": 10887
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.275192774895556e-06,
+      "loss": 0.4433,
+      "step": 10888
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.273705267735635e-06,
+      "loss": 0.4223,
+      "step": 10889
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2722181846221948e-06,
+      "loss": 0.3812,
+      "step": 10890
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2707315256368434e-06,
+      "loss": 0.4339,
+      "step": 10891
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2692452908611818e-06,
+      "loss": 0.4172,
+      "step": 10892
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2677594803767743e-06,
+      "loss": 0.3636,
+      "step": 10893
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2662740942651705e-06,
+      "loss": 0.4199,
+      "step": 10894
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.264789132607892e-06,
+      "loss": 0.4514,
+      "step": 10895
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2633045954864373e-06,
+      "loss": 0.4216,
+      "step": 10896
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.261820482982283e-06,
+      "loss": 0.4235,
+      "step": 10897
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.260336795176887e-06,
+      "loss": 0.423,
+      "step": 10898
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.258853532151675e-06,
+      "loss": 0.3914,
+      "step": 10899
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.257370693988056e-06,
+      "loss": 0.423,
+      "step": 10900
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2558882807674132e-06,
+      "loss": 0.4151,
+      "step": 10901
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2544062925711064e-06,
+      "loss": 0.3832,
+      "step": 10902
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2529247294804725e-06,
+      "loss": 0.4154,
+      "step": 10903
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.251443591576824e-06,
+      "loss": 0.3676,
+      "step": 10904
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.249962878941452e-06,
+      "loss": 0.4282,
+      "step": 10905
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2484825916556207e-06,
+      "loss": 0.379,
+      "step": 10906
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2470027298005748e-06,
+      "loss": 0.3916,
+      "step": 10907
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2455232934575376e-06,
+      "loss": 0.4059,
+      "step": 10908
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.244044282707705e-06,
+      "loss": 0.3862,
+      "step": 10909
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2425656976322473e-06,
+      "loss": 0.4115,
+      "step": 10910
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2410875383123166e-06,
+      "loss": 0.4172,
+      "step": 10911
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2396098048290395e-06,
+      "loss": 0.3976,
+      "step": 10912
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2381324972635177e-06,
+      "loss": 0.3978,
+      "step": 10913
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.23665561569683e-06,
+      "loss": 0.4109,
+      "step": 10914
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.235179160210037e-06,
+      "loss": 0.4084,
+      "step": 10915
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.233703130884168e-06,
+      "loss": 0.413,
+      "step": 10916
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.232227527800237e-06,
+      "loss": 0.4461,
+      "step": 10917
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.230752351039228e-06,
+      "loss": 0.3871,
+      "step": 10918
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.229277600682104e-06,
+      "loss": 0.4184,
+      "step": 10919
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2278032768098034e-06,
+      "loss": 0.4262,
+      "step": 10920
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2263293795032436e-06,
+      "loss": 0.404,
+      "step": 10921
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2248559088433177e-06,
+      "loss": 0.3777,
+      "step": 10922
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.22338286491089e-06,
+      "loss": 0.3997,
+      "step": 10923
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2219102477868126e-06,
+      "loss": 0.4117,
+      "step": 10924
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.220438057551908e-06,
+      "loss": 0.4187,
+      "step": 10925
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.218966294286973e-06,
+      "loss": 0.4245,
+      "step": 10926
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2174949580727835e-06,
+      "loss": 0.3895,
+      "step": 10927
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2160240489900918e-06,
+      "loss": 0.397,
+      "step": 10928
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.214553567119626e-06,
+      "loss": 0.4221,
+      "step": 10929
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2130835125420926e-06,
+      "loss": 0.4011,
+      "step": 10930
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2116138853381686e-06,
+      "loss": 0.3847,
+      "step": 10931
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.21014468558852e-06,
+      "loss": 0.4086,
+      "step": 10932
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2086759133737744e-06,
+      "loss": 0.4074,
+      "step": 10933
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.207207568774551e-06,
+      "loss": 0.4162,
+      "step": 10934
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2057396518714324e-06,
+      "loss": 0.4285,
+      "step": 10935
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2042721627449848e-06,
+      "loss": 0.4079,
+      "step": 10936
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.202805101475749e-06,
+      "loss": 0.3637,
+      "step": 10937
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.201338468144242e-06,
+      "loss": 0.4479,
+      "step": 10938
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1998722628309587e-06,
+      "loss": 0.4158,
+      "step": 10939
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1984064856163657e-06,
+      "loss": 0.4187,
+      "step": 10940
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1969411365809124e-06,
+      "loss": 0.4358,
+      "step": 10941
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.195476215805028e-06,
+      "loss": 0.3869,
+      "step": 10942
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1940117233691073e-06,
+      "loss": 0.4339,
+      "step": 10943
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1925476593535265e-06,
+      "loss": 0.3965,
+      "step": 10944
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.19108402383864e-06,
+      "loss": 0.3666,
+      "step": 10945
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.189620816904777e-06,
+      "loss": 0.3924,
+      "step": 10946
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.188158038632244e-06,
+      "loss": 0.4243,
+      "step": 10947
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1866956891013192e-06,
+      "loss": 0.4245,
+      "step": 10948
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1852337683922664e-06,
+      "loss": 0.4318,
+      "step": 10949
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.183772276585321e-06,
+      "loss": 0.4198,
+      "step": 10950
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.182311213760695e-06,
+      "loss": 0.4233,
+      "step": 10951
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1808505799985745e-06,
+      "loss": 0.3891,
+      "step": 10952
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1793903753791258e-06,
+      "loss": 0.3829,
+      "step": 10953
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1779305999824884e-06,
+      "loss": 0.4553,
+      "step": 10954
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1764712538887824e-06,
+      "loss": 0.401,
+      "step": 10955
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1750123371780964e-06,
+      "loss": 0.4154,
+      "step": 10956
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1735538499305076e-06,
+      "loss": 0.4283,
+      "step": 10957
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.172095792226059e-06,
+      "loss": 0.4037,
+      "step": 10958
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.170638164144777e-06,
+      "loss": 0.4094,
+      "step": 10959
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1691809657666595e-06,
+      "loss": 0.3838,
+      "step": 10960
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1677241971716843e-06,
+      "loss": 0.4049,
+      "step": 10961
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1662678584398024e-06,
+      "loss": 0.3745,
+      "step": 10962
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1648119496509423e-06,
+      "loss": 0.4287,
+      "step": 10963
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1633564708850073e-06,
+      "loss": 0.4313,
+      "step": 10964
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.161901422221886e-06,
+      "loss": 0.4188,
+      "step": 10965
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1604468037414293e-06,
+      "loss": 0.3819,
+      "step": 10966
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1589926155234786e-06,
+      "loss": 0.4328,
+      "step": 10967
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1575388576478416e-06,
+      "loss": 0.4149,
+      "step": 10968
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.156085530194306e-06,
+      "loss": 0.3875,
+      "step": 10969
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.154632633242635e-06,
+      "loss": 0.4039,
+      "step": 10970
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1531801668725695e-06,
+      "loss": 0.4293,
+      "step": 10971
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1517281311638214e-06,
+      "loss": 0.4029,
+      "step": 10972
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.150276526196092e-06,
+      "loss": 0.435,
+      "step": 10973
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1488253520490463e-06,
+      "loss": 0.407,
+      "step": 10974
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1473746088023263e-06,
+      "loss": 0.4267,
+      "step": 10975
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.145924296535561e-06,
+      "loss": 0.4237,
+      "step": 10976
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1444744153283447e-06,
+      "loss": 0.4316,
+      "step": 10977
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.143024965260253e-06,
+      "loss": 0.4143,
+      "step": 10978
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1415759464108353e-06,
+      "loss": 0.4057,
+      "step": 10979
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.140127358859618e-06,
+      "loss": 0.4074,
+      "step": 10980
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1386792026861103e-06,
+      "loss": 0.4628,
+      "step": 10981
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1372314779697877e-06,
+      "loss": 0.4025,
+      "step": 10982
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.135784184790105e-06,
+      "loss": 0.3808,
+      "step": 10983
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1343373232264996e-06,
+      "loss": 0.4337,
+      "step": 10984
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.132890893358378e-06,
+      "loss": 0.413,
+      "step": 10985
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1314448952651246e-06,
+      "loss": 0.4003,
+      "step": 10986
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.129999329026102e-06,
+      "loss": 0.4266,
+      "step": 10987
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.128554194720646e-06,
+      "loss": 0.4501,
+      "step": 10988
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1271094924280744e-06,
+      "loss": 0.4281,
+      "step": 10989
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1256652222276753e-06,
+      "loss": 0.4287,
+      "step": 10990
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1242213841987156e-06,
+      "loss": 0.4618,
+      "step": 10991
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.122777978420436e-06,
+      "loss": 0.4008,
+      "step": 10992
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1213350049720604e-06,
+      "loss": 0.3898,
+      "step": 10993
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.119892463932781e-06,
+      "loss": 0.4138,
+      "step": 10994
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1184503553817714e-06,
+      "loss": 0.4277,
+      "step": 10995
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1170086793981792e-06,
+      "loss": 0.4489,
+      "step": 10996
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1155674360611246e-06,
+      "loss": 0.374,
+      "step": 10997
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.114126625449715e-06,
+      "loss": 0.3812,
+      "step": 10998
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1126862476430245e-06,
+      "loss": 0.4257,
+      "step": 10999
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1112463027201015e-06,
+      "loss": 0.4147,
+      "step": 11000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.109806790759983e-06,
+      "loss": 0.4328,
+      "step": 11001
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1083677118416703e-06,
+      "loss": 0.402,
+      "step": 11002
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.106929066044147e-06,
+      "loss": 0.3969,
+      "step": 11003
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.105490853446368e-06,
+      "loss": 0.3929,
+      "step": 11004
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.104053074127268e-06,
+      "loss": 0.4008,
+      "step": 11005
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1026157281657613e-06,
+      "loss": 0.439,
+      "step": 11006
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.101178815640731e-06,
+      "loss": 0.3952,
+      "step": 11007
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.09974233663104e-06,
+      "loss": 0.431,
+      "step": 11008
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0983062912155296e-06,
+      "loss": 0.4245,
+      "step": 11009
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0968706794730144e-06,
+      "loss": 0.3932,
+      "step": 11010
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.095435501482286e-06,
+      "loss": 0.4331,
+      "step": 11011
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.09400075732211e-06,
+      "loss": 0.4356,
+      "step": 11012
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0925664470712304e-06,
+      "loss": 0.4148,
+      "step": 11013
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.091132570808371e-06,
+      "loss": 0.3995,
+      "step": 11014
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.089699128612224e-06,
+      "loss": 0.3939,
+      "step": 11015
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.088266120561464e-06,
+      "loss": 0.4096,
+      "step": 11016
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0868335467347367e-06,
+      "loss": 0.4399,
+      "step": 11017
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.085401407210671e-06,
+      "loss": 0.4386,
+      "step": 11018
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0839697020678663e-06,
+      "loss": 0.4042,
+      "step": 11019
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0825384313848995e-06,
+      "loss": 0.4205,
+      "step": 11020
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0811075952403193e-06,
+      "loss": 0.4281,
+      "step": 11021
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.079677193712664e-06,
+      "loss": 0.4288,
+      "step": 11022
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.078247226880432e-06,
+      "loss": 0.3831,
+      "step": 11023
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0768176948221088e-06,
+      "loss": 0.3732,
+      "step": 11024
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.075388597616148e-06,
+      "loss": 0.4172,
+      "step": 11025
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.073959935340988e-06,
+      "loss": 0.4197,
+      "step": 11026
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0725317080750373e-06,
+      "loss": 0.3638,
+      "step": 11027
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.071103915896683e-06,
+      "loss": 0.3975,
+      "step": 11028
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0696765588842814e-06,
+      "loss": 0.4183,
+      "step": 11029
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.06824963711618e-06,
+      "loss": 0.3805,
+      "step": 11030
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0668231506706893e-06,
+      "loss": 0.4,
+      "step": 11031
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0653970996260987e-06,
+      "loss": 0.3995,
+      "step": 11032
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.063971484060677e-06,
+      "loss": 0.4059,
+      "step": 11033
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0625463040526626e-06,
+      "loss": 0.4198,
+      "step": 11034
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.06112155968028e-06,
+      "loss": 0.3743,
+      "step": 11035
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.059697251021724e-06,
+      "loss": 0.3862,
+      "step": 11036
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.058273378155159e-06,
+      "loss": 0.3787,
+      "step": 11037
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0568499411587405e-06,
+      "loss": 0.389,
+      "step": 11038
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0554269401105888e-06,
+      "loss": 0.4156,
+      "step": 11039
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.054004375088803e-06,
+      "loss": 0.4105,
+      "step": 11040
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0525822461714575e-06,
+      "loss": 0.3937,
+      "step": 11041
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.051160553436601e-06,
+      "loss": 0.3784,
+      "step": 11042
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0497392969622687e-06,
+      "loss": 0.4172,
+      "step": 11043
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.04831847682646e-06,
+      "loss": 0.4226,
+      "step": 11044
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.04689809310715e-06,
+      "loss": 0.4469,
+      "step": 11045
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0454781458823038e-06,
+      "loss": 0.4081,
+      "step": 11046
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0440586352298475e-06,
+      "loss": 0.366,
+      "step": 11047
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0426395612276896e-06,
+      "loss": 0.4238,
+      "step": 11048
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.041220923953714e-06,
+      "loss": 0.4296,
+      "step": 11049
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0398027234857808e-06,
+      "loss": 0.4461,
+      "step": 11050
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0383849599017238e-06,
+      "loss": 0.4196,
+      "step": 11051
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0369676332793586e-06,
+      "loss": 0.3991,
+      "step": 11052
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0355507436964684e-06,
+      "loss": 0.378,
+      "step": 11053
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.034134291230824e-06,
+      "loss": 0.3864,
+      "step": 11054
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0327182759601592e-06,
+      "loss": 0.4329,
+      "step": 11055
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.031302697962194e-06,
+      "loss": 0.4193,
+      "step": 11056
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.029887557314617e-06,
+      "loss": 0.3996,
+      "step": 11057
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0284728540950982e-06,
+      "loss": 0.4029,
+      "step": 11058
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.027058588381279e-06,
+      "loss": 0.3907,
+      "step": 11059
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0256447602507835e-06,
+      "loss": 0.4243,
+      "step": 11060
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0242313697812043e-06,
+      "loss": 0.4438,
+      "step": 11061
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.022818417050113e-06,
+      "loss": 0.3853,
+      "step": 11062
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0214059021350605e-06,
+      "loss": 0.3657,
+      "step": 11063
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.01999382511357e-06,
+      "loss": 0.4221,
+      "step": 11064
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0185821860631395e-06,
+      "loss": 0.4244,
+      "step": 11065
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.017170985061245e-06,
+      "loss": 0.4187,
+      "step": 11066
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.015760222185337e-06,
+      "loss": 0.4203,
+      "step": 11067
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.014349897512847e-06,
+      "loss": 0.4063,
+      "step": 11068
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.012940011121176e-06,
+      "loss": 0.4073,
+      "step": 11069
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0115305630877024e-06,
+      "loss": 0.3843,
+      "step": 11070
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0101215534897856e-06,
+      "loss": 0.4098,
+      "step": 11071
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0087129824047545e-06,
+      "loss": 0.4198,
+      "step": 11072
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.007304849909918e-06,
+      "loss": 0.4133,
+      "step": 11073
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.005897156082557e-06,
+      "loss": 0.3913,
+      "step": 11074
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.004489900999932e-06,
+      "loss": 0.4494,
+      "step": 11075
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0030830847392755e-06,
+      "loss": 0.4124,
+      "step": 11076
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0016767073778044e-06,
+      "loss": 0.4142,
+      "step": 11077
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0002707689927e-06,
+      "loss": 0.4215,
+      "step": 11078
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.99886526966113e-06,
+      "loss": 0.4184,
+      "step": 11079
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.99746020946023e-06,
+      "loss": 0.3573,
+      "step": 11080
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9960555884671163e-06,
+      "loss": 0.3966,
+      "step": 11081
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9946514067588787e-06,
+      "loss": 0.4172,
+      "step": 11082
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.993247664412583e-06,
+      "loss": 0.3937,
+      "step": 11083
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.991844361505271e-06,
+      "loss": 0.3945,
+      "step": 11084
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.990441498113963e-06,
+      "loss": 0.3997,
+      "step": 11085
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9890390743156506e-06,
+      "loss": 0.3824,
+      "step": 11086
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.987637090187309e-06,
+      "loss": 0.3851,
+      "step": 11087
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9862355458058792e-06,
+      "loss": 0.3918,
+      "step": 11088
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9848344412482856e-06,
+      "loss": 0.4129,
+      "step": 11089
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9834337765914235e-06,
+      "loss": 0.4045,
+      "step": 11090
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.982033551912167e-06,
+      "loss": 0.4038,
+      "step": 11091
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9806337672873664e-06,
+      "loss": 0.422,
+      "step": 11092
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9792344227938433e-06,
+      "loss": 0.4223,
+      "step": 11093
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.977835518508402e-06,
+      "loss": 0.387,
+      "step": 11094
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9764370545078214e-06,
+      "loss": 0.3668,
+      "step": 11095
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9750390308688517e-06,
+      "loss": 0.3901,
+      "step": 11096
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.973641447668221e-06,
+      "loss": 0.4308,
+      "step": 11097
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9722443049826346e-06,
+      "loss": 0.438,
+      "step": 11098
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9708476028887713e-06,
+      "loss": 0.396,
+      "step": 11099
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.969451341463289e-06,
+      "loss": 0.3843,
+      "step": 11100
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9680555207828146e-06,
+      "loss": 0.4343,
+      "step": 11101
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9666601409239617e-06,
+      "loss": 0.4461,
+      "step": 11102
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.965265201963309e-06,
+      "loss": 0.3759,
+      "step": 11103
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.963870703977422e-06,
+      "loss": 0.3806,
+      "step": 11104
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.96247664704283e-06,
+      "loss": 0.3877,
+      "step": 11105
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9610830312360475e-06,
+      "loss": 0.3968,
+      "step": 11106
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9596898566335575e-06,
+      "loss": 0.4098,
+      "step": 11107
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9582971233118244e-06,
+      "loss": 0.3994,
+      "step": 11108
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.956904831347286e-06,
+      "loss": 0.3942,
+      "step": 11109
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.955512980816354e-06,
+      "loss": 0.3912,
+      "step": 11110
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9541215717954208e-06,
+      "loss": 0.4258,
+      "step": 11111
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.952730604360853e-06,
+      "loss": 0.4115,
+      "step": 11112
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.95134007858899e-06,
+      "loss": 0.3966,
+      "step": 11113
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9499499945561504e-06,
+      "loss": 0.42,
+      "step": 11114
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9485603523386243e-06,
+      "loss": 0.39,
+      "step": 11115
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9471711520126823e-06,
+      "loss": 0.455,
+      "step": 11116
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.945782393654567e-06,
+      "loss": 0.3888,
+      "step": 11117
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9443940773404966e-06,
+      "loss": 0.3984,
+      "step": 11118
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9430062031466722e-06,
+      "loss": 0.3888,
+      "step": 11119
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9416187711492597e-06,
+      "loss": 0.4036,
+      "step": 11120
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.940231781424412e-06,
+      "loss": 0.4191,
+      "step": 11121
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.938845234048249e-06,
+      "loss": 0.4357,
+      "step": 11122
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.937459129096869e-06,
+      "loss": 0.3941,
+      "step": 11123
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9360734666463464e-06,
+      "loss": 0.3517,
+      "step": 11124
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9346882467727323e-06,
+      "loss": 0.383,
+      "step": 11125
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9333034695520503e-06,
+      "loss": 0.3809,
+      "step": 11126
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.931919135060304e-06,
+      "loss": 0.4013,
+      "step": 11127
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9305352433734683e-06,
+      "loss": 0.4181,
+      "step": 11128
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.929151794567501e-06,
+      "loss": 0.4089,
+      "step": 11129
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.927768788718326e-06,
+      "loss": 0.4456,
+      "step": 11130
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9263862259018506e-06,
+      "loss": 0.4396,
+      "step": 11131
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9250041061939516e-06,
+      "loss": 0.4309,
+      "step": 11132
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9236224296704876e-06,
+      "loss": 0.4425,
+      "step": 11133
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9222411964072885e-06,
+      "loss": 0.4391,
+      "step": 11134
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9208604064801574e-06,
+      "loss": 0.3879,
+      "step": 11135
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.919480059964882e-06,
+      "loss": 0.43,
+      "step": 11136
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9181001569372217e-06,
+      "loss": 0.4028,
+      "step": 11137
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.916720697472908e-06,
+      "loss": 0.4069,
+      "step": 11138
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.915341681647651e-06,
+      "loss": 0.4113,
+      "step": 11139
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.913963109537136e-06,
+      "loss": 0.4018,
+      "step": 11140
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.912584981217024e-06,
+      "loss": 0.4496,
+      "step": 11141
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9112072967629503e-06,
+      "loss": 0.406,
+      "step": 11142
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.4085,
+      "step": 11143
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.908453259755344e-06,
+      "loss": 0.4041,
+      "step": 11144
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9070769073529615e-06,
+      "loss": 0.4016,
+      "step": 11145
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9057009991189246e-06,
+      "loss": 0.3951,
+      "step": 11146
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9043255351287438e-06,
+      "loss": 0.4192,
+      "step": 11147
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9029505154579098e-06,
+      "loss": 0.4291,
+      "step": 11148
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.901575940181889e-06,
+      "loss": 0.3783,
+      "step": 11149
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9002018093761222e-06,
+      "loss": 0.425,
+      "step": 11150
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8988281231160244e-06,
+      "loss": 0.4105,
+      "step": 11151
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8974548814769945e-06,
+      "loss": 0.389,
+      "step": 11152
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.896082084534394e-06,
+      "loss": 0.3722,
+      "step": 11153
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8947097323635743e-06,
+      "loss": 0.4582,
+      "step": 11154
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8933378250398493e-06,
+      "loss": 0.4066,
+      "step": 11155
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8919663626385165e-06,
+      "loss": 0.4463,
+      "step": 11156
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.890595345234847e-06,
+      "loss": 0.3843,
+      "step": 11157
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8892247729040848e-06,
+      "loss": 0.4132,
+      "step": 11158
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8878546457214509e-06,
+      "loss": 0.3656,
+      "step": 11159
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8864849637621463e-06,
+      "loss": 0.4134,
+      "step": 11160
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8851157271013443e-06,
+      "loss": 0.4046,
+      "step": 11161
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.883746935814188e-06,
+      "loss": 0.4212,
+      "step": 11162
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8823785899758085e-06,
+      "loss": 0.4121,
+      "step": 11163
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8810106896613013e-06,
+      "loss": 0.4022,
+      "step": 11164
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.879643234945744e-06,
+      "loss": 0.4207,
+      "step": 11165
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8782762259041843e-06,
+      "loss": 0.4037,
+      "step": 11166
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.876909662611649e-06,
+      "loss": 0.4138,
+      "step": 11167
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8755435451431426e-06,
+      "loss": 0.4294,
+      "step": 11168
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8741778735736415e-06,
+      "loss": 0.435,
+      "step": 11169
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8728126479780951e-06,
+      "loss": 0.3641,
+      "step": 11170
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8714478684314375e-06,
+      "loss": 0.4054,
+      "step": 11171
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.87008353500857e-06,
+      "loss": 0.3997,
+      "step": 11172
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8687196477843716e-06,
+      "loss": 0.417,
+      "step": 11173
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8673562068336982e-06,
+      "loss": 0.4214,
+      "step": 11174
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8659932122313774e-06,
+      "loss": 0.4138,
+      "step": 11175
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8646306640522193e-06,
+      "loss": 0.3698,
+      "step": 11176
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8632685623710046e-06,
+      "loss": 0.4263,
+      "step": 11177
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8619069072624884e-06,
+      "loss": 0.3848,
+      "step": 11178
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8605456988014014e-06,
+      "loss": 0.387,
+      "step": 11179
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8591849370624581e-06,
+      "loss": 0.4232,
+      "step": 11180
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8578246221203367e-06,
+      "loss": 0.3655,
+      "step": 11181
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.856464754049697e-06,
+      "loss": 0.4375,
+      "step": 11182
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8551053329251723e-06,
+      "loss": 0.4161,
+      "step": 11183
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8537463588213767e-06,
+      "loss": 0.4069,
+      "step": 11184
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.852387831812893e-06,
+      "loss": 0.3845,
+      "step": 11185
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8510297519742803e-06,
+      "loss": 0.404,
+      "step": 11186
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.849672119380075e-06,
+      "loss": 0.3948,
+      "step": 11187
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8483149341047923e-06,
+      "loss": 0.3985,
+      "step": 11188
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8469581962229167e-06,
+      "loss": 0.4447,
+      "step": 11189
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8456019058089124e-06,
+      "loss": 0.4255,
+      "step": 11190
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8442460629372128e-06,
+      "loss": 0.4199,
+      "step": 11191
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8428906676822377e-06,
+      "loss": 0.3867,
+      "step": 11192
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8415357201183725e-06,
+      "loss": 0.3937,
+      "step": 11193
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.840181220319983e-06,
+      "loss": 0.4011,
+      "step": 11194
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8388271683614056e-06,
+      "loss": 0.4056,
+      "step": 11195
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8374735643169606e-06,
+      "loss": 0.4074,
+      "step": 11196
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8361204082609353e-06,
+      "loss": 0.4081,
+      "step": 11197
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8347677002675967e-06,
+      "loss": 0.3625,
+      "step": 11198
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.833415440411187e-06,
+      "loss": 0.3602,
+      "step": 11199
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8320636287659177e-06,
+      "loss": 0.3741,
+      "step": 11200
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8307122654059895e-06,
+      "loss": 0.3753,
+      "step": 11201
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.829361350405564e-06,
+      "loss": 0.4025,
+      "step": 11202
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8280108838387878e-06,
+      "loss": 0.4188,
+      "step": 11203
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8266608657797735e-06,
+      "loss": 0.3703,
+      "step": 11204
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.825311296302622e-06,
+      "loss": 0.3946,
+      "step": 11205
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8239621754813997e-06,
+      "loss": 0.3872,
+      "step": 11206
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8226135033901505e-06,
+      "loss": 0.4044,
+      "step": 11207
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.821265280102892e-06,
+      "loss": 0.4253,
+      "step": 11208
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8199175056936247e-06,
+      "loss": 0.3779,
+      "step": 11209
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8185701802363166e-06,
+      "loss": 0.4268,
+      "step": 11210
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8172233038049137e-06,
+      "loss": 0.3813,
+      "step": 11211
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8158768764733347e-06,
+      "loss": 0.404,
+      "step": 11212
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8145308983154819e-06,
+      "loss": 0.4041,
+      "step": 11213
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8131853694052248e-06,
+      "loss": 0.3873,
+      "step": 11214
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.811840289816409e-06,
+      "loss": 0.3691,
+      "step": 11215
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8104956596228573e-06,
+      "loss": 0.4299,
+      "step": 11216
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8091514788983711e-06,
+      "loss": 0.4125,
+      "step": 11217
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8078077477167232e-06,
+      "loss": 0.3824,
+      "step": 11218
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8064644661516594e-06,
+      "loss": 0.4073,
+      "step": 11219
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8051216342769063e-06,
+      "loss": 0.4233,
+      "step": 11220
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.803779252166159e-06,
+      "loss": 0.4005,
+      "step": 11221
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8024373198930989e-06,
+      "loss": 0.3815,
+      "step": 11222
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8010958375313726e-06,
+      "loss": 0.3872,
+      "step": 11223
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7997548051546032e-06,
+      "loss": 0.4268,
+      "step": 11224
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7984142228363954e-06,
+      "loss": 0.3887,
+      "step": 11225
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7970740906503237e-06,
+      "loss": 0.4304,
+      "step": 11226
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.795734408669938e-06,
+      "loss": 0.4348,
+      "step": 11227
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7943951769687673e-06,
+      "loss": 0.3834,
+      "step": 11228
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7930563956203084e-06,
+      "loss": 0.4174,
+      "step": 11229
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7917180646980437e-06,
+      "loss": 0.4072,
+      "step": 11230
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7903801842754244e-06,
+      "loss": 0.4059,
+      "step": 11231
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7890427544258748e-06,
+      "loss": 0.4293,
+      "step": 11232
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7877057752228022e-06,
+      "loss": 0.4338,
+      "step": 11233
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7863692467395843e-06,
+      "loss": 0.3905,
+      "step": 11234
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7850331690495715e-06,
+      "loss": 0.4155,
+      "step": 11235
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7836975422260948e-06,
+      "loss": 0.4121,
+      "step": 11236
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7823623663424583e-06,
+      "loss": 0.3893,
+      "step": 11237
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7810276414719375e-06,
+      "loss": 0.3839,
+      "step": 11238
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7796933676877914e-06,
+      "loss": 0.3988,
+      "step": 11239
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7783595450632464e-06,
+      "loss": 0.4151,
+      "step": 11240
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.77702617367151e-06,
+      "loss": 0.4346,
+      "step": 11241
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7756932535857629e-06,
+      "loss": 0.374,
+      "step": 11242
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.774360784879159e-06,
+      "loss": 0.3868,
+      "step": 11243
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.773028767624827e-06,
+      "loss": 0.4401,
+      "step": 11244
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7716972018958767e-06,
+      "loss": 0.4209,
+      "step": 11245
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7703660877653827e-06,
+      "loss": 0.3816,
+      "step": 11246
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7690354253064101e-06,
+      "loss": 0.4006,
+      "step": 11247
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7677052145919815e-06,
+      "loss": 0.3882,
+      "step": 11248
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7663754556951118e-06,
+      "loss": 0.4026,
+      "step": 11249
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7650461486887781e-06,
+      "loss": 0.4094,
+      "step": 11250
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.763717293645939e-06,
+      "loss": 0.4179,
+      "step": 11251
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.762388890639526e-06,
+      "loss": 0.3769,
+      "step": 11252
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7610609397424473e-06,
+      "loss": 0.4165,
+      "step": 11253
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7597334410275823e-06,
+      "loss": 0.3716,
+      "step": 11254
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7584063945677932e-06,
+      "loss": 0.4475,
+      "step": 11255
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7570798004359101e-06,
+      "loss": 0.3621,
+      "step": 11256
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7557536587047441e-06,
+      "loss": 0.3798,
+      "step": 11257
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7544279694470778e-06,
+      "loss": 0.4228,
+      "step": 11258
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7531027327356686e-06,
+      "loss": 0.352,
+      "step": 11259
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7517779486432495e-06,
+      "loss": 0.3832,
+      "step": 11260
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7504536172425314e-06,
+      "loss": 0.3986,
+      "step": 11261
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7491297386061978e-06,
+      "loss": 0.3877,
+      "step": 11262
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7478063128069033e-06,
+      "loss": 0.4196,
+      "step": 11263
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7464833399172875e-06,
+      "loss": 0.4117,
+      "step": 11264
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.74516082000996e-06,
+      "loss": 0.4009,
+      "step": 11265
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7438387531575051e-06,
+      "loss": 0.4342,
+      "step": 11266
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7425171394324803e-06,
+      "loss": 0.397,
+      "step": 11267
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7411959789074228e-06,
+      "loss": 0.4195,
+      "step": 11268
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7398752716548395e-06,
+      "loss": 0.3729,
+      "step": 11269
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7385550177472187e-06,
+      "loss": 0.3942,
+      "step": 11270
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.737235217257016e-06,
+      "loss": 0.4191,
+      "step": 11271
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7359158702566736e-06,
+      "loss": 0.3771,
+      "step": 11272
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7345969768185944e-06,
+      "loss": 0.3991,
+      "step": 11273
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7332785370151716e-06,
+      "loss": 0.3901,
+      "step": 11274
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.731960550918761e-06,
+      "loss": 0.4465,
+      "step": 11275
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7306430186016998e-06,
+      "loss": 0.3707,
+      "step": 11276
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7293259401362984e-06,
+      "loss": 0.3747,
+      "step": 11277
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7280093155948429e-06,
+      "loss": 0.4056,
+      "step": 11278
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.726693145049595e-06,
+      "loss": 0.4385,
+      "step": 11279
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7253774285727875e-06,
+      "loss": 0.3971,
+      "step": 11280
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.724062166236633e-06,
+      "loss": 0.4041,
+      "step": 11281
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7227473581133225e-06,
+      "loss": 0.388,
+      "step": 11282
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7214330042750138e-06,
+      "loss": 0.4229,
+      "step": 11283
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7201191047938426e-06,
+      "loss": 0.4045,
+      "step": 11284
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.718805659741921e-06,
+      "loss": 0.4134,
+      "step": 11285
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7174926691913363e-06,
+      "loss": 0.3516,
+      "step": 11286
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7161801332141492e-06,
+      "loss": 0.4133,
+      "step": 11287
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7148680518823946e-06,
+      "loss": 0.4231,
+      "step": 11288
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.713556425268087e-06,
+      "loss": 0.4046,
+      "step": 11289
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.712245253443211e-06,
+      "loss": 0.379,
+      "step": 11290
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7109345364797314e-06,
+      "loss": 0.4537,
+      "step": 11291
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.709624274449584e-06,
+      "loss": 0.3921,
+      "step": 11292
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7083144674246787e-06,
+      "loss": 0.394,
+      "step": 11293
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.707005115476904e-06,
+      "loss": 0.3793,
+      "step": 11294
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.705696218678121e-06,
+      "loss": 0.4283,
+      "step": 11295
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.704387777100165e-06,
+      "loss": 0.3982,
+      "step": 11296
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.703079790814851e-06,
+      "loss": 0.3999,
+      "step": 11297
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7017722598939635e-06,
+      "loss": 0.3933,
+      "step": 11298
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7004651844092668e-06,
+      "loss": 0.4028,
+      "step": 11299
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6991585644324982e-06,
+      "loss": 0.4076,
+      "step": 11300
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6978524000353668e-06,
+      "loss": 0.4136,
+      "step": 11301
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6965466912895612e-06,
+      "loss": 0.4187,
+      "step": 11302
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6952414382667437e-06,
+      "loss": 0.4241,
+      "step": 11303
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6939366410385493e-06,
+      "loss": 0.3709,
+      "step": 11304
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6926322996765899e-06,
+      "loss": 0.4351,
+      "step": 11305
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6913284142524522e-06,
+      "loss": 0.4168,
+      "step": 11306
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6900249848377026e-06,
+      "loss": 0.3987,
+      "step": 11307
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6887220115038748e-06,
+      "loss": 0.4197,
+      "step": 11308
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6874194943224808e-06,
+      "loss": 0.4249,
+      "step": 11309
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6861174333650065e-06,
+      "loss": 0.382,
+      "step": 11310
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.684815828702915e-06,
+      "loss": 0.3971,
+      "step": 11311
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.683514680407642e-06,
+      "loss": 0.3988,
+      "step": 11312
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.682213988550596e-06,
+      "loss": 0.392,
+      "step": 11313
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6809137532031705e-06,
+      "loss": 0.4064,
+      "step": 11314
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6796139744367214e-06,
+      "loss": 0.3992,
+      "step": 11315
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.678314652322589e-06,
+      "loss": 0.3738,
+      "step": 11316
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6770157869320826e-06,
+      "loss": 0.423,
+      "step": 11317
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.67571737833649e-06,
+      "loss": 0.4337,
+      "step": 11318
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6744194266070702e-06,
+      "loss": 0.4036,
+      "step": 11319
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6731219318150616e-06,
+      "loss": 0.4021,
+      "step": 11320
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6718248940316707e-06,
+      "loss": 0.4383,
+      "step": 11321
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6705283133280902e-06,
+      "loss": 0.389,
+      "step": 11322
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.669232189775476e-06,
+      "loss": 0.4054,
+      "step": 11323
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6679365234449684e-06,
+      "loss": 0.3923,
+      "step": 11324
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6666413144076755e-06,
+      "loss": 0.3953,
+      "step": 11325
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6653465627346822e-06,
+      "loss": 0.3817,
+      "step": 11326
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6640522684970518e-06,
+      "loss": 0.4124,
+      "step": 11327
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6627584317658163e-06,
+      "loss": 0.4209,
+      "step": 11328
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6614650526119868e-06,
+      "loss": 0.3953,
+      "step": 11329
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.660172131106551e-06,
+      "loss": 0.3873,
+      "step": 11330
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6588796673204677e-06,
+      "loss": 0.3531,
+      "step": 11331
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6575876613246699e-06,
+      "loss": 0.4014,
+      "step": 11332
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6562961131900711e-06,
+      "loss": 0.3792,
+      "step": 11333
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6550050229875547e-06,
+      "loss": 0.3919,
+      "step": 11334
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6537143907879793e-06,
+      "loss": 0.4012,
+      "step": 11335
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6524242166621795e-06,
+      "loss": 0.4054,
+      "step": 11336
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6511345006809664e-06,
+      "loss": 0.3726,
+      "step": 11337
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.64984524291512e-06,
+      "loss": 0.418,
+      "step": 11338
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.648556443435404e-06,
+      "loss": 0.3764,
+      "step": 11339
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6472681023125492e-06,
+      "loss": 0.3854,
+      "step": 11340
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.645980219617267e-06,
+      "loss": 0.3796,
+      "step": 11341
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6446927954202396e-06,
+      "loss": 0.3767,
+      "step": 11342
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6434058297921262e-06,
+      "loss": 0.4047,
+      "step": 11343
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.642119322803558e-06,
+      "loss": 0.403,
+      "step": 11344
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6408332745251454e-06,
+      "loss": 0.4102,
+      "step": 11345
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6395476850274682e-06,
+      "loss": 0.3947,
+      "step": 11346
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6382625543810881e-06,
+      "loss": 0.4338,
+      "step": 11347
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6369778826565352e-06,
+      "loss": 0.3914,
+      "step": 11348
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6356936699243153e-06,
+      "loss": 0.3923,
+      "step": 11349
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6344099162549143e-06,
+      "loss": 0.3577,
+      "step": 11350
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.633126621718788e-06,
+      "loss": 0.4214,
+      "step": 11351
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.631843786386368e-06,
+      "loss": 0.4221,
+      "step": 11352
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6305614103280598e-06,
+      "loss": 0.3918,
+      "step": 11353
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6292794936142431e-06,
+      "loss": 0.3463,
+      "step": 11354
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6279980363152781e-06,
+      "loss": 0.4074,
+      "step": 11355
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6267170385014952e-06,
+      "loss": 0.4099,
+      "step": 11356
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.625436500243196e-06,
+      "loss": 0.4036,
+      "step": 11357
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6241564216106653e-06,
+      "loss": 0.3892,
+      "step": 11358
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.622876802674158e-06,
+      "loss": 0.4042,
+      "step": 11359
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6215976435039027e-06,
+      "loss": 0.4079,
+      "step": 11360
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.620318944170104e-06,
+      "loss": 0.4026,
+      "step": 11361
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6190407047429402e-06,
+      "loss": 0.4368,
+      "step": 11362
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6177629252925697e-06,
+      "loss": 0.367,
+      "step": 11363
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6164856058891187e-06,
+      "loss": 0.387,
+      "step": 11364
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.61520874660269e-06,
+      "loss": 0.4094,
+      "step": 11365
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6139323475033663e-06,
+      "loss": 0.4106,
+      "step": 11366
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6126564086611974e-06,
+      "loss": 0.4347,
+      "step": 11367
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6113809301462125e-06,
+      "loss": 0.3892,
+      "step": 11368
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6101059120284147e-06,
+      "loss": 0.3622,
+      "step": 11369
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.608831354377779e-06,
+      "loss": 0.3847,
+      "step": 11370
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6075572572642617e-06,
+      "loss": 0.3762,
+      "step": 11371
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.606283620757787e-06,
+      "loss": 0.4314,
+      "step": 11372
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6050104449282578e-06,
+      "loss": 0.3722,
+      "step": 11373
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.603737729845548e-06,
+      "loss": 0.3853,
+      "step": 11374
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6024654755795121e-06,
+      "loss": 0.4136,
+      "step": 11375
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6011936821999762e-06,
+      "loss": 0.3996,
+      "step": 11376
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.5999223497767379e-06,
+      "loss": 0.4127,
+      "step": 11377
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5986514783795726e-06,
+      "loss": 0.4526,
+      "step": 11378
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.597381068078232e-06,
+      "loss": 0.4105,
+      "step": 11379
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5961111189424416e-06,
+      "loss": 0.379,
+      "step": 11380
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.594841631041899e-06,
+      "loss": 0.4349,
+      "step": 11381
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5935726044462763e-06,
+      "loss": 0.4066,
+      "step": 11382
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5923040392252253e-06,
+      "loss": 0.4141,
+      "step": 11383
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5910359354483696e-06,
+      "loss": 0.3915,
+      "step": 11384
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5897682931853065e-06,
+      "loss": 0.4069,
+      "step": 11385
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.588501112505605e-06,
+      "loss": 0.4007,
+      "step": 11386
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5872343934788181e-06,
+      "loss": 0.3744,
+      "step": 11387
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.585968136174465e-06,
+      "loss": 0.39,
+      "step": 11388
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5847023406620431e-06,
+      "loss": 0.3921,
+      "step": 11389
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.583437007011024e-06,
+      "loss": 0.4165,
+      "step": 11390
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5821721352908503e-06,
+      "loss": 0.4102,
+      "step": 11391
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5809077255709482e-06,
+      "loss": 0.4041,
+      "step": 11392
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5796437779207108e-06,
+      "loss": 0.4647,
+      "step": 11393
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.578380292409505e-06,
+      "loss": 0.4274,
+      "step": 11394
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5771172691066793e-06,
+      "loss": 0.3881,
+      "step": 11395
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5758547080815522e-06,
+      "loss": 0.3925,
+      "step": 11396
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.574592609403417e-06,
+      "loss": 0.3547,
+      "step": 11397
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5733309731415414e-06,
+      "loss": 0.4097,
+      "step": 11398
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.572069799365167e-06,
+      "loss": 0.3713,
+      "step": 11399
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5708090881435168e-06,
+      "loss": 0.381,
+      "step": 11400
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5695488395457791e-06,
+      "loss": 0.4256,
+      "step": 11401
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.568289053641119e-06,
+      "loss": 0.4103,
+      "step": 11402
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5670297304986837e-06,
+      "loss": 0.3953,
+      "step": 11403
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.565770870187585e-06,
+      "loss": 0.3871,
+      "step": 11404
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5645124727769168e-06,
+      "loss": 0.399,
+      "step": 11405
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5632545383357412e-06,
+      "loss": 0.4031,
+      "step": 11406
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5619970669330997e-06,
+      "loss": 0.4237,
+      "step": 11407
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5607400586380038e-06,
+      "loss": 0.4138,
+      "step": 11408
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5594835135194475e-06,
+      "loss": 0.4041,
+      "step": 11409
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.558227431646393e-06,
+      "loss": 0.4085,
+      "step": 11410
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5569718130877742e-06,
+      "loss": 0.4164,
+      "step": 11411
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5557166579125105e-06,
+      "loss": 0.4184,
+      "step": 11412
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5544619661894866e-06,
+      "loss": 0.4099,
+      "step": 11413
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5532077379875642e-06,
+      "loss": 0.4539,
+      "step": 11414
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5519539733755784e-06,
+      "loss": 0.4335,
+      "step": 11415
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5507006724223405e-06,
+      "loss": 0.3921,
+      "step": 11416
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.549447835196639e-06,
+      "loss": 0.4001,
+      "step": 11417
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5481954617672323e-06,
+      "loss": 0.4341,
+      "step": 11418
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5469435522028519e-06,
+      "loss": 0.3767,
+      "step": 11419
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.545692106572213e-06,
+      "loss": 0.4223,
+      "step": 11420
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.544441124943996e-06,
+      "loss": 0.4072,
+      "step": 11421
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5431906073868607e-06,
+      "loss": 0.3961,
+      "step": 11422
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5419405539694377e-06,
+      "loss": 0.4207,
+      "step": 11423
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.540690964760334e-06,
+      "loss": 0.3982,
+      "step": 11424
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5394418398281353e-06,
+      "loss": 0.3501,
+      "step": 11425
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5381931792413962e-06,
+      "loss": 0.4231,
+      "step": 11426
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5369449830686445e-06,
+      "loss": 0.3857,
+      "step": 11427
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5356972513783908e-06,
+      "loss": 0.3969,
+      "step": 11428
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5344499842391125e-06,
+      "loss": 0.3791,
+      "step": 11429
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.533203181719264e-06,
+      "loss": 0.3895,
+      "step": 11430
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5319568438872746e-06,
+      "loss": 0.4352,
+      "step": 11431
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5307109708115476e-06,
+      "loss": 0.3939,
+      "step": 11432
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5294655625604581e-06,
+      "loss": 0.4192,
+      "step": 11433
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.528220619202364e-06,
+      "loss": 0.4171,
+      "step": 11434
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5269761408055872e-06,
+      "loss": 0.3904,
+      "step": 11435
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.525732127438433e-06,
+      "loss": 0.392,
+      "step": 11436
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5244885791691766e-06,
+      "loss": 0.4093,
+      "step": 11437
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.523245496066067e-06,
+      "loss": 0.3943,
+      "step": 11438
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5220028781973295e-06,
+      "loss": 0.3911,
+      "step": 11439
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5207607256311641e-06,
+      "loss": 0.4007,
+      "step": 11440
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5195190384357405e-06,
+      "loss": 0.3548,
+      "step": 11441
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5182778166792124e-06,
+      "loss": 0.4149,
+      "step": 11442
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.517037060429698e-06,
+      "loss": 0.4444,
+      "step": 11443
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5157967697553e-06,
+      "loss": 0.4051,
+      "step": 11444
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5145569447240848e-06,
+      "loss": 0.3944,
+      "step": 11445
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5133175854041016e-06,
+      "loss": 0.3785,
+      "step": 11446
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.512078691863369e-06,
+      "loss": 0.44,
+      "step": 11447
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5108402641698815e-06,
+      "loss": 0.446,
+      "step": 11448
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5096023023916096e-06,
+      "loss": 0.3931,
+      "step": 11449
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.508364806596495e-06,
+      "loss": 0.3978,
+      "step": 11450
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5071277768524572e-06,
+      "loss": 0.4141,
+      "step": 11451
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5058912132273907e-06,
+      "loss": 0.3932,
+      "step": 11452
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.504655115789161e-06,
+      "loss": 0.4189,
+      "step": 11453
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5034194846056093e-06,
+      "loss": 0.4533,
+      "step": 11454
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.502184319744553e-06,
+      "loss": 0.3862,
+      "step": 11455
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5009496212737795e-06,
+      "loss": 0.4059,
+      "step": 11456
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.499715389261055e-06,
+      "loss": 0.425,
+      "step": 11457
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4984816237741151e-06,
+      "loss": 0.3881,
+      "step": 11458
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4972483248806803e-06,
+      "loss": 0.3992,
+      "step": 11459
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4960154926484317e-06,
+      "loss": 0.4173,
+      "step": 11460
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4947831271450363e-06,
+      "loss": 0.4097,
+      "step": 11461
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4935512284381294e-06,
+      "loss": 0.4161,
+      "step": 11462
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4923197965953207e-06,
+      "loss": 0.37,
+      "step": 11463
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4910888316841954e-06,
+      "loss": 0.4133,
+      "step": 11464
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.489858333772315e-06,
+      "loss": 0.3961,
+      "step": 11465
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.488628302927213e-06,
+      "loss": 0.4376,
+      "step": 11466
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4873987392163946e-06,
+      "loss": 0.3952,
+      "step": 11467
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4861696427073458e-06,
+      "loss": 0.3492,
+      "step": 11468
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4849410134675257e-06,
+      "loss": 0.4055,
+      "step": 11469
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4837128515643628e-06,
+      "loss": 0.3997,
+      "step": 11470
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4824851570652644e-06,
+      "loss": 0.4515,
+      "step": 11471
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4812579300376095e-06,
+      "loss": 0.4496,
+      "step": 11472
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4800311705487547e-06,
+      "loss": 0.3971,
+      "step": 11473
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4788048786660259e-06,
+      "loss": 0.392,
+      "step": 11474
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.477579054456727e-06,
+      "loss": 0.3673,
+      "step": 11475
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4763536979881355e-06,
+      "loss": 0.4056,
+      "step": 11476
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4751288093275073e-06,
+      "loss": 0.438,
+      "step": 11477
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4739043885420645e-06,
+      "loss": 0.4058,
+      "step": 11478
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4726804356990099e-06,
+      "loss": 0.401,
+      "step": 11479
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4714569508655175e-06,
+      "loss": 0.4185,
+      "step": 11480
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4702339341087357e-06,
+      "loss": 0.3883,
+      "step": 11481
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4690113854957889e-06,
+      "loss": 0.3902,
+      "step": 11482
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4677893050937719e-06,
+      "loss": 0.3634,
+      "step": 11483
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4665676929697614e-06,
+      "loss": 0.3958,
+      "step": 11484
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4653465491908003e-06,
+      "loss": 0.4093,
+      "step": 11485
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4641258738239139e-06,
+      "loss": 0.3933,
+      "step": 11486
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.462905666936093e-06,
+      "loss": 0.4137,
+      "step": 11487
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4616859285943085e-06,
+      "loss": 0.3783,
+      "step": 11488
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4604666588655036e-06,
+      "loss": 0.3615,
+      "step": 11489
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.459247857816597e-06,
+      "loss": 0.4204,
+      "step": 11490
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4580295255144784e-06,
+      "loss": 0.4013,
+      "step": 11491
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4568116620260142e-06,
+      "loss": 0.3844,
+      "step": 11492
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4555942674180467e-06,
+      "loss": 0.4015,
+      "step": 11493
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4543773417573926e-06,
+      "loss": 0.4025,
+      "step": 11494
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4531608851108404e-06,
+      "loss": 0.3968,
+      "step": 11495
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4519448975451512e-06,
+      "loss": 0.3919,
+      "step": 11496
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4507293791270637e-06,
+      "loss": 0.3876,
+      "step": 11497
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4495143299232916e-06,
+      "loss": 0.4212,
+      "step": 11498
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.448299750000518e-06,
+      "loss": 0.3692,
+      "step": 11499
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.447085639425404e-06,
+      "loss": 0.3999,
+      "step": 11500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4458719982645863e-06,
+      "loss": 0.4213,
+      "step": 11501
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4446588265846717e-06,
+      "loss": 0.3958,
+      "step": 11502
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4434461244522458e-06,
+      "loss": 0.3698,
+      "step": 11503
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4422338919338652e-06,
+      "loss": 0.4157,
+      "step": 11504
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4410221290960613e-06,
+      "loss": 0.38,
+      "step": 11505
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4398108360053387e-06,
+      "loss": 0.4179,
+      "step": 11506
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4386000127281795e-06,
+      "loss": 0.4136,
+      "step": 11507
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4373896593310333e-06,
+      "loss": 0.3736,
+      "step": 11508
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4361797758803352e-06,
+      "loss": 0.416,
+      "step": 11509
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4349703624424827e-06,
+      "loss": 0.3733,
+      "step": 11510
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4337614190838567e-06,
+      "loss": 0.3946,
+      "step": 11511
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4325529458708066e-06,
+      "loss": 0.3857,
+      "step": 11512
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.431344942869658e-06,
+      "loss": 0.348,
+      "step": 11513
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4301374101467091e-06,
+      "loss": 0.3913,
+      "step": 11514
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4289303477682348e-06,
+      "loss": 0.3931,
+      "step": 11515
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4277237558004809e-06,
+      "loss": 0.3951,
+      "step": 11516
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4265176343096732e-06,
+      "loss": 0.4009,
+      "step": 11517
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4253119833620055e-06,
+      "loss": 0.4301,
+      "step": 11518
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.424106803023646e-06,
+      "loss": 0.3831,
+      "step": 11519
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4229020933607452e-06,
+      "loss": 0.4191,
+      "step": 11520
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4216978544394177e-06,
+      "loss": 0.4096,
+      "step": 11521
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4204940863257565e-06,
+      "loss": 0.41,
+      "step": 11522
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.419290789085831e-06,
+      "loss": 0.4053,
+      "step": 11523
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.418087962785678e-06,
+      "loss": 0.4174,
+      "step": 11524
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4168856074913174e-06,
+      "loss": 0.4004,
+      "step": 11525
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.415683723268737e-06,
+      "loss": 0.4175,
+      "step": 11526
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4144823101838989e-06,
+      "loss": 0.4068,
+      "step": 11527
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4132813683027436e-06,
+      "loss": 0.3852,
+      "step": 11528
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4120808976911825e-06,
+      "loss": 0.3924,
+      "step": 11529
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4108808984151023e-06,
+      "loss": 0.3815,
+      "step": 11530
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.409681370540361e-06,
+      "loss": 0.4152,
+      "step": 11531
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4084823141327918e-06,
+      "loss": 0.3725,
+      "step": 11532
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4072837292582076e-06,
+      "loss": 0.4266,
+      "step": 11533
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4060856159823888e-06,
+      "loss": 0.3969,
+      "step": 11534
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4048879743710897e-06,
+      "loss": 0.4088,
+      "step": 11535
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4036908044900455e-06,
+      "loss": 0.3999,
+      "step": 11536
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4024941064049592e-06,
+      "loss": 0.3935,
+      "step": 11537
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4012978801815102e-06,
+      "loss": 0.402,
+      "step": 11538
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.400102125885351e-06,
+      "loss": 0.4264,
+      "step": 11539
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3989068435821062e-06,
+      "loss": 0.377,
+      "step": 11540
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3977120333373829e-06,
+      "loss": 0.4232,
+      "step": 11541
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3965176952167536e-06,
+      "loss": 0.3889,
+      "step": 11542
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3953238292857675e-06,
+      "loss": 0.4147,
+      "step": 11543
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3941304356099462e-06,
+      "loss": 0.3914,
+      "step": 11544
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.392937514254792e-06,
+      "loss": 0.395,
+      "step": 11545
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3917450652857746e-06,
+      "loss": 0.4388,
+      "step": 11546
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3905530887683393e-06,
+      "loss": 0.3926,
+      "step": 11547
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3893615847679065e-06,
+      "loss": 0.4546,
+      "step": 11548
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3881705533498669e-06,
+      "loss": 0.4394,
+      "step": 11549
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3869799945795947e-06,
+      "loss": 0.4018,
+      "step": 11550
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3857899085224291e-06,
+      "loss": 0.3946,
+      "step": 11551
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3846002952436833e-06,
+      "loss": 0.4116,
+      "step": 11552
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3834111548086526e-06,
+      "loss": 0.4088,
+      "step": 11553
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3822224872825984e-06,
+      "loss": 0.3758,
+      "step": 11554
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3810342927307595e-06,
+      "loss": 0.3955,
+      "step": 11555
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3798465712183496e-06,
+      "loss": 0.4205,
+      "step": 11556
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3786593228105494e-06,
+      "loss": 0.3845,
+      "step": 11557
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.377472547572527e-06,
+      "loss": 0.4291,
+      "step": 11558
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3762862455694137e-06,
+      "loss": 0.3615,
+      "step": 11559
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3751004168663163e-06,
+      "loss": 0.4392,
+      "step": 11560
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3739150615283182e-06,
+      "loss": 0.3934,
+      "step": 11561
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3727301796204774e-06,
+      "loss": 0.4127,
+      "step": 11562
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3715457712078227e-06,
+      "loss": 0.3602,
+      "step": 11563
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3703618363553605e-06,
+      "loss": 0.3836,
+      "step": 11564
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3691783751280663e-06,
+      "loss": 0.3822,
+      "step": 11565
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3679953875908959e-06,
+      "loss": 0.362,
+      "step": 11566
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.366812873808775e-06,
+      "loss": 0.4136,
+      "step": 11567
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.365630833846603e-06,
+      "loss": 0.4079,
+      "step": 11568
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3644492677692534e-06,
+      "loss": 0.4084,
+      "step": 11569
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3632681756415777e-06,
+      "loss": 0.3813,
+      "step": 11570
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3620875575283988e-06,
+      "loss": 0.4061,
+      "step": 11571
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.36090741349451e-06,
+      "loss": 0.3921,
+      "step": 11572
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3597277436046818e-06,
+      "loss": 0.4074,
+      "step": 11573
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3585485479236626e-06,
+      "loss": 0.4184,
+      "step": 11574
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3573698265161683e-06,
+      "loss": 0.4232,
+      "step": 11575
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3561915794468917e-06,
+      "loss": 0.3584,
+      "step": 11576
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3550138067804986e-06,
+      "loss": 0.405,
+      "step": 11577
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3538365085816274e-06,
+      "loss": 0.415,
+      "step": 11578
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3526596849148977e-06,
+      "loss": 0.4152,
+      "step": 11579
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3514833358448943e-06,
+      "loss": 0.3932,
+      "step": 11580
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3503074614361778e-06,
+      "loss": 0.4087,
+      "step": 11581
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3491320617532877e-06,
+      "loss": 0.415,
+      "step": 11582
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3479571368607337e-06,
+      "loss": 0.4129,
+      "step": 11583
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3467826868229995e-06,
+      "loss": 0.353,
+      "step": 11584
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3456087117045413e-06,
+      "loss": 0.4183,
+      "step": 11585
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3444352115697912e-06,
+      "loss": 0.4282,
+      "step": 11586
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3432621864831575e-06,
+      "loss": 0.3989,
+      "step": 11587
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3420896365090197e-06,
+      "loss": 0.4065,
+      "step": 11588
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3409175617117277e-06,
+      "loss": 0.398,
+      "step": 11589
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.4265,
+      "step": 11590
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.338574837904978e-06,
+      "loss": 0.4141,
+      "step": 11591
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.337404189024094e-06,
+      "loss": 0.4077,
+      "step": 11592
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3362340155772147e-06,
+      "loss": 0.3906,
+      "step": 11593
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.335064317628557e-06,
+      "loss": 0.3858,
+      "step": 11594
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3338950952423257e-06,
+      "loss": 0.4312,
+      "step": 11595
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.332726348482688e-06,
+      "loss": 0.395,
+      "step": 11596
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3315580774137883e-06,
+      "loss": 0.3972,
+      "step": 11597
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3303902820997471e-06,
+      "loss": 0.3749,
+      "step": 11598
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3292229626046572e-06,
+      "loss": 0.3714,
+      "step": 11599
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3280561189925844e-06,
+      "loss": 0.4385,
+      "step": 11600
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3268897513275692e-06,
+      "loss": 0.4039,
+      "step": 11601
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3257238596736266e-06,
+      "loss": 0.4138,
+      "step": 11602
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3245584440947423e-06,
+      "loss": 0.421,
+      "step": 11603
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.323393504654883e-06,
+      "loss": 0.3928,
+      "step": 11604
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3222290414179794e-06,
+      "loss": 0.3733,
+      "step": 11605
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3210650544479464e-06,
+      "loss": 0.3855,
+      "step": 11606
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.319901543808666e-06,
+      "loss": 0.4217,
+      "step": 11607
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.318738509563996e-06,
+      "loss": 0.4203,
+      "step": 11608
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3175759517777664e-06,
+      "loss": 0.3726,
+      "step": 11609
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.316413870513784e-06,
+      "loss": 0.375,
+      "step": 11610
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3152522658358247e-06,
+      "loss": 0.4011,
+      "step": 11611
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3140911378076459e-06,
+      "loss": 0.4062,
+      "step": 11612
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3129304864929715e-06,
+      "loss": 0.3883,
+      "step": 11613
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3117703119555047e-06,
+      "loss": 0.3999,
+      "step": 11614
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.310610614258918e-06,
+      "loss": 0.3913,
+      "step": 11615
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3094513934668618e-06,
+      "loss": 0.3896,
+      "step": 11616
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.308292649642956e-06,
+      "loss": 0.4354,
+      "step": 11617
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3071343828507977e-06,
+      "loss": 0.387,
+      "step": 11618
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3059765931539558e-06,
+      "loss": 0.3845,
+      "step": 11619
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3048192806159721e-06,
+      "loss": 0.4238,
+      "step": 11620
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3036624453003688e-06,
+      "loss": 0.4458,
+      "step": 11621
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.302506087270633e-06,
+      "loss": 0.4142,
+      "step": 11622
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3013502065902328e-06,
+      "loss": 0.3997,
+      "step": 11623
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3001948033226054e-06,
+      "loss": 0.3883,
+      "step": 11624
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.299039877531164e-06,
+      "loss": 0.3792,
+      "step": 11625
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2978854292792952e-06,
+      "loss": 0.4335,
+      "step": 11626
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2967314586303592e-06,
+      "loss": 0.4314,
+      "step": 11627
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2955779656476864e-06,
+      "loss": 0.4176,
+      "step": 11628
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2944249503945894e-06,
+      "loss": 0.3954,
+      "step": 11629
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2932724129343466e-06,
+      "loss": 0.3912,
+      "step": 11630
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2921203533302174e-06,
+      "loss": 0.3953,
+      "step": 11631
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2909687716454278e-06,
+      "loss": 0.4104,
+      "step": 11632
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2898176679431817e-06,
+      "loss": 0.391,
+      "step": 11633
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2886670422866543e-06,
+      "loss": 0.3937,
+      "step": 11634
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2875168947389982e-06,
+      "loss": 0.3909,
+      "step": 11635
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2863672253633353e-06,
+      "loss": 0.4035,
+      "step": 11636
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2852180342227637e-06,
+      "loss": 0.4034,
+      "step": 11637
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2840693213803545e-06,
+      "loss": 0.3744,
+      "step": 11638
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2829210868991582e-06,
+      "loss": 0.4211,
+      "step": 11639
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2817733308421898e-06,
+      "loss": 0.4374,
+      "step": 11640
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2806260532724435e-06,
+      "loss": 0.3836,
+      "step": 11641
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2794792542528843e-06,
+      "loss": 0.45,
+      "step": 11642
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2783329338464545e-06,
+      "loss": 0.4076,
+      "step": 11643
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2771870921160668e-06,
+      "loss": 0.3756,
+      "step": 11644
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2760417291246074e-06,
+      "loss": 0.4413,
+      "step": 11645
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.274896844934943e-06,
+      "loss": 0.3902,
+      "step": 11646
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2737524396099033e-06,
+      "loss": 0.44,
+      "step": 11647
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2726085132123022e-06,
+      "loss": 0.3801,
+      "step": 11648
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.27146506580492e-06,
+      "loss": 0.4025,
+      "step": 11649
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2703220974505137e-06,
+      "loss": 0.3891,
+      "step": 11650
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2691796082118135e-06,
+      "loss": 0.3932,
+      "step": 11651
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2680375981515225e-06,
+      "loss": 0.3953,
+      "step": 11652
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2668960673323172e-06,
+      "loss": 0.3739,
+      "step": 11653
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2657550158168519e-06,
+      "loss": 0.4029,
+      "step": 11654
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2646144436677477e-06,
+      "loss": 0.4248,
+      "step": 11655
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2634743509476088e-06,
+      "loss": 0.3777,
+      "step": 11656
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2623347377190032e-06,
+      "loss": 0.4371,
+      "step": 11657
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2611956040444784e-06,
+      "loss": 0.3933,
+      "step": 11658
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2600569499865545e-06,
+      "loss": 0.427,
+      "step": 11659
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2589187756077236e-06,
+      "loss": 0.4351,
+      "step": 11660
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2577810809704527e-06,
+      "loss": 0.385,
+      "step": 11661
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2566438661371815e-06,
+      "loss": 0.4325,
+      "step": 11662
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.255507131170326e-06,
+      "loss": 0.4294,
+      "step": 11663
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2543708761322759e-06,
+      "loss": 0.3808,
+      "step": 11664
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2532351010853916e-06,
+      "loss": 0.3975,
+      "step": 11665
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2520998060920087e-06,
+      "loss": 0.426,
+      "step": 11666
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.250964991214435e-06,
+      "loss": 0.4332,
+      "step": 11667
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2498306565149532e-06,
+      "loss": 0.3945,
+      "step": 11668
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.248696802055821e-06,
+      "loss": 0.4362,
+      "step": 11669
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2475634278992654e-06,
+      "loss": 0.4029,
+      "step": 11670
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2464305341074934e-06,
+      "loss": 0.4199,
+      "step": 11671
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2452981207426795e-06,
+      "loss": 0.4624,
+      "step": 11672
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2441661878669786e-06,
+      "loss": 0.4077,
+      "step": 11673
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.243034735542512e-06,
+      "loss": 0.3856,
+      "step": 11674
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.241903763831379e-06,
+      "loss": 0.3964,
+      "step": 11675
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.24077327279565e-06,
+      "loss": 0.3924,
+      "step": 11676
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.239643262497372e-06,
+      "loss": 0.3891,
+      "step": 11677
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2385137329985607e-06,
+      "loss": 0.3937,
+      "step": 11678
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2373846843612135e-06,
+      "loss": 0.3742,
+      "step": 11679
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.236256116647292e-06,
+      "loss": 0.3794,
+      "step": 11680
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2351280299187395e-06,
+      "loss": 0.3904,
+      "step": 11681
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2340004242374693e-06,
+      "loss": 0.4027,
+      "step": 11682
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.232873299665367e-06,
+      "loss": 0.41,
+      "step": 11683
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2317466562642932e-06,
+      "loss": 0.386,
+      "step": 11684
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.230620494096082e-06,
+      "loss": 0.3881,
+      "step": 11685
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2294948132225416e-06,
+      "loss": 0.4046,
+      "step": 11686
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2283696137054513e-06,
+      "loss": 0.4049,
+      "step": 11687
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2272448956065709e-06,
+      "loss": 0.3964,
+      "step": 11688
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2261206589876228e-06,
+      "loss": 0.4161,
+      "step": 11689
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2249969039103148e-06,
+      "loss": 0.4081,
+      "step": 11690
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2238736304363197e-06,
+      "loss": 0.4044,
+      "step": 11691
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.222750838627288e-06,
+      "loss": 0.422,
+      "step": 11692
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2216285285448404e-06,
+      "loss": 0.4407,
+      "step": 11693
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2205067002505756e-06,
+      "loss": 0.3966,
+      "step": 11694
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2193853538060597e-06,
+      "loss": 0.4081,
+      "step": 11695
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2182644892728413e-06,
+      "loss": 0.4272,
+      "step": 11696
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2171441067124323e-06,
+      "loss": 0.4162,
+      "step": 11697
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2160242061863282e-06,
+      "loss": 0.421,
+      "step": 11698
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2149047877559916e-06,
+      "loss": 0.3873,
+      "step": 11699
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2137858514828583e-06,
+      "loss": 0.3695,
+      "step": 11700
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.212667397428342e-06,
+      "loss": 0.3612,
+      "step": 11701
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2115494256538252e-06,
+      "loss": 0.3897,
+      "step": 11702
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2104319362206651e-06,
+      "loss": 0.3763,
+      "step": 11703
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2093149291901973e-06,
+      "loss": 0.4101,
+      "step": 11704
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2081984046237248e-06,
+      "loss": 0.4055,
+      "step": 11705
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2070823625825255e-06,
+      "loss": 0.3891,
+      "step": 11706
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.205966803127855e-06,
+      "loss": 0.439,
+      "step": 11707
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2048517263209369e-06,
+      "loss": 0.3994,
+      "step": 11708
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.203737132222972e-06,
+      "loss": 0.4011,
+      "step": 11709
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2026230208951307e-06,
+      "loss": 0.367,
+      "step": 11710
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.20150939239856e-06,
+      "loss": 0.388,
+      "step": 11711
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.200396246794382e-06,
+      "loss": 0.4077,
+      "step": 11712
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1992835841436878e-06,
+      "loss": 0.4075,
+      "step": 11713
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1981714045075443e-06,
+      "loss": 0.4578,
+      "step": 11714
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1970597079469947e-06,
+      "loss": 0.3616,
+      "step": 11715
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1959484945230505e-06,
+      "loss": 0.4256,
+      "step": 11716
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.194837764296699e-06,
+      "loss": 0.3797,
+      "step": 11717
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.193727517328902e-06,
+      "loss": 0.397,
+      "step": 11718
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1926177536805905e-06,
+      "loss": 0.4041,
+      "step": 11719
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1915084734126781e-06,
+      "loss": 0.3899,
+      "step": 11720
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1903996765860415e-06,
+      "loss": 0.4252,
+      "step": 11721
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1892913632615355e-06,
+      "loss": 0.412,
+      "step": 11722
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1881835334999915e-06,
+      "loss": 0.4288,
+      "step": 11723
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1870761873622094e-06,
+      "loss": 0.3625,
+      "step": 11724
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1859693249089642e-06,
+      "loss": 0.3615,
+      "step": 11725
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1848629462010042e-06,
+      "loss": 0.3939,
+      "step": 11726
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1837570512990494e-06,
+      "loss": 0.4036,
+      "step": 11727
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.182651640263799e-06,
+      "loss": 0.3805,
+      "step": 11728
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1815467131559212e-06,
+      "loss": 0.4026,
+      "step": 11729
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1804422700360563e-06,
+      "loss": 0.3995,
+      "step": 11730
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1793383109648194e-06,
+      "loss": 0.4147,
+      "step": 11731
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.178234836002804e-06,
+      "loss": 0.4468,
+      "step": 11732
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1771318452105696e-06,
+      "loss": 0.3832,
+      "step": 11733
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1760293386486531e-06,
+      "loss": 0.3836,
+      "step": 11734
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1749273163775609e-06,
+      "loss": 0.3993,
+      "step": 11735
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.173825778457781e-06,
+      "loss": 0.3977,
+      "step": 11736
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1727247249497686e-06,
+      "loss": 0.3869,
+      "step": 11737
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1716241559139507e-06,
+      "loss": 0.4117,
+      "step": 11738
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1705240714107301e-06,
+      "loss": 0.4128,
+      "step": 11739
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1694244715004878e-06,
+      "loss": 0.3999,
+      "step": 11740
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1683253562435704e-06,
+      "loss": 0.4529,
+      "step": 11741
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1672267257003024e-06,
+      "loss": 0.445,
+      "step": 11742
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1661285799309775e-06,
+      "loss": 0.4205,
+      "step": 11743
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1650309189958708e-06,
+      "loss": 0.3874,
+      "step": 11744
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1639337429552245e-06,
+      "loss": 0.3898,
+      "step": 11745
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1628370518692533e-06,
+      "loss": 0.3886,
+      "step": 11746
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1617408457981482e-06,
+      "loss": 0.4077,
+      "step": 11747
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1606451248020722e-06,
+      "loss": 0.3575,
+      "step": 11748
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.159549888941165e-06,
+      "loss": 0.3904,
+      "step": 11749
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1584551382755361e-06,
+      "loss": 0.3974,
+      "step": 11750
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1573608728652663e-06,
+      "loss": 0.3839,
+      "step": 11751
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1562670927704179e-06,
+      "loss": 0.4575,
+      "step": 11752
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.155173798051018e-06,
+      "loss": 0.3885,
+      "step": 11753
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1540809887670713e-06,
+      "loss": 0.4407,
+      "step": 11754
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.152988664978556e-06,
+      "loss": 0.363,
+      "step": 11755
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.151896826745419e-06,
+      "loss": 0.4101,
+      "step": 11756
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1508054741275888e-06,
+      "loss": 0.3973,
+      "step": 11757
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1497146071849608e-06,
+      "loss": 0.3867,
+      "step": 11758
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1486242259774049e-06,
+      "loss": 0.3921,
+      "step": 11759
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1475343305647656e-06,
+      "loss": 0.4054,
+      "step": 11760
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1464449210068618e-06,
+      "loss": 0.4026,
+      "step": 11761
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.145355997363482e-06,
+      "loss": 0.4129,
+      "step": 11762
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.144267559694392e-06,
+      "loss": 0.3829,
+      "step": 11763
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1431796080593284e-06,
+      "loss": 0.4333,
+      "step": 11764
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1420921425179987e-06,
+      "loss": 0.3806,
+      "step": 11765
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1410051631300934e-06,
+      "loss": 0.4072,
+      "step": 11766
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1399186699552644e-06,
+      "loss": 0.4282,
+      "step": 11767
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1388326630531433e-06,
+      "loss": 0.4288,
+      "step": 11768
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1377471424833363e-06,
+      "loss": 0.3866,
+      "step": 11769
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1366621083054196e-06,
+      "loss": 0.4102,
+      "step": 11770
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.135577560578943e-06,
+      "loss": 0.4461,
+      "step": 11771
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1344934993634304e-06,
+      "loss": 0.4011,
+      "step": 11772
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1334099247183782e-06,
+      "loss": 0.3811,
+      "step": 11773
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.132326836703258e-06,
+      "loss": 0.3863,
+      "step": 11774
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1312442353775145e-06,
+      "loss": 0.3992,
+      "step": 11775
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.130162120800561e-06,
+      "loss": 0.4261,
+      "step": 11776
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1290804930317923e-06,
+      "loss": 0.385,
+      "step": 11777
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.12799935213057e-06,
+      "loss": 0.4027,
+      "step": 11778
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1269186981562308e-06,
+      "loss": 0.4327,
+      "step": 11779
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1258385311680852e-06,
+      "loss": 0.4132,
+      "step": 11780
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1247588512254138e-06,
+      "loss": 0.3641,
+      "step": 11781
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1236796583874787e-06,
+      "loss": 0.3956,
+      "step": 11782
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1226009527135062e-06,
+      "loss": 0.3915,
+      "step": 11783
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1215227342626978e-06,
+      "loss": 0.4388,
+      "step": 11784
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.120445003094235e-06,
+      "loss": 0.3967,
+      "step": 11785
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1193677592672648e-06,
+      "loss": 0.4252,
+      "step": 11786
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1182910028409099e-06,
+      "loss": 0.4189,
+      "step": 11787
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1172147338742668e-06,
+      "loss": 0.4302,
+      "step": 11788
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1161389524264055e-06,
+      "loss": 0.4246,
+      "step": 11789
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1150636585563658e-06,
+      "loss": 0.4444,
+      "step": 11790
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.113988852323168e-06,
+      "loss": 0.4124,
+      "step": 11791
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1129145337857982e-06,
+      "loss": 0.3942,
+      "step": 11792
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1118407030032207e-06,
+      "loss": 0.4165,
+      "step": 11793
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1107673600343715e-06,
+      "loss": 0.4143,
+      "step": 11794
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1096945049381569e-06,
+      "loss": 0.3621,
+      "step": 11795
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1086221377734618e-06,
+      "loss": 0.4326,
+      "step": 11796
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1075502585991394e-06,
+      "loss": 0.4052,
+      "step": 11797
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1064788674740179e-06,
+      "loss": 0.4009,
+      "step": 11798
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1054079644569016e-06,
+      "loss": 0.4435,
+      "step": 11799
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.104337549606561e-06,
+      "loss": 0.3828,
+      "step": 11800
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1032676229817507e-06,
+      "loss": 0.4027,
+      "step": 11801
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1021981846411889e-06,
+      "loss": 0.4389,
+      "step": 11802
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1011292346435687e-06,
+      "loss": 0.4108,
+      "step": 11803
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1000607730475598e-06,
+      "loss": 0.3765,
+      "step": 11804
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0989927999118022e-06,
+      "loss": 0.3607,
+      "step": 11805
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.09792531529491e-06,
+      "loss": 0.3641,
+      "step": 11806
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0968583192554693e-06,
+      "loss": 0.4262,
+      "step": 11807
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0957918118520417e-06,
+      "loss": 0.4073,
+      "step": 11808
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0947257931431644e-06,
+      "loss": 0.4106,
+      "step": 11809
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0936602631873406e-06,
+      "loss": 0.4134,
+      "step": 11810
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0925952220430513e-06,
+      "loss": 0.3769,
+      "step": 11811
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.091530669768749e-06,
+      "loss": 0.4089,
+      "step": 11812
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0904666064228608e-06,
+      "loss": 0.4221,
+      "step": 11813
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0894030320637861e-06,
+      "loss": 0.3674,
+      "step": 11814
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0883399467498957e-06,
+      "loss": 0.4191,
+      "step": 11815
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0872773505395396e-06,
+      "loss": 0.4004,
+      "step": 11816
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0862152434910322e-06,
+      "loss": 0.4056,
+      "step": 11817
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0851536256626705e-06,
+      "loss": 0.4256,
+      "step": 11818
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0840924971127176e-06,
+      "loss": 0.408,
+      "step": 11819
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0830318578994114e-06,
+      "loss": 0.4094,
+      "step": 11820
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0819717080809643e-06,
+      "loss": 0.3777,
+      "step": 11821
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.080912047715561e-06,
+      "loss": 0.3965,
+      "step": 11822
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0798528768613592e-06,
+      "loss": 0.3881,
+      "step": 11823
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.078794195576488e-06,
+      "loss": 0.3722,
+      "step": 11824
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0777360039190531e-06,
+      "loss": 0.4443,
+      "step": 11825
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.076678301947135e-06,
+      "loss": 0.3772,
+      "step": 11826
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0756210897187812e-06,
+      "loss": 0.3988,
+      "step": 11827
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.074564367292016e-06,
+      "loss": 0.403,
+      "step": 11828
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0735081347248343e-06,
+      "loss": 0.3786,
+      "step": 11829
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0724523920752083e-06,
+      "loss": 0.381,
+      "step": 11830
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0713971394010791e-06,
+      "loss": 0.4345,
+      "step": 11831
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0703423767603615e-06,
+      "loss": 0.3743,
+      "step": 11832
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0692881042109471e-06,
+      "loss": 0.4023,
+      "step": 11833
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0682343218106993e-06,
+      "loss": 0.3754,
+      "step": 11834
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0671810296174513e-06,
+      "loss": 0.3498,
+      "step": 11835
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0661282276890129e-06,
+      "loss": 0.4018,
+      "step": 11836
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0650759160831637e-06,
+      "loss": 0.4118,
+      "step": 11837
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0640240948576586e-06,
+      "loss": 0.3881,
+      "step": 11838
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.062972764070227e-06,
+      "loss": 0.4031,
+      "step": 11839
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.061921923778566e-06,
+      "loss": 0.4036,
+      "step": 11840
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0608715740403542e-06,
+      "loss": 0.4317,
+      "step": 11841
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0598217149132339e-06,
+      "loss": 0.405,
+      "step": 11842
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0587723464548294e-06,
+      "loss": 0.3812,
+      "step": 11843
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0577234687227322e-06,
+      "loss": 0.3662,
+      "step": 11844
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0566750817745076e-06,
+      "loss": 0.4127,
+      "step": 11845
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0556271856676959e-06,
+      "loss": 0.3889,
+      "step": 11846
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.054579780459808e-06,
+      "loss": 0.416,
+      "step": 11847
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.053532866208331e-06,
+      "loss": 0.4545,
+      "step": 11848
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0524864429707193e-06,
+      "loss": 0.4246,
+      "step": 11849
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.051440510804408e-06,
+      "loss": 0.4276,
+      "step": 11850
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.050395069766802e-06,
+      "loss": 0.4022,
+      "step": 11851
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0493501199152778e-06,
+      "loss": 0.4132,
+      "step": 11852
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0483056613071863e-06,
+      "loss": 0.4108,
+      "step": 11853
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0472616939998493e-06,
+      "loss": 0.4026,
+      "step": 11854
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0462182180505664e-06,
+      "loss": 0.4471,
+      "step": 11855
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0451752335166043e-06,
+      "loss": 0.4195,
+      "step": 11856
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.044132740455206e-06,
+      "loss": 0.3622,
+      "step": 11857
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.043090738923589e-06,
+      "loss": 0.4151,
+      "step": 11858
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0420492289789397e-06,
+      "loss": 0.4135,
+      "step": 11859
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0410082106784236e-06,
+      "loss": 0.397,
+      "step": 11860
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0399676840791729e-06,
+      "loss": 0.4091,
+      "step": 11861
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.038927649238296e-06,
+      "loss": 0.3986,
+      "step": 11862
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0378881062128732e-06,
+      "loss": 0.4274,
+      "step": 11863
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0368490550599597e-06,
+      "loss": 0.4082,
+      "step": 11864
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0358104958365778e-06,
+      "loss": 0.3918,
+      "step": 11865
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.034772428599733e-06,
+      "loss": 0.3968,
+      "step": 11866
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.033734853406394e-06,
+      "loss": 0.4242,
+      "step": 11867
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.032697770313511e-06,
+      "loss": 0.4018,
+      "step": 11868
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0316611793779996e-06,
+      "loss": 0.4027,
+      "step": 11869
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.030625080656752e-06,
+      "loss": 0.3765,
+      "step": 11870
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0295894742066326e-06,
+      "loss": 0.3859,
+      "step": 11871
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0285543600844805e-06,
+      "loss": 0.4452,
+      "step": 11872
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0275197383471047e-06,
+      "loss": 0.3816,
+      "step": 11873
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0264856090512908e-06,
+      "loss": 0.392,
+      "step": 11874
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0254519722537948e-06,
+      "loss": 0.4047,
+      "step": 11875
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0244188280113432e-06,
+      "loss": 0.401,
+      "step": 11876
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0233861763806452e-06,
+      "loss": 0.4089,
+      "step": 11877
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.022354017418371e-06,
+      "loss": 0.4232,
+      "step": 11878
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.021322351181172e-06,
+      "loss": 0.3723,
+      "step": 11879
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0202911777256684e-06,
+      "loss": 0.3909,
+      "step": 11880
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.019260497108453e-06,
+      "loss": 0.3629,
+      "step": 11881
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0182303093860969e-06,
+      "loss": 0.3738,
+      "step": 11882
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0172006146151391e-06,
+      "loss": 0.3724,
+      "step": 11883
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0161714128520895e-06,
+      "loss": 0.3861,
+      "step": 11884
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0151427041534402e-06,
+      "loss": 0.4043,
+      "step": 11885
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0141144885756472e-06,
+      "loss": 0.3945,
+      "step": 11886
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.013086766175142e-06,
+      "loss": 0.3781,
+      "step": 11887
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.012059537008332e-06,
+      "loss": 0.4177,
+      "step": 11888
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0110328011315906e-06,
+      "loss": 0.4043,
+      "step": 11889
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.010006558601274e-06,
+      "loss": 0.3945,
+      "step": 11890
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0089808094737042e-06,
+      "loss": 0.3753,
+      "step": 11891
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0079555538051755e-06,
+      "loss": 0.4233,
+      "step": 11892
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.006930791651961e-06,
+      "loss": 0.3848,
+      "step": 11893
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0059065230703025e-06,
+      "loss": 0.4597,
+      "step": 11894
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0048827481164147e-06,
+      "loss": 0.3745,
+      "step": 11895
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0038594668464874e-06,
+      "loss": 0.4151,
+      "step": 11896
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.00283667931668e-06,
+      "loss": 0.4244,
+      "step": 11897
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0018143855831253e-06,
+      "loss": 0.3907,
+      "step": 11898
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.000792585701934e-06,
+      "loss": 0.4119,
+      "step": 11899
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.99771279729186e-07,
+      "loss": 0.4209,
+      "step": 11900
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.987504677209304e-07,
+      "loss": 0.4271,
+      "step": 11901
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.97730149733197e-07,
+      "loss": 0.3709,
+      "step": 11902
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.967103258219824e-07,
+      "loss": 0.4057,
+      "step": 11903
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.956909960432603e-07,
+      "loss": 0.3847,
+      "step": 11904
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.946721604529718e-07,
+      "loss": 0.4202,
+      "step": 11905
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.936538191070355e-07,
+      "loss": 0.384,
+      "step": 11906
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.926359720613433e-07,
+      "loss": 0.3875,
+      "step": 11907
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.916186193717569e-07,
+      "loss": 0.3976,
+      "step": 11908
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.906017610941109e-07,
+      "loss": 0.426,
+      "step": 11909
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.895853972842173e-07,
+      "loss": 0.4299,
+      "step": 11910
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.885695279978557e-07,
+      "loss": 0.4001,
+      "step": 11911
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.875541532907806e-07,
+      "loss": 0.4236,
+      "step": 11912
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.865392732187207e-07,
+      "loss": 0.41,
+      "step": 11913
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.855248878373724e-07,
+      "loss": 0.4504,
+      "step": 11914
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.845109972024136e-07,
+      "loss": 0.414,
+      "step": 11915
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.834976013694875e-07,
+      "loss": 0.3674,
+      "step": 11916
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.824847003942152e-07,
+      "loss": 0.3793,
+      "step": 11917
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.81472294332183e-07,
+      "loss": 0.3755,
+      "step": 11918
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.804603832389615e-07,
+      "loss": 0.4436,
+      "step": 11919
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.794489671700857e-07,
+      "loss": 0.4186,
+      "step": 11920
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.78438046181065e-07,
+      "loss": 0.408,
+      "step": 11921
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.774276203273814e-07,
+      "loss": 0.397,
+      "step": 11922
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.764176896644929e-07,
+      "loss": 0.3831,
+      "step": 11923
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.754082542478282e-07,
+      "loss": 0.3801,
+      "step": 11924
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.743993141327868e-07,
+      "loss": 0.3819,
+      "step": 11925
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.73390869374743e-07,
+      "loss": 0.3627,
+      "step": 11926
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.723829200290457e-07,
+      "loss": 0.4265,
+      "step": 11927
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.713754661510133e-07,
+      "loss": 0.3992,
+      "step": 11928
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.703685077959402e-07,
+      "loss": 0.3878,
+      "step": 11929
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.693620450190877e-07,
+      "loss": 0.4098,
+      "step": 11930
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.683560778756984e-07,
+      "loss": 0.4001,
+      "step": 11931
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.673506064209826e-07,
+      "loss": 0.3742,
+      "step": 11932
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.663456307101227e-07,
+      "loss": 0.403,
+      "step": 11933
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.653411507982757e-07,
+      "loss": 0.3676,
+      "step": 11934
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.6433716674057e-07,
+      "loss": 0.3956,
+      "step": 11935
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.633336785921098e-07,
+      "loss": 0.4148,
+      "step": 11936
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.623306864079707e-07,
+      "loss": 0.4067,
+      "step": 11937
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.61328190243197e-07,
+      "loss": 0.4233,
+      "step": 11938
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.603261901528127e-07,
+      "loss": 0.4094,
+      "step": 11939
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.593246861918104e-07,
+      "loss": 0.4285,
+      "step": 11940
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.583236784151551e-07,
+      "loss": 0.4058,
+      "step": 11941
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.57323166877786e-07,
+      "loss": 0.3797,
+      "step": 11942
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.563231516346139e-07,
+      "loss": 0.3973,
+      "step": 11943
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.553236327405246e-07,
+      "loss": 0.4039,
+      "step": 11944
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.543246102503756e-07,
+      "loss": 0.395,
+      "step": 11945
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.533260842189951e-07,
+      "loss": 0.3958,
+      "step": 11946
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.523280547011871e-07,
+      "loss": 0.3944,
+      "step": 11947
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.513305217517266e-07,
+      "loss": 0.3644,
+      "step": 11948
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.503334854253621e-07,
+      "loss": 0.3612,
+      "step": 11949
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.493369457768143e-07,
+      "loss": 0.415,
+      "step": 11950
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.483409028607748e-07,
+      "loss": 0.4021,
+      "step": 11951
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.473453567319136e-07,
+      "loss": 0.4396,
+      "step": 11952
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.463503074448676e-07,
+      "loss": 0.391,
+      "step": 11953
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.45355755054248e-07,
+      "loss": 0.4237,
+      "step": 11954
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.44361699614641e-07,
+      "loss": 0.3465,
+      "step": 11955
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.433681411806039e-07,
+      "loss": 0.3915,
+      "step": 11956
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.423750798066666e-07,
+      "loss": 0.3963,
+      "step": 11957
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.413825155473311e-07,
+      "loss": 0.4153,
+      "step": 11958
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.403904484570725e-07,
+      "loss": 0.4257,
+      "step": 11959
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.393988785903385e-07,
+      "loss": 0.3867,
+      "step": 11960
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.384078060015522e-07,
+      "loss": 0.378,
+      "step": 11961
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.374172307451068e-07,
+      "loss": 0.3808,
+      "step": 11962
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.364271528753654e-07,
+      "loss": 0.4074,
+      "step": 11963
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.354375724466713e-07,
+      "loss": 0.415,
+      "step": 11964
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.344484895133343e-07,
+      "loss": 0.3735,
+      "step": 11965
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.334599041296388e-07,
+      "loss": 0.3842,
+      "step": 11966
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.324718163498414e-07,
+      "loss": 0.3759,
+      "step": 11967
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.314842262281709e-07,
+      "loss": 0.3947,
+      "step": 11968
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.304971338188329e-07,
+      "loss": 0.4017,
+      "step": 11969
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.295105391760007e-07,
+      "loss": 0.3911,
+      "step": 11970
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.285244423538198e-07,
+      "loss": 0.4768,
+      "step": 11971
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.275388434064158e-07,
+      "loss": 0.4356,
+      "step": 11972
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.26553742387879e-07,
+      "loss": 0.4254,
+      "step": 11973
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.255691393522748e-07,
+      "loss": 0.3885,
+      "step": 11974
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.245850343536422e-07,
+      "loss": 0.4002,
+      "step": 11975
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.236014274459937e-07,
+      "loss": 0.4063,
+      "step": 11976
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.226183186833093e-07,
+      "loss": 0.4334,
+      "step": 11977
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.216357081195493e-07,
+      "loss": 0.4032,
+      "step": 11978
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.206535958086405e-07,
+      "loss": 0.4019,
+      "step": 11979
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.196719818044886e-07,
+      "loss": 0.3844,
+      "step": 11980
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.186908661609639e-07,
+      "loss": 0.3952,
+      "step": 11981
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.177102489319156e-07,
+      "loss": 0.3933,
+      "step": 11982
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.167301301711629e-07,
+      "loss": 0.4138,
+      "step": 11983
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.157505099324982e-07,
+      "loss": 0.4107,
+      "step": 11984
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.14771388269684e-07,
+      "loss": 0.3913,
+      "step": 11985
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.13792765236462e-07,
+      "loss": 0.4192,
+      "step": 11986
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.128146408865402e-07,
+      "loss": 0.3704,
+      "step": 11987
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.118370152736033e-07,
+      "loss": 0.3858,
+      "step": 11988
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.108598884513053e-07,
+      "loss": 0.3914,
+      "step": 11989
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.098832604732754e-07,
+      "loss": 0.4179,
+      "step": 11990
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.089071313931142e-07,
+      "loss": 0.3609,
+      "step": 11991
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.079315012643952e-07,
+      "loss": 0.4219,
+      "step": 11992
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.069563701406637e-07,
+      "loss": 0.3806,
+      "step": 11993
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.059817380754366e-07,
+      "loss": 0.4132,
+      "step": 11994
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.050076051222067e-07,
+      "loss": 0.4417,
+      "step": 11995
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.040339713344415e-07,
+      "loss": 0.4229,
+      "step": 11996
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.030608367655736e-07,
+      "loss": 0.3713,
+      "step": 11997
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.020882014690136e-07,
+      "loss": 0.4034,
+      "step": 11998
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.011160654981422e-07,
+      "loss": 0.3809,
+      "step": 11999
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.001444289063143e-07,
+      "loss": 0.4425,
+      "step": 12000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.991732917468554e-07,
+      "loss": 0.3914,
+      "step": 12001
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.982026540730659e-07,
+      "loss": 0.3763,
+      "step": 12002
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.972325159382167e-07,
+      "loss": 0.3953,
+      "step": 12003
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.962628773955551e-07,
+      "loss": 0.401,
+      "step": 12004
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.952937384982974e-07,
+      "loss": 0.3765,
+      "step": 12005
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.943250992996333e-07,
+      "loss": 0.4096,
+      "step": 12006
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.933569598527247e-07,
+      "loss": 0.4297,
+      "step": 12007
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.923893202107059e-07,
+      "loss": 0.4168,
+      "step": 12008
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.914221804266854e-07,
+      "loss": 0.4018,
+      "step": 12009
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.904555405537408e-07,
+      "loss": 0.429,
+      "step": 12010
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.894894006449283e-07,
+      "loss": 0.3946,
+      "step": 12011
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.885237607532704e-07,
+      "loss": 0.4236,
+      "step": 12012
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.875586209317677e-07,
+      "loss": 0.3998,
+      "step": 12013
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.86593981233389e-07,
+      "loss": 0.4363,
+      "step": 12014
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.856298417110776e-07,
+      "loss": 0.3951,
+      "step": 12015
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.846662024177477e-07,
+      "loss": 0.401,
+      "step": 12016
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.837030634062871e-07,
+      "loss": 0.3832,
+      "step": 12017
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.827404247295579e-07,
+      "loss": 0.379,
+      "step": 12018
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.817782864403901e-07,
+      "loss": 0.373,
+      "step": 12019
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.808166485915915e-07,
+      "loss": 0.3879,
+      "step": 12020
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.798555112359419e-07,
+      "loss": 0.4248,
+      "step": 12021
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.788948744261905e-07,
+      "loss": 0.4134,
+      "step": 12022
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.779347382150605e-07,
+      "loss": 0.3989,
+      "step": 12023
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.769751026552464e-07,
+      "loss": 0.3848,
+      "step": 12024
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.760159677994174e-07,
+      "loss": 0.428,
+      "step": 12025
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.750573337002132e-07,
+      "loss": 0.3785,
+      "step": 12026
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.740992004102478e-07,
+      "loss": 0.393,
+      "step": 12027
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.731415679821076e-07,
+      "loss": 0.3848,
+      "step": 12028
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.721844364683485e-07,
+      "loss": 0.3923,
+      "step": 12029
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.712278059215051e-07,
+      "loss": 0.4101,
+      "step": 12030
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.702716763940788e-07,
+      "loss": 0.3504,
+      "step": 12031
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.693160479385443e-07,
+      "loss": 0.4292,
+      "step": 12032
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.683609206073518e-07,
+      "loss": 0.4071,
+      "step": 12033
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.674062944529216e-07,
+      "loss": 0.382,
+      "step": 12034
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.664521695276451e-07,
+      "loss": 0.4049,
+      "step": 12035
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.654985458838882e-07,
+      "loss": 0.4338,
+      "step": 12036
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 0.4238,
+      "step": 12037
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.635928026502638e-07,
+      "loss": 0.3968,
+      "step": 12038
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.626406831649914e-07,
+      "loss": 0.4081,
+      "step": 12039
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.616890651704268e-07,
+      "loss": 0.3905,
+      "step": 12040
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.607379487187984e-07,
+      "loss": 0.3852,
+      "step": 12041
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.597873338623086e-07,
+      "loss": 0.4027,
+      "step": 12042
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.588372206531293e-07,
+      "loss": 0.3781,
+      "step": 12043
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.578876091434041e-07,
+      "loss": 0.4106,
+      "step": 12044
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.569384993852547e-07,
+      "loss": 0.4793,
+      "step": 12045
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.559898914307696e-07,
+      "loss": 0.392,
+      "step": 12046
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.550417853320126e-07,
+      "loss": 0.4032,
+      "step": 12047
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.540941811410197e-07,
+      "loss": 0.3916,
+      "step": 12048
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.531470789097973e-07,
+      "loss": 0.3993,
+      "step": 12049
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.522004786903271e-07,
+      "loss": 0.3885,
+      "step": 12050
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.512543805345608e-07,
+      "loss": 0.4174,
+      "step": 12051
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.503087844944213e-07,
+      "loss": 0.3945,
+      "step": 12052
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.493636906218117e-07,
+      "loss": 0.3538,
+      "step": 12053
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.484190989685969e-07,
+      "loss": 0.4067,
+      "step": 12054
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.474750095866236e-07,
+      "loss": 0.4115,
+      "step": 12055
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.465314225277055e-07,
+      "loss": 0.4368,
+      "step": 12056
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.455883378436292e-07,
+      "loss": 0.4287,
+      "step": 12057
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.446457555861542e-07,
+      "loss": 0.4107,
+      "step": 12058
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.437036758070139e-07,
+      "loss": 0.431,
+      "step": 12059
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.42762098557911e-07,
+      "loss": 0.3684,
+      "step": 12060
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.418210238905256e-07,
+      "loss": 0.4289,
+      "step": 12061
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.408804518565039e-07,
+      "loss": 0.3927,
+      "step": 12062
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.399403825074725e-07,
+      "loss": 0.3871,
+      "step": 12063
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.390008158950213e-07,
+      "loss": 0.3778,
+      "step": 12064
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.38061752070719e-07,
+      "loss": 0.372,
+      "step": 12065
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.371231910861055e-07,
+      "loss": 0.4157,
+      "step": 12066
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.361851329926896e-07,
+      "loss": 0.4157,
+      "step": 12067
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.352475778419566e-07,
+      "loss": 0.3638,
+      "step": 12068
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.343105256853634e-07,
+      "loss": 0.4378,
+      "step": 12069
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.333739765743399e-07,
+      "loss": 0.4143,
+      "step": 12070
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.324379305602836e-07,
+      "loss": 0.4321,
+      "step": 12071
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.315023876945715e-07,
+      "loss": 0.3716,
+      "step": 12072
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.305673480285492e-07,
+      "loss": 0.3731,
+      "step": 12073
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.296328116135333e-07,
+      "loss": 0.4086,
+      "step": 12074
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.286987785008161e-07,
+      "loss": 0.4191,
+      "step": 12075
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.277652487416565e-07,
+      "loss": 0.4287,
+      "step": 12076
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.26832222387296e-07,
+      "loss": 0.3735,
+      "step": 12077
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.258996994889401e-07,
+      "loss": 0.4221,
+      "step": 12078
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.249676800977657e-07,
+      "loss": 0.4318,
+      "step": 12079
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.240361642649308e-07,
+      "loss": 0.411,
+      "step": 12080
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.231051520415578e-07,
+      "loss": 0.3995,
+      "step": 12081
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.221746434787426e-07,
+      "loss": 0.3564,
+      "step": 12082
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.212446386275563e-07,
+      "loss": 0.4443,
+      "step": 12083
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.203151375390395e-07,
+      "loss": 0.411,
+      "step": 12084
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.193861402642089e-07,
+      "loss": 0.4225,
+      "step": 12085
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.184576468540495e-07,
+      "loss": 0.4041,
+      "step": 12086
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.175296573595215e-07,
+      "loss": 0.4372,
+      "step": 12087
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.166021718315531e-07,
+      "loss": 0.3914,
+      "step": 12088
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.156751903210525e-07,
+      "loss": 0.4306,
+      "step": 12089
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.147487128788934e-07,
+      "loss": 0.4203,
+      "step": 12090
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.13822739555924e-07,
+      "loss": 0.3929,
+      "step": 12091
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.128972704029637e-07,
+      "loss": 0.3693,
+      "step": 12092
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.119723054708084e-07,
+      "loss": 0.4117,
+      "step": 12093
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.110478448102221e-07,
+      "loss": 0.4415,
+      "step": 12094
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.101238884719442e-07,
+      "loss": 0.4017,
+      "step": 12095
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.092004365066797e-07,
+      "loss": 0.4011,
+      "step": 12096
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.082774889651168e-07,
+      "loss": 0.3916,
+      "step": 12097
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.073550458979074e-07,
+      "loss": 0.3621,
+      "step": 12098
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.064331073556786e-07,
+      "loss": 0.4078,
+      "step": 12099
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.0551167338903e-07,
+      "loss": 0.3521,
+      "step": 12100
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.045907440485313e-07,
+      "loss": 0.3873,
+      "step": 12101
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.036703193847295e-07,
+      "loss": 0.4379,
+      "step": 12102
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.027503994481401e-07,
+      "loss": 0.4245,
+      "step": 12103
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.018309842892491e-07,
+      "loss": 0.4108,
+      "step": 12104
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.009120739585186e-07,
+      "loss": 0.4116,
+      "step": 12105
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.999936685063836e-07,
+      "loss": 0.4131,
+      "step": 12106
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.990757679832473e-07,
+      "loss": 0.3601,
+      "step": 12107
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.98158372439487e-07,
+      "loss": 0.4031,
+      "step": 12108
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.972414819254537e-07,
+      "loss": 0.387,
+      "step": 12109
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.963250964914693e-07,
+      "loss": 0.3576,
+      "step": 12110
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.95409216187829e-07,
+      "loss": 0.4234,
+      "step": 12111
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.944938410647996e-07,
+      "loss": 0.405,
+      "step": 12112
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.935789711726171e-07,
+      "loss": 0.4433,
+      "step": 12113
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.926646065614974e-07,
+      "loss": 0.4012,
+      "step": 12114
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.91750747281621e-07,
+      "loss": 0.4192,
+      "step": 12115
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.90837393383146e-07,
+      "loss": 0.3901,
+      "step": 12116
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.899245449161963e-07,
+      "loss": 0.416,
+      "step": 12117
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.890122019308766e-07,
+      "loss": 0.4107,
+      "step": 12118
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.881003644772578e-07,
+      "loss": 0.4042,
+      "step": 12119
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.871890326053854e-07,
+      "loss": 0.3996,
+      "step": 12120
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.862782063652752e-07,
+      "loss": 0.4513,
+      "step": 12121
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.85367885806918e-07,
+      "loss": 0.4035,
+      "step": 12122
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.844580709802752e-07,
+      "loss": 0.3936,
+      "step": 12123
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.83548761935281e-07,
+      "loss": 0.3918,
+      "step": 12124
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.82639958721838e-07,
+      "loss": 0.3918,
+      "step": 12125
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.817316613898296e-07,
+      "loss": 0.3884,
+      "step": 12126
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.808238699891046e-07,
+      "loss": 0.4036,
+      "step": 12127
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.799165845694845e-07,
+      "loss": 0.4069,
+      "step": 12128
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.79009805180766e-07,
+      "loss": 0.3899,
+      "step": 12129
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.781035318727137e-07,
+      "loss": 0.3872,
+      "step": 12130
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.771977646950712e-07,
+      "loss": 0.3844,
+      "step": 12131
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.762925036975466e-07,
+      "loss": 0.4213,
+      "step": 12132
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.753877489298245e-07,
+      "loss": 0.4338,
+      "step": 12133
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.74483500441564e-07,
+      "loss": 0.3661,
+      "step": 12134
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.735797582823912e-07,
+      "loss": 0.4361,
+      "step": 12135
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.726765225019062e-07,
+      "loss": 0.3938,
+      "step": 12136
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.717737931496827e-07,
+      "loss": 0.433,
+      "step": 12137
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.708715702752634e-07,
+      "loss": 0.3802,
+      "step": 12138
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.699698539281686e-07,
+      "loss": 0.3788,
+      "step": 12139
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.690686441578876e-07,
+      "loss": 0.3771,
+      "step": 12140
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.681679410138775e-07,
+      "loss": 0.3451,
+      "step": 12141
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.672677445455789e-07,
+      "loss": 0.3818,
+      "step": 12142
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.663680548023945e-07,
+      "loss": 0.3947,
+      "step": 12143
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.654688718337011e-07,
+      "loss": 0.406,
+      "step": 12144
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.645701956888508e-07,
+      "loss": 0.4259,
+      "step": 12145
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.63672026417166e-07,
+      "loss": 0.4023,
+      "step": 12146
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.627743640679386e-07,
+      "loss": 0.3728,
+      "step": 12147
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.618772086904402e-07,
+      "loss": 0.3861,
+      "step": 12148
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.60980560333906e-07,
+      "loss": 0.4261,
+      "step": 12149
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.600844190475498e-07,
+      "loss": 0.3725,
+      "step": 12150
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.591887848805545e-07,
+      "loss": 0.4215,
+      "step": 12151
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.582936578820754e-07,
+      "loss": 0.3908,
+      "step": 12152
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.573990381012408e-07,
+      "loss": 0.4088,
+      "step": 12153
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.565049255871482e-07,
+      "loss": 0.4327,
+      "step": 12154
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.556113203888716e-07,
+      "loss": 0.3771,
+      "step": 12155
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.54718222555455e-07,
+      "loss": 0.3842,
+      "step": 12156
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.538256321359139e-07,
+      "loss": 0.4084,
+      "step": 12157
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.529335491792389e-07,
+      "loss": 0.4175,
+      "step": 12158
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.520419737343909e-07,
+      "loss": 0.3999,
+      "step": 12159
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.511509058502997e-07,
+      "loss": 0.3663,
+      "step": 12160
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.502603455758728e-07,
+      "loss": 0.4258,
+      "step": 12161
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.493702929599855e-07,
+      "loss": 0.3826,
+      "step": 12162
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.484807480514889e-07,
+      "loss": 0.3975,
+      "step": 12163
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.475917108992014e-07,
+      "loss": 0.3916,
+      "step": 12164
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.467031815519188e-07,
+      "loss": 0.4306,
+      "step": 12165
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.458151600584073e-07,
+      "loss": 0.374,
+      "step": 12166
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.449276464674049e-07,
+      "loss": 0.4264,
+      "step": 12167
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.440406408276191e-07,
+      "loss": 0.4101,
+      "step": 12168
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.431541431877343e-07,
+      "loss": 0.3905,
+      "step": 12169
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.422681535964038e-07,
+      "loss": 0.3837,
+      "step": 12170
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.413826721022532e-07,
+      "loss": 0.4014,
+      "step": 12171
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.404976987538792e-07,
+      "loss": 0.3644,
+      "step": 12172
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.396132335998563e-07,
+      "loss": 0.3699,
+      "step": 12173
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.387292766887244e-07,
+      "loss": 0.3722,
+      "step": 12174
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.378458280689993e-07,
+      "loss": 0.4336,
+      "step": 12175
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.369628877891688e-07,
+      "loss": 0.4078,
+      "step": 12176
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.360804558976897e-07,
+      "loss": 0.4147,
+      "step": 12177
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.351985324429933e-07,
+      "loss": 0.4015,
+      "step": 12178
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.343171174734843e-07,
+      "loss": 0.4186,
+      "step": 12179
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.334362110375337e-07,
+      "loss": 0.404,
+      "step": 12180
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.325558131834942e-07,
+      "loss": 0.4036,
+      "step": 12181
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.316759239596816e-07,
+      "loss": 0.403,
+      "step": 12182
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.307965434143893e-07,
+      "loss": 0.3685,
+      "step": 12183
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.299176715958811e-07,
+      "loss": 0.4397,
+      "step": 12184
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.290393085523906e-07,
+      "loss": 0.4734,
+      "step": 12185
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.281614543321269e-07,
+      "loss": 0.4199,
+      "step": 12186
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.272841089832694e-07,
+      "loss": 0.3848,
+      "step": 12187
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.264072725539695e-07,
+      "loss": 0.4004,
+      "step": 12188
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.255309450923487e-07,
+      "loss": 0.3789,
+      "step": 12189
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.246551266465074e-07,
+      "loss": 0.4231,
+      "step": 12190
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.237798172645127e-07,
+      "loss": 0.4131,
+      "step": 12191
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.229050169944041e-07,
+      "loss": 0.3936,
+      "step": 12192
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.22030725884193e-07,
+      "loss": 0.4079,
+      "step": 12193
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.211569439818645e-07,
+      "loss": 0.4092,
+      "step": 12194
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.202836713353745e-07,
+      "loss": 0.3937,
+      "step": 12195
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.194109079926515e-07,
+      "loss": 0.3907,
+      "step": 12196
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.185386540015937e-07,
+      "loss": 0.4042,
+      "step": 12197
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.176669094100763e-07,
+      "loss": 0.4272,
+      "step": 12198
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.167956742659421e-07,
+      "loss": 0.4151,
+      "step": 12199
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.159249486170095e-07,
+      "loss": 0.3904,
+      "step": 12200
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.150547325110668e-07,
+      "loss": 0.4234,
+      "step": 12201
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.141850259958727e-07,
+      "loss": 0.3862,
+      "step": 12202
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.1331582911916e-07,
+      "loss": 0.4117,
+      "step": 12203
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.12447141928635e-07,
+      "loss": 0.3902,
+      "step": 12204
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.115789644719728e-07,
+      "loss": 0.3757,
+      "step": 12205
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.107112967968221e-07,
+      "loss": 0.4091,
+      "step": 12206
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.098441389508026e-07,
+      "loss": 0.399,
+      "step": 12207
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.089774909815106e-07,
+      "loss": 0.4203,
+      "step": 12208
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.081113529365091e-07,
+      "loss": 0.4124,
+      "step": 12209
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.072457248633335e-07,
+      "loss": 0.4123,
+      "step": 12210
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.063806068094948e-07,
+      "loss": 0.3895,
+      "step": 12211
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.055159988224714e-07,
+      "loss": 0.4156,
+      "step": 12212
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.046519009497165e-07,
+      "loss": 0.3781,
+      "step": 12213
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.037883132386547e-07,
+      "loss": 0.3461,
+      "step": 12214
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.029252357366845e-07,
+      "loss": 0.3304,
+      "step": 12215
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.020626684911724e-07,
+      "loss": 0.4127,
+      "step": 12216
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.012006115494619e-07,
+      "loss": 0.4083,
+      "step": 12217
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.003390649588648e-07,
+      "loss": 0.405,
+      "step": 12218
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.994780287666647e-07,
+      "loss": 0.4417,
+      "step": 12219
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.986175030201181e-07,
+      "loss": 0.4171,
+      "step": 12220
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.977574877664561e-07,
+      "loss": 0.4091,
+      "step": 12221
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.968979830528744e-07,
+      "loss": 0.4197,
+      "step": 12222
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.960389889265517e-07,
+      "loss": 0.4129,
+      "step": 12223
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.951805054346283e-07,
+      "loss": 0.4326,
+      "step": 12224
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.943225326242242e-07,
+      "loss": 0.3705,
+      "step": 12225
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.934650705424262e-07,
+      "loss": 0.4258,
+      "step": 12226
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.926081192362943e-07,
+      "loss": 0.3999,
+      "step": 12227
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.917516787528622e-07,
+      "loss": 0.4512,
+      "step": 12228
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.908957491391343e-07,
+      "loss": 0.4205,
+      "step": 12229
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.900403304420844e-07,
+      "loss": 0.3838,
+      "step": 12230
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.891854227086648e-07,
+      "loss": 0.4006,
+      "step": 12231
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.883310259857944e-07,
+      "loss": 0.4103,
+      "step": 12232
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.874771403203639e-07,
+      "loss": 0.4014,
+      "step": 12233
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.8662376575924e-07,
+      "loss": 0.3886,
+      "step": 12234
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.857709023492587e-07,
+      "loss": 0.42,
+      "step": 12235
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.849185501372268e-07,
+      "loss": 0.4022,
+      "step": 12236
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.84066709169926e-07,
+      "loss": 0.4147,
+      "step": 12237
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.832153794941066e-07,
+      "loss": 0.3824,
+      "step": 12238
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.82364561156491e-07,
+      "loss": 0.4168,
+      "step": 12239
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.8151425420378e-07,
+      "loss": 0.3786,
+      "step": 12240
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.806644586826383e-07,
+      "loss": 0.3894,
+      "step": 12241
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.798151746397064e-07,
+      "loss": 0.3945,
+      "step": 12242
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.78966402121597e-07,
+      "loss": 0.441,
+      "step": 12243
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.781181411748916e-07,
+      "loss": 0.3909,
+      "step": 12244
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.772703918461476e-07,
+      "loss": 0.4201,
+      "step": 12245
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.764231541818922e-07,
+      "loss": 0.4295,
+      "step": 12246
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.755764282286225e-07,
+      "loss": 0.4212,
+      "step": 12247
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.747302140328127e-07,
+      "loss": 0.4072,
+      "step": 12248
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.738845116409043e-07,
+      "loss": 0.3905,
+      "step": 12249
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.730393210993147e-07,
+      "loss": 0.4108,
+      "step": 12250
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.72194642454429e-07,
+      "loss": 0.389,
+      "step": 12251
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.713504757526079e-07,
+      "loss": 0.3965,
+      "step": 12252
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.705068210401799e-07,
+      "loss": 0.4053,
+      "step": 12253
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.696636783634491e-07,
+      "loss": 0.4073,
+      "step": 12254
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.688210477686874e-07,
+      "loss": 0.3968,
+      "step": 12255
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.679789293021466e-07,
+      "loss": 0.4285,
+      "step": 12256
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.67137323010042e-07,
+      "loss": 0.4383,
+      "step": 12257
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.662962289385633e-07,
+      "loss": 0.4459,
+      "step": 12258
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.654556471338747e-07,
+      "loss": 0.4127,
+      "step": 12259
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.646155776421104e-07,
+      "loss": 0.3821,
+      "step": 12260
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.637760205093746e-07,
+      "loss": 0.3928,
+      "step": 12261
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.629369757817461e-07,
+      "loss": 0.4026,
+      "step": 12262
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.620984435052735e-07,
+      "loss": 0.3704,
+      "step": 12263
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.612604237259801e-07,
+      "loss": 0.4046,
+      "step": 12264
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.60422916489859e-07,
+      "loss": 0.3936,
+      "step": 12265
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.595859218428746e-07,
+      "loss": 0.4313,
+      "step": 12266
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.587494398309657e-07,
+      "loss": 0.3778,
+      "step": 12267
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.579134705000412e-07,
+      "loss": 0.3887,
+      "step": 12268
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.57078013895981e-07,
+      "loss": 0.3651,
+      "step": 12269
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.562430700646383e-07,
+      "loss": 0.3562,
+      "step": 12270
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.554086390518355e-07,
+      "loss": 0.367,
+      "step": 12271
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.545747209033749e-07,
+      "loss": 0.4117,
+      "step": 12272
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.537413156650197e-07,
+      "loss": 0.4083,
+      "step": 12273
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.529084233825134e-07,
+      "loss": 0.393,
+      "step": 12274
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.520760441015639e-07,
+      "loss": 0.3875,
+      "step": 12275
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.512441778678591e-07,
+      "loss": 0.4261,
+      "step": 12276
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.504128247270547e-07,
+      "loss": 0.4068,
+      "step": 12277
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.495819847247764e-07,
+      "loss": 0.4182,
+      "step": 12278
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.487516579066233e-07,
+      "loss": 0.4254,
+      "step": 12279
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.47921844318169e-07,
+      "loss": 0.3809,
+      "step": 12280
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.470925440049547e-07,
+      "loss": 0.4092,
+      "step": 12281
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.462637570124964e-07,
+      "loss": 0.4063,
+      "step": 12282
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.454354833862797e-07,
+      "loss": 0.4137,
+      "step": 12283
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.446077231717652e-07,
+      "loss": 0.4329,
+      "step": 12284
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.437804764143807e-07,
+      "loss": 0.3963,
+      "step": 12285
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.429537431595312e-07,
+      "loss": 0.4308,
+      "step": 12286
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.421275234525881e-07,
+      "loss": 0.3873,
+      "step": 12287
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.413018173388996e-07,
+      "loss": 0.4398,
+      "step": 12288
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.404766248637817e-07,
+      "loss": 0.4339,
+      "step": 12289
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.396519460725248e-07,
+      "loss": 0.415,
+      "step": 12290
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.388277810103905e-07,
+      "loss": 0.3868,
+      "step": 12291
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.380041297226091e-07,
+      "loss": 0.4087,
+      "step": 12292
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.371809922543892e-07,
+      "loss": 0.3903,
+      "step": 12293
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.363583686509067e-07,
+      "loss": 0.417,
+      "step": 12294
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.355362589573078e-07,
+      "loss": 0.3804,
+      "step": 12295
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.347146632187163e-07,
+      "loss": 0.3946,
+      "step": 12296
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.338935814802216e-07,
+      "loss": 0.4092,
+      "step": 12297
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.33073013786889e-07,
+      "loss": 0.4324,
+      "step": 12298
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.322529601837524e-07,
+      "loss": 0.4069,
+      "step": 12299
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.314334207158201e-07,
+      "loss": 0.3833,
+      "step": 12300
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.30614395428073e-07,
+      "loss": 0.3297,
+      "step": 12301
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.297958843654606e-07,
+      "loss": 0.3961,
+      "step": 12302
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.289778875729035e-07,
+      "loss": 0.4294,
+      "step": 12303
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.281604050953016e-07,
+      "loss": 0.4043,
+      "step": 12304
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.273434369775178e-07,
+      "loss": 0.403,
+      "step": 12305
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.265269832643905e-07,
+      "loss": 0.4041,
+      "step": 12306
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.257110440007297e-07,
+      "loss": 0.3727,
+      "step": 12307
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.24895619231316e-07,
+      "loss": 0.4157,
+      "step": 12308
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.240807090009049e-07,
+      "loss": 0.4192,
+      "step": 12309
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.232663133542205e-07,
+      "loss": 0.4249,
+      "step": 12310
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.224524323359604e-07,
+      "loss": 0.4322,
+      "step": 12311
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.2163906599079e-07,
+      "loss": 0.4128,
+      "step": 12312
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.208262143633548e-07,
+      "loss": 0.379,
+      "step": 12313
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.200138774982645e-07,
+      "loss": 0.3722,
+      "step": 12314
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.192020554401035e-07,
+      "loss": 0.3858,
+      "step": 12315
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.183907482334262e-07,
+      "loss": 0.4062,
+      "step": 12316
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.175799559227602e-07,
+      "loss": 0.4217,
+      "step": 12317
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.167696785526067e-07,
+      "loss": 0.3992,
+      "step": 12318
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.159599161674346e-07,
+      "loss": 0.3969,
+      "step": 12319
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.15150668811686e-07,
+      "loss": 0.371,
+      "step": 12320
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.143419365297787e-07,
+      "loss": 0.3916,
+      "step": 12321
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.135337193660962e-07,
+      "loss": 0.3942,
+      "step": 12322
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.127260173649963e-07,
+      "loss": 0.4265,
+      "step": 12323
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.119188305708091e-07,
+      "loss": 0.4354,
+      "step": 12324
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.111121590278346e-07,
+      "loss": 0.4175,
+      "step": 12325
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.103060027803476e-07,
+      "loss": 0.4234,
+      "step": 12326
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.095003618725926e-07,
+      "loss": 0.3768,
+      "step": 12327
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.086952363487841e-07,
+      "loss": 0.407,
+      "step": 12328
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.078906262531137e-07,
+      "loss": 0.4066,
+      "step": 12329
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.070865316297392e-07,
+      "loss": 0.3978,
+      "step": 12330
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.062829525227909e-07,
+      "loss": 0.4196,
+      "step": 12331
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.054798889763746e-07,
+      "loss": 0.4253,
+      "step": 12332
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.046773410345641e-07,
+      "loss": 0.3999,
+      "step": 12333
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.03875308741404e-07,
+      "loss": 0.4273,
+      "step": 12334
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.4155,
+      "step": 12335
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.022727912770887e-07,
+      "loss": 0.4206,
+      "step": 12336
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.014723061938843e-07,
+      "loss": 0.3899,
+      "step": 12337
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.006723369352374e-07,
+      "loss": 0.4161,
+      "step": 12338
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.998728835450507e-07,
+      "loss": 0.363,
+      "step": 12339
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.990739460672024e-07,
+      "loss": 0.4129,
+      "step": 12340
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.982755245455407e-07,
+      "loss": 0.4145,
+      "step": 12341
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.974776190238851e-07,
+      "loss": 0.4146,
+      "step": 12342
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.966802295460284e-07,
+      "loss": 0.4063,
+      "step": 12343
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.958833561557331e-07,
+      "loss": 0.42,
+      "step": 12344
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.950869988967378e-07,
+      "loss": 0.3992,
+      "step": 12345
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.942911578127464e-07,
+      "loss": 0.373,
+      "step": 12346
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.934958329474372e-07,
+      "loss": 0.4082,
+      "step": 12347
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.92701024344462e-07,
+      "loss": 0.3674,
+      "step": 12348
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.919067320474425e-07,
+      "loss": 0.386,
+      "step": 12349
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.911129560999707e-07,
+      "loss": 0.3891,
+      "step": 12350
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.903196965456126e-07,
+      "loss": 0.4117,
+      "step": 12351
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.895269534279047e-07,
+      "loss": 0.4006,
+      "step": 12352
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.887347267903587e-07,
+      "loss": 0.4108,
+      "step": 12353
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.879430166764511e-07,
+      "loss": 0.4087,
+      "step": 12354
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.871518231296359e-07,
+      "loss": 0.4296,
+      "step": 12355
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.863611461933361e-07,
+      "loss": 0.435,
+      "step": 12356
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.85570985910946e-07,
+      "loss": 0.4222,
+      "step": 12357
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.84781342325833e-07,
+      "loss": 0.3621,
+      "step": 12358
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.839922154813337e-07,
+      "loss": 0.3629,
+      "step": 12359
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.832036054207602e-07,
+      "loss": 0.4098,
+      "step": 12360
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.824155121873953e-07,
+      "loss": 0.4184,
+      "step": 12361
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.816279358244914e-07,
+      "loss": 0.3871,
+      "step": 12362
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.808408763752738e-07,
+      "loss": 0.4316,
+      "step": 12363
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.800543338829368e-07,
+      "loss": 0.4269,
+      "step": 12364
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.792683083906514e-07,
+      "loss": 0.4576,
+      "step": 12365
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.784827999415565e-07,
+      "loss": 0.3811,
+      "step": 12366
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.77697808578761e-07,
+      "loss": 0.4011,
+      "step": 12367
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.769133343453526e-07,
+      "loss": 0.3673,
+      "step": 12368
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.761293772843823e-07,
+      "loss": 0.4302,
+      "step": 12369
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.753459374388804e-07,
+      "loss": 0.4395,
+      "step": 12370
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.745630148518411e-07,
+      "loss": 0.4338,
+      "step": 12371
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.737806095662357e-07,
+      "loss": 0.4217,
+      "step": 12372
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.729987216250055e-07,
+      "loss": 0.3979,
+      "step": 12373
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.722173510710627e-07,
+      "loss": 0.3957,
+      "step": 12374
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.71436497947292e-07,
+      "loss": 0.4449,
+      "step": 12375
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.706561622965468e-07,
+      "loss": 0.3678,
+      "step": 12376
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.698763441616583e-07,
+      "loss": 0.4152,
+      "step": 12377
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.690970435854249e-07,
+      "loss": 0.3793,
+      "step": 12378
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.683182606106175e-07,
+      "loss": 0.4025,
+      "step": 12379
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.675399952799787e-07,
+      "loss": 0.3933,
+      "step": 12380
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.6676224763622e-07,
+      "loss": 0.4226,
+      "step": 12381
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.659850177220305e-07,
+      "loss": 0.4126,
+      "step": 12382
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.652083055800639e-07,
+      "loss": 0.4295,
+      "step": 12383
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.644321112529505e-07,
+      "loss": 0.3911,
+      "step": 12384
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.636564347832907e-07,
+      "loss": 0.4576,
+      "step": 12385
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.62881276213656e-07,
+      "loss": 0.3918,
+      "step": 12386
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.6210663558659e-07,
+      "loss": 0.3478,
+      "step": 12387
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.61332512944609e-07,
+      "loss": 0.3679,
+      "step": 12388
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.605589083301987e-07,
+      "loss": 0.3953,
+      "step": 12389
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.597858217858165e-07,
+      "loss": 0.3712,
+      "step": 12390
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.590132533538928e-07,
+      "loss": 0.377,
+      "step": 12391
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.582412030768291e-07,
+      "loss": 0.4303,
+      "step": 12392
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.574696709969963e-07,
+      "loss": 0.4021,
+      "step": 12393
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.566986571567401e-07,
+      "loss": 0.4022,
+      "step": 12394
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.559281615983791e-07,
+      "loss": 0.3891,
+      "step": 12395
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.551581843641973e-07,
+      "loss": 0.4216,
+      "step": 12396
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.543887254964564e-07,
+      "loss": 0.4079,
+      "step": 12397
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.536197850373848e-07,
+      "loss": 0.3982,
+      "step": 12398
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.528513630291854e-07,
+      "loss": 0.4032,
+      "step": 12399
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.520834595140323e-07,
+      "loss": 0.415,
+      "step": 12400
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.513160745340695e-07,
+      "loss": 0.4236,
+      "step": 12401
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.505492081314156e-07,
+      "loss": 0.4344,
+      "step": 12402
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.49782860348157e-07,
+      "loss": 0.3984,
+      "step": 12403
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.490170312263565e-07,
+      "loss": 0.3656,
+      "step": 12404
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.48251720808044e-07,
+      "loss": 0.3861,
+      "step": 12405
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.474869291352214e-07,
+      "loss": 0.4219,
+      "step": 12406
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.467226562498651e-07,
+      "loss": 0.3526,
+      "step": 12407
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.459589021939193e-07,
+      "loss": 0.3785,
+      "step": 12408
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.451956670093006e-07,
+      "loss": 0.4036,
+      "step": 12409
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.44432950737902e-07,
+      "loss": 0.3957,
+      "step": 12410
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.436707534215801e-07,
+      "loss": 0.3968,
+      "step": 12411
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.429090751021704e-07,
+      "loss": 0.3812,
+      "step": 12412
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.421479158214738e-07,
+      "loss": 0.4137,
+      "step": 12413
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.413872756212679e-07,
+      "loss": 0.4416,
+      "step": 12414
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.406271545432973e-07,
+      "loss": 0.3963,
+      "step": 12415
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.398675526292796e-07,
+      "loss": 0.4386,
+      "step": 12416
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.391084699209059e-07,
+      "loss": 0.4082,
+      "step": 12417
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.383499064598374e-07,
+      "loss": 0.4384,
+      "step": 12418
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.375918622877052e-07,
+      "loss": 0.3832,
+      "step": 12419
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.36834337446116e-07,
+      "loss": 0.4141,
+      "step": 12420
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.360773319766443e-07,
+      "loss": 0.355,
+      "step": 12421
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.35320845920837e-07,
+      "loss": 0.3721,
+      "step": 12422
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.34564879320213e-07,
+      "loss": 0.4199,
+      "step": 12423
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.338094322162613e-07,
+      "loss": 0.3949,
+      "step": 12424
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.330545046504432e-07,
+      "loss": 0.4024,
+      "step": 12425
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.323000966641956e-07,
+      "loss": 0.3762,
+      "step": 12426
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.315462082989198e-07,
+      "loss": 0.408,
+      "step": 12427
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.307928395959905e-07,
+      "loss": 0.4013,
+      "step": 12428
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.300399905967591e-07,
+      "loss": 0.4286,
+      "step": 12429
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.292876613425435e-07,
+      "loss": 0.3706,
+      "step": 12430
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.285358518746331e-07,
+      "loss": 0.4261,
+      "step": 12431
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.277845622342903e-07,
+      "loss": 0.4057,
+      "step": 12432
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.270337924627467e-07,
+      "loss": 0.4439,
+      "step": 12433
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.262835426012113e-07,
+      "loss": 0.3935,
+      "step": 12434
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.255338126908582e-07,
+      "loss": 0.4025,
+      "step": 12435
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.247846027728332e-07,
+      "loss": 0.4505,
+      "step": 12436
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.2403591288826e-07,
+      "loss": 0.4053,
+      "step": 12437
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.232877430782258e-07,
+      "loss": 0.3755,
+      "step": 12438
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.225400933837954e-07,
+      "loss": 0.4108,
+      "step": 12439
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.217929638460006e-07,
+      "loss": 0.4529,
+      "step": 12440
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.210463545058453e-07,
+      "loss": 0.3951,
+      "step": 12441
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.203002654043099e-07,
+      "loss": 0.4086,
+      "step": 12442
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.195546965823394e-07,
+      "loss": 0.4239,
+      "step": 12443
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.188096480808547e-07,
+      "loss": 0.4183,
+      "step": 12444
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.180651199407449e-07,
+      "loss": 0.3877,
+      "step": 12445
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.173211122028754e-07,
+      "loss": 0.4036,
+      "step": 12446
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.165776249080779e-07,
+      "loss": 0.4013,
+      "step": 12447
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.158346580971574e-07,
+      "loss": 0.4117,
+      "step": 12448
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.150922118108926e-07,
+      "loss": 0.3873,
+      "step": 12449
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.143502860900274e-07,
+      "loss": 0.4364,
+      "step": 12450
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.13608880975286e-07,
+      "loss": 0.4411,
+      "step": 12451
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.128679965073569e-07,
+      "loss": 0.4066,
+      "step": 12452
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.12127632726902e-07,
+      "loss": 0.4194,
+      "step": 12453
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.113877896745567e-07,
+      "loss": 0.3892,
+      "step": 12454
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.106484673909262e-07,
+      "loss": 0.3759,
+      "step": 12455
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.099096659165859e-07,
+      "loss": 0.3959,
+      "step": 12456
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.091713852920854e-07,
+      "loss": 0.4077,
+      "step": 12457
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.084336255579414e-07,
+      "loss": 0.4285,
+      "step": 12458
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.076963867546469e-07,
+      "loss": 0.3635,
+      "step": 12459
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.069596689226653e-07,
+      "loss": 0.4032,
+      "step": 12460
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.062234721024283e-07,
+      "loss": 0.3945,
+      "step": 12461
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.054877963343396e-07,
+      "loss": 0.4495,
+      "step": 12462
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.0475264165878e-07,
+      "loss": 0.3675,
+      "step": 12463
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.040180081160939e-07,
+      "loss": 0.3854,
+      "step": 12464
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.032838957466024e-07,
+      "loss": 0.4267,
+      "step": 12465
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.025503045905933e-07,
+      "loss": 0.3997,
+      "step": 12466
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.018172346883332e-07,
+      "loss": 0.4132,
+      "step": 12467
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.010846860800534e-07,
+      "loss": 0.3919,
+      "step": 12468
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.003526588059582e-07,
+      "loss": 0.3776,
+      "step": 12469
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.996211529062223e-07,
+      "loss": 0.3791,
+      "step": 12470
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.988901684209979e-07,
+      "loss": 0.4029,
+      "step": 12471
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.981597053904008e-07,
+      "loss": 0.3623,
+      "step": 12472
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.974297638545223e-07,
+      "loss": 0.3887,
+      "step": 12473
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.967003438534223e-07,
+      "loss": 0.3893,
+      "step": 12474
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.95971445427137e-07,
+      "loss": 0.3819,
+      "step": 12475
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.952430686156695e-07,
+      "loss": 0.357,
+      "step": 12476
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.945152134589959e-07,
+      "loss": 0.4363,
+      "step": 12477
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.937878799970619e-07,
+      "loss": 0.3662,
+      "step": 12478
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.930610682697889e-07,
+      "loss": 0.4088,
+      "step": 12479
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.923347783170651e-07,
+      "loss": 0.3487,
+      "step": 12480
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.916090101787529e-07,
+      "loss": 0.4007,
+      "step": 12481
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.908837638946828e-07,
+      "loss": 0.4029,
+      "step": 12482
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.901590395046619e-07,
+      "loss": 0.4092,
+      "step": 12483
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.4271,
+      "step": 12484
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.887111565658376e-07,
+      "loss": 0.4278,
+      "step": 12485
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.879879980964986e-07,
+      "loss": 0.4159,
+      "step": 12486
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.872653616801348e-07,
+      "loss": 0.3935,
+      "step": 12487
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.865432473564125e-07,
+      "loss": 0.4005,
+      "step": 12488
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.858216551649608e-07,
+      "loss": 0.395,
+      "step": 12489
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.851005851453816e-07,
+      "loss": 0.4089,
+      "step": 12490
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.843800373372531e-07,
+      "loss": 0.4401,
+      "step": 12491
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.836600117801204e-07,
+      "loss": 0.3918,
+      "step": 12492
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.829405085134997e-07,
+      "loss": 0.3809,
+      "step": 12493
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.822215275768816e-07,
+      "loss": 0.3968,
+      "step": 12494
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.815030690097244e-07,
+      "loss": 0.4102,
+      "step": 12495
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.807851328514624e-07,
+      "loss": 0.4532,
+      "step": 12496
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.80067719141496e-07,
+      "loss": 0.4274,
+      "step": 12497
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.793508279192005e-07,
+      "loss": 0.3697,
+      "step": 12498
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.786344592239211e-07,
+      "loss": 0.4117,
+      "step": 12499
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.779186130949753e-07,
+      "loss": 0.4272,
+      "step": 12500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.772032895716517e-07,
+      "loss": 0.42,
+      "step": 12501
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.764884886932086e-07,
+      "loss": 0.4093,
+      "step": 12502
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7577421049887605e-07,
+      "loss": 0.4203,
+      "step": 12503
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7506045502785704e-07,
+      "loss": 0.374,
+      "step": 12504
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7434722231932686e-07,
+      "loss": 0.405,
+      "step": 12505
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.736345124124264e-07,
+      "loss": 0.411,
+      "step": 12506
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.729223253462756e-07,
+      "loss": 0.401,
+      "step": 12507
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.722106611599597e-07,
+      "loss": 0.399,
+      "step": 12508
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7149951989253763e-07,
+      "loss": 0.4052,
+      "step": 12509
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7078890158304023e-07,
+      "loss": 0.3919,
+      "step": 12510
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.700788062704686e-07,
+      "loss": 0.3688,
+      "step": 12511
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.693692339937916e-07,
+      "loss": 0.3987,
+      "step": 12512
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.686601847919592e-07,
+      "loss": 0.4338,
+      "step": 12513
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6795165870388235e-07,
+      "loss": 0.4239,
+      "step": 12514
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.672436557684479e-07,
+      "loss": 0.3857,
+      "step": 12515
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.665361760245157e-07,
+      "loss": 0.4191,
+      "step": 12516
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6582921951091377e-07,
+      "loss": 0.401,
+      "step": 12517
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.65122786266442e-07,
+      "loss": 0.4221,
+      "step": 12518
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6441687632987065e-07,
+      "loss": 0.4293,
+      "step": 12519
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.637114897399453e-07,
+      "loss": 0.4292,
+      "step": 12520
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6300662653537607e-07,
+      "loss": 0.3717,
+      "step": 12521
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6230228675485325e-07,
+      "loss": 0.3916,
+      "step": 12522
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.615984704370302e-07,
+      "loss": 0.383,
+      "step": 12523
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6089517762053616e-07,
+      "loss": 0.4341,
+      "step": 12524
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.601924083439702e-07,
+      "loss": 0.3954,
+      "step": 12525
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5949016264590254e-07,
+      "loss": 0.4202,
+      "step": 12526
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.587884405648757e-07,
+      "loss": 0.4326,
+      "step": 12527
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5808724213940113e-07,
+      "loss": 0.408,
+      "step": 12528
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.573865674079625e-07,
+      "loss": 0.3859,
+      "step": 12529
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.56686416409019e-07,
+      "loss": 0.4068,
+      "step": 12530
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5598678918099325e-07,
+      "loss": 0.3673,
+      "step": 12531
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.552876857622868e-07,
+      "loss": 0.3809,
+      "step": 12532
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5458910619126664e-07,
+      "loss": 0.4046,
+      "step": 12533
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5389105050627434e-07,
+      "loss": 0.4131,
+      "step": 12534
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5319351874562156e-07,
+      "loss": 0.3987,
+      "step": 12535
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5249651094758984e-07,
+      "loss": 0.4027,
+      "step": 12536
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.518000271504353e-07,
+      "loss": 0.3493,
+      "step": 12537
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5110406739238276e-07,
+      "loss": 0.3895,
+      "step": 12538
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.504086317116274e-07,
+      "loss": 0.3491,
+      "step": 12539
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4971372014634084e-07,
+      "loss": 0.4056,
+      "step": 12540
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4901933273466036e-07,
+      "loss": 0.4311,
+      "step": 12541
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4832546951469546e-07,
+      "loss": 0.4174,
+      "step": 12542
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.476321305245301e-07,
+      "loss": 0.3918,
+      "step": 12543
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4693931580221395e-07,
+      "loss": 0.3671,
+      "step": 12544
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4624702538577424e-07,
+      "loss": 0.4401,
+      "step": 12545
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4555525931320285e-07,
+      "loss": 0.3926,
+      "step": 12546
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4486401762246947e-07,
+      "loss": 0.3873,
+      "step": 12547
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.441733003515114e-07,
+      "loss": 0.3749,
+      "step": 12548
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4348310753823734e-07,
+      "loss": 0.4093,
+      "step": 12549
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.42793439220528e-07,
+      "loss": 0.4201,
+      "step": 12550
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.421042954362331e-07,
+      "loss": 0.427,
+      "step": 12551
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4141567622317696e-07,
+      "loss": 0.4225,
+      "step": 12552
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.407275816191536e-07,
+      "loss": 0.4039,
+      "step": 12553
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4004001166192523e-07,
+      "loss": 0.3778,
+      "step": 12554
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3935296638923153e-07,
+      "loss": 0.4208,
+      "step": 12555
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3866644583877794e-07,
+      "loss": 0.3791,
+      "step": 12556
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.379804500482443e-07,
+      "loss": 0.4261,
+      "step": 12557
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.372949790552805e-07,
+      "loss": 0.4331,
+      "step": 12558
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.366100328975076e-07,
+      "loss": 0.4129,
+      "step": 12559
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3592561161251766e-07,
+      "loss": 0.4014,
+      "step": 12560
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3524171523787406e-07,
+      "loss": 0.4071,
+      "step": 12561
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3455834381111115e-07,
+      "loss": 0.3906,
+      "step": 12562
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3387549736973347e-07,
+      "loss": 0.4187,
+      "step": 12563
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.331931759512198e-07,
+      "loss": 0.3987,
+      "step": 12564
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3251137959302023e-07,
+      "loss": 0.388,
+      "step": 12565
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.318301083325516e-07,
+      "loss": 0.4366,
+      "step": 12566
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.311493622072038e-07,
+      "loss": 0.3922,
+      "step": 12567
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.3046914125434157e-07,
+      "loss": 0.4115,
+      "step": 12568
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2978944551129496e-07,
+      "loss": 0.3953,
+      "step": 12569
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2911027501536973e-07,
+      "loss": 0.4038,
+      "step": 12570
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2843162980383936e-07,
+      "loss": 0.4094,
+      "step": 12571
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.277535099139529e-07,
+      "loss": 0.4048,
+      "step": 12572
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2707591538292516e-07,
+      "loss": 0.4277,
+      "step": 12573
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2639884624794846e-07,
+      "loss": 0.4178,
+      "step": 12574
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2572230254617983e-07,
+      "loss": 0.389,
+      "step": 12575
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.250462843147518e-07,
+      "loss": 0.431,
+      "step": 12576
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2437079159076577e-07,
+      "loss": 0.373,
+      "step": 12577
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2369582441129655e-07,
+      "loss": 0.3779,
+      "step": 12578
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2302138281338554e-07,
+      "loss": 0.4105,
+      "step": 12579
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2234746683405215e-07,
+      "loss": 0.3732,
+      "step": 12580
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2167407651028e-07,
+      "loss": 0.3723,
+      "step": 12581
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2100121187903075e-07,
+      "loss": 0.397,
+      "step": 12582
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.203288729772326e-07,
+      "loss": 0.418,
+      "step": 12583
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1965705984178374e-07,
+      "loss": 0.3778,
+      "step": 12584
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1898577250955806e-07,
+      "loss": 0.3724,
+      "step": 12585
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.183150110173972e-07,
+      "loss": 0.3835,
+      "step": 12586
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.17644775402114e-07,
+      "loss": 0.3979,
+      "step": 12587
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.169750657004934e-07,
+      "loss": 0.4161,
+      "step": 12588
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.163058819492938e-07,
+      "loss": 0.3946,
+      "step": 12589
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.156372241852402e-07,
+      "loss": 0.4234,
+      "step": 12590
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1496909244503223e-07,
+      "loss": 0.3914,
+      "step": 12591
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.143014867653383e-07,
+      "loss": 0.3726,
+      "step": 12592
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.136344071827991e-07,
+      "loss": 0.4137,
+      "step": 12593
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.129678537340276e-07,
+      "loss": 0.4,
+      "step": 12594
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.123018264556056e-07,
+      "loss": 0.4014,
+      "step": 12595
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1163632538408493e-07,
+      "loss": 0.4236,
+      "step": 12596
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1097135055599535e-07,
+      "loss": 0.3908,
+      "step": 12597
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1030690200782765e-07,
+      "loss": 0.3953,
+      "step": 12598
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0964297977605483e-07,
+      "loss": 0.4257,
+      "step": 12599
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0897958389711115e-07,
+      "loss": 0.3809,
+      "step": 12600
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.083167144074074e-07,
+      "loss": 0.4016,
+      "step": 12601
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0765437134332453e-07,
+      "loss": 0.3886,
+      "step": 12602
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0699255474121457e-07,
+      "loss": 0.3997,
+      "step": 12603
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.063312646373985e-07,
+      "loss": 0.3699,
+      "step": 12604
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.056705010681716e-07,
+      "loss": 0.4037,
+      "step": 12605
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.050102640697984e-07,
+      "loss": 0.4044,
+      "step": 12606
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0435055367851636e-07,
+      "loss": 0.4193,
+      "step": 12607
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0369136993053226e-07,
+      "loss": 0.4032,
+      "step": 12608
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.030327128620237e-07,
+      "loss": 0.4292,
+      "step": 12609
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.023745825091407e-07,
+      "loss": 0.3891,
+      "step": 12610
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.017169789080033e-07,
+      "loss": 0.4094,
+      "step": 12611
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.010599020947026e-07,
+      "loss": 0.3784,
+      "step": 12612
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.00403352105303e-07,
+      "loss": 0.3828,
+      "step": 12613
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.997473289758369e-07,
+      "loss": 0.3933,
+      "step": 12614
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.990918327423099e-07,
+      "loss": 0.4069,
+      "step": 12615
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9843686344069765e-07,
+      "loss": 0.4227,
+      "step": 12616
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9778242110694696e-07,
+      "loss": 0.3927,
+      "step": 12617
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.971285057769769e-07,
+      "loss": 0.4159,
+      "step": 12618
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9647511748667653e-07,
+      "loss": 0.4349,
+      "step": 12619
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9582225627190386e-07,
+      "loss": 0.3702,
+      "step": 12620
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9516992216849237e-07,
+      "loss": 0.403,
+      "step": 12621
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.945181152122446e-07,
+      "loss": 0.3802,
+      "step": 12622
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9386683543893303e-07,
+      "loss": 0.4091,
+      "step": 12623
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9321608288430234e-07,
+      "loss": 0.3759,
+      "step": 12624
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9256585758406964e-07,
+      "loss": 0.3447,
+      "step": 12625
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.919161595739207e-07,
+      "loss": 0.4204,
+      "step": 12626
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9126698888951156e-07,
+      "loss": 0.4099,
+      "step": 12627
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9061834556647247e-07,
+      "loss": 0.3947,
+      "step": 12628
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.89970229640404e-07,
+      "loss": 0.4148,
+      "step": 12629
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8932264114687646e-07,
+      "loss": 0.4112,
+      "step": 12630
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8867558012143145e-07,
+      "loss": 0.3927,
+      "step": 12631
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8802904659958063e-07,
+      "loss": 0.4199,
+      "step": 12632
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8738304061681107e-07,
+      "loss": 0.3915,
+      "step": 12633
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.867375622085767e-07,
+      "loss": 0.3839,
+      "step": 12634
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8609261141030364e-07,
+      "loss": 0.4086,
+      "step": 12635
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8544818825738683e-07,
+      "loss": 0.4264,
+      "step": 12636
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.84804292785198e-07,
+      "loss": 0.4172,
+      "step": 12637
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8416092502907455e-07,
+      "loss": 0.4316,
+      "step": 12638
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8351808502432697e-07,
+      "loss": 0.4148,
+      "step": 12639
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.82875772806236e-07,
+      "loss": 0.4337,
+      "step": 12640
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8223398841005565e-07,
+      "loss": 0.399,
+      "step": 12641
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.815927318710089e-07,
+      "loss": 0.4158,
+      "step": 12642
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.809520032242897e-07,
+      "loss": 0.4005,
+      "step": 12643
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8031180250506226e-07,
+      "loss": 0.3932,
+      "step": 12644
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.796721297484651e-07,
+      "loss": 0.3929,
+      "step": 12645
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7903298498960574e-07,
+      "loss": 0.3656,
+      "step": 12646
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7839436826356045e-07,
+      "loss": 0.3791,
+      "step": 12647
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.777562796053813e-07,
+      "loss": 0.4338,
+      "step": 12648
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.771187190500869e-07,
+      "loss": 0.4121,
+      "step": 12649
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7648168663267037e-07,
+      "loss": 0.3632,
+      "step": 12650
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7584518238809484e-07,
+      "loss": 0.431,
+      "step": 12651
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.752092063512913e-07,
+      "loss": 0.3953,
+      "step": 12652
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7457375855716517e-07,
+      "loss": 0.4554,
+      "step": 12653
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7393883904059405e-07,
+      "loss": 0.405,
+      "step": 12654
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.733044478364234e-07,
+      "loss": 0.4393,
+      "step": 12655
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7267058497947096e-07,
+      "loss": 0.4032,
+      "step": 12656
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7203725050452333e-07,
+      "loss": 0.3957,
+      "step": 12657
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7140444444634383e-07,
+      "loss": 0.4214,
+      "step": 12658
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.707721668396613e-07,
+      "loss": 0.4157,
+      "step": 12659
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7014041771917696e-07,
+      "loss": 0.4099,
+      "step": 12660
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.69509197119563e-07,
+      "loss": 0.3962,
+      "step": 12661
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6887850507546506e-07,
+      "loss": 0.4491,
+      "step": 12662
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.682483416214955e-07,
+      "loss": 0.4262,
+      "step": 12663
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.676187067922421e-07,
+      "loss": 0.3996,
+      "step": 12664
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6698960062225953e-07,
+      "loss": 0.405,
+      "step": 12665
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.663610231460768e-07,
+      "loss": 0.395,
+      "step": 12666
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.657329743981919e-07,
+      "loss": 0.3966,
+      "step": 12667
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6510545441307387e-07,
+      "loss": 0.428,
+      "step": 12668
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.64478463225163e-07,
+      "loss": 0.4014,
+      "step": 12669
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6385200086887287e-07,
+      "loss": 0.4437,
+      "step": 12670
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.632260673785837e-07,
+      "loss": 0.4194,
+      "step": 12671
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6260066278865025e-07,
+      "loss": 0.3977,
+      "step": 12672
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.619757871333973e-07,
+      "loss": 0.4128,
+      "step": 12673
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6135144044711743e-07,
+      "loss": 0.4245,
+      "step": 12674
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6072762276408103e-07,
+      "loss": 0.4148,
+      "step": 12675
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6010433411852283e-07,
+      "loss": 0.3997,
+      "step": 12676
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.5948157454465005e-07,
+      "loss": 0.437,
+      "step": 12677
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5885934407664523e-07,
+      "loss": 0.4032,
+      "step": 12678
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.582376427486578e-07,
+      "loss": 0.3957,
+      "step": 12679
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.576164705948071e-07,
+      "loss": 0.4275,
+      "step": 12680
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5699582764918695e-07,
+      "loss": 0.4138,
+      "step": 12681
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.563757139458579e-07,
+      "loss": 0.3727,
+      "step": 12682
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.557561295188583e-07,
+      "loss": 0.409,
+      "step": 12683
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.551370744021898e-07,
+      "loss": 0.4223,
+      "step": 12684
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.545185486298275e-07,
+      "loss": 0.4299,
+      "step": 12685
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5390055223572197e-07,
+      "loss": 0.417,
+      "step": 12686
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.532830852537883e-07,
+      "loss": 0.3975,
+      "step": 12687
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5266614771791717e-07,
+      "loss": 0.3894,
+      "step": 12688
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.520497396619671e-07,
+      "loss": 0.402,
+      "step": 12689
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5143386111976765e-07,
+      "loss": 0.3654,
+      "step": 12690
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5081851212512176e-07,
+      "loss": 0.4157,
+      "step": 12691
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5020369271180244e-07,
+      "loss": 0.3735,
+      "step": 12692
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.495894029135516e-07,
+      "loss": 0.4312,
+      "step": 12693
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4897564276408556e-07,
+      "loss": 0.422,
+      "step": 12694
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4836241229708854e-07,
+      "loss": 0.3873,
+      "step": 12695
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.477497115462181e-07,
+      "loss": 0.4245,
+      "step": 12696
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.471375405450994e-07,
+      "loss": 0.3979,
+      "step": 12697
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.465258993273313e-07,
+      "loss": 0.414,
+      "step": 12698
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4591478792648237e-07,
+      "loss": 0.4148,
+      "step": 12699
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4530420637609365e-07,
+      "loss": 0.3965,
+      "step": 12700
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4469415470967494e-07,
+      "loss": 0.4073,
+      "step": 12701
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.440846329607095e-07,
+      "loss": 0.3864,
+      "step": 12702
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.434756411626494e-07,
+      "loss": 0.4057,
+      "step": 12703
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.42867179348918e-07,
+      "loss": 0.4003,
+      "step": 12704
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.422592475529096e-07,
+      "loss": 0.3869,
+      "step": 12705
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4165184580799094e-07,
+      "loss": 0.4092,
+      "step": 12706
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4104497414749637e-07,
+      "loss": 0.3863,
+      "step": 12707
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4043863260473373e-07,
+      "loss": 0.4203,
+      "step": 12708
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.398328212129809e-07,
+      "loss": 0.4009,
+      "step": 12709
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3922754000549007e-07,
+      "loss": 0.3855,
+      "step": 12710
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3862278901547805e-07,
+      "loss": 0.3965,
+      "step": 12711
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3801856827613724e-07,
+      "loss": 0.3692,
+      "step": 12712
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3741487782062766e-07,
+      "loss": 0.3999,
+      "step": 12713
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3681171768208397e-07,
+      "loss": 0.3626,
+      "step": 12714
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3620908789360863e-07,
+      "loss": 0.3696,
+      "step": 12715
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.356069884882751e-07,
+      "loss": 0.4042,
+      "step": 12716
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3500541949913035e-07,
+      "loss": 0.3932,
+      "step": 12717
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3440438095919126e-07,
+      "loss": 0.3542,
+      "step": 12718
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.338038729014426e-07,
+      "loss": 0.4085,
+      "step": 12719
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3320389535884565e-07,
+      "loss": 0.3956,
+      "step": 12720
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.326044483643265e-07,
+      "loss": 0.38,
+      "step": 12721
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3200553195078533e-07,
+      "loss": 0.3955,
+      "step": 12722
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.314071461510937e-07,
+      "loss": 0.4227,
+      "step": 12723
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.308092909980909e-07,
+      "loss": 0.4028,
+      "step": 12724
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.302119665245929e-07,
+      "loss": 0.4458,
+      "step": 12725
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2961517276338005e-07,
+      "loss": 0.4217,
+      "step": 12726
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.290189097472096e-07,
+      "loss": 0.4133,
+      "step": 12727
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2842317750880317e-07,
+      "loss": 0.3892,
+      "step": 12728
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2782797608085894e-07,
+      "loss": 0.359,
+      "step": 12729
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2723330549604306e-07,
+      "loss": 0.4143,
+      "step": 12730
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.266391657869927e-07,
+      "loss": 0.4236,
+      "step": 12731
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2604555698631636e-07,
+      "loss": 0.4119,
+      "step": 12732
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.254524791265934e-07,
+      "loss": 0.433,
+      "step": 12733
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2485993224037337e-07,
+      "loss": 0.3949,
+      "step": 12734
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.242679163601803e-07,
+      "loss": 0.3879,
+      "step": 12735
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.236764315185037e-07,
+      "loss": 0.4282,
+      "step": 12736
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.230854777478065e-07,
+      "loss": 0.3952,
+      "step": 12737
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.224950550805228e-07,
+      "loss": 0.376,
+      "step": 12738
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2190516354905663e-07,
+      "loss": 0.387,
+      "step": 12739
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.213158031857844e-07,
+      "loss": 0.3935,
+      "step": 12740
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2072697402305006e-07,
+      "loss": 0.4305,
+      "step": 12741
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.201386760931735e-07,
+      "loss": 0.3857,
+      "step": 12742
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.195509094284399e-07,
+      "loss": 0.4066,
+      "step": 12743
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.189636740611102e-07,
+      "loss": 0.3935,
+      "step": 12744
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1837697002341293e-07,
+      "loss": 0.3919,
+      "step": 12745
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.17790797347548e-07,
+      "loss": 0.3546,
+      "step": 12746
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.172051560656886e-07,
+      "loss": 0.3837,
+      "step": 12747
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1662004620997443e-07,
+      "loss": 0.3869,
+      "step": 12748
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1603546781251994e-07,
+      "loss": 0.4248,
+      "step": 12749
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.154514209054071e-07,
+      "loss": 0.3711,
+      "step": 12750
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1486790552069157e-07,
+      "loss": 0.4179,
+      "step": 12751
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.142849216903998e-07,
+      "loss": 0.419,
+      "step": 12752
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1370246944652736e-07,
+      "loss": 0.4005,
+      "step": 12753
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.131205488210409e-07,
+      "loss": 0.3662,
+      "step": 12754
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.125391598458793e-07,
+      "loss": 0.4176,
+      "step": 12755
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1195830255294933e-07,
+      "loss": 0.403,
+      "step": 12756
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.113779769741321e-07,
+      "loss": 0.4058,
+      "step": 12757
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.107981831412765e-07,
+      "loss": 0.425,
+      "step": 12758
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.10218921086205e-07,
+      "loss": 0.3945,
+      "step": 12759
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.0964019084070764e-07,
+      "loss": 0.429,
+      "step": 12760
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.090619924365512e-07,
+      "loss": 0.4262,
+      "step": 12761
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.08484325905466e-07,
+      "loss": 0.3814,
+      "step": 12762
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0790719127915647e-07,
+      "loss": 0.3744,
+      "step": 12763
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0733058858929967e-07,
+      "loss": 0.3915,
+      "step": 12764
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.067545178675391e-07,
+      "loss": 0.4137,
+      "step": 12765
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.061789791454928e-07,
+      "loss": 0.4152,
+      "step": 12766
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.056039724547488e-07,
+      "loss": 0.3992,
+      "step": 12767
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.050294978268653e-07,
+      "loss": 0.3937,
+      "step": 12768
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0445555529337143e-07,
+      "loss": 0.3684,
+      "step": 12769
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.038821448857676e-07,
+      "loss": 0.3654,
+      "step": 12770
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.033092666355242e-07,
+      "loss": 0.4166,
+      "step": 12771
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0273692057408264e-07,
+      "loss": 0.3848,
+      "step": 12772
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.021651067328546e-07,
+      "loss": 0.4086,
+      "step": 12773
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.015938251432238e-07,
+      "loss": 0.3966,
+      "step": 12774
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.010230758365462e-07,
+      "loss": 0.4355,
+      "step": 12775
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0045285884414246e-07,
+      "loss": 0.3585,
+      "step": 12776
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.998831741973129e-07,
+      "loss": 0.3917,
+      "step": 12777
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.993140219273205e-07,
+      "loss": 0.4092,
+      "step": 12778
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.987454020654035e-07,
+      "loss": 0.3724,
+      "step": 12779
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9817731464276914e-07,
+      "loss": 0.4104,
+      "step": 12780
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.976097596905969e-07,
+      "loss": 0.38,
+      "step": 12781
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.970427372400353e-07,
+      "loss": 0.4074,
+      "step": 12782
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9647624732220605e-07,
+      "loss": 0.4516,
+      "step": 12783
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9591028996819873e-07,
+      "loss": 0.378,
+      "step": 12784
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.95344865209074e-07,
+      "loss": 0.4079,
+      "step": 12785
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9477997307586716e-07,
+      "loss": 0.4019,
+      "step": 12786
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9421561359957997e-07,
+      "loss": 0.3662,
+      "step": 12787
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.936517868111877e-07,
+      "loss": 0.378,
+      "step": 12788
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.930884927416333e-07,
+      "loss": 0.3953,
+      "step": 12789
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9252573142183327e-07,
+      "loss": 0.3631,
+      "step": 12790
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9196350288267284e-07,
+      "loss": 0.3908,
+      "step": 12791
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9140180715501176e-07,
+      "loss": 0.4315,
+      "step": 12792
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.908406442696754e-07,
+      "loss": 0.4265,
+      "step": 12793
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.902800142574635e-07,
+      "loss": 0.395,
+      "step": 12794
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.897199171491449e-07,
+      "loss": 0.4034,
+      "step": 12795
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8916035297546054e-07,
+      "loss": 0.4226,
+      "step": 12796
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.886013217671213e-07,
+      "loss": 0.3935,
+      "step": 12797
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.880428235548083e-07,
+      "loss": 0.4004,
+      "step": 12798
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.874848583691714e-07,
+      "loss": 0.3818,
+      "step": 12799
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.869274262408384e-07,
+      "loss": 0.404,
+      "step": 12800
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8637052720040026e-07,
+      "loss": 0.4426,
+      "step": 12801
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.858141612784215e-07,
+      "loss": 0.4109,
+      "step": 12802
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8525832850544e-07,
+      "loss": 0.4213,
+      "step": 12803
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.84703028911959e-07,
+      "loss": 0.3814,
+      "step": 12804
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.841482625284564e-07,
+      "loss": 0.3978,
+      "step": 12805
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8359402938538005e-07,
+      "loss": 0.4107,
+      "step": 12806
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8304032951314675e-07,
+      "loss": 0.4131,
+      "step": 12807
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.824871629421477e-07,
+      "loss": 0.3649,
+      "step": 12808
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8193452970274094e-07,
+      "loss": 0.3733,
+      "step": 12809
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8138242982525765e-07,
+      "loss": 0.3824,
+      "step": 12810
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.808308633399992e-07,
+      "loss": 0.3877,
+      "step": 12811
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8027983027723694e-07,
+      "loss": 0.3964,
+      "step": 12812
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.797293306672144e-07,
+      "loss": 0.4173,
+      "step": 12813
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7917936454014525e-07,
+      "loss": 0.4057,
+      "step": 12814
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.786299319262109e-07,
+      "loss": 0.4244,
+      "step": 12815
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7808103285556834e-07,
+      "loss": 0.4098,
+      "step": 12816
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.775326673583434e-07,
+      "loss": 0.3883,
+      "step": 12817
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.769848354646321e-07,
+      "loss": 0.4427,
+      "step": 12818
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.764375372044992e-07,
+      "loss": 0.436,
+      "step": 12819
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.758907726079851e-07,
+      "loss": 0.4043,
+      "step": 12820
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7534454170509816e-07,
+      "loss": 0.4087,
+      "step": 12821
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.747988445258165e-07,
+      "loss": 0.456,
+      "step": 12822
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7425368110008845e-07,
+      "loss": 0.4097,
+      "step": 12823
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7370905145783664e-07,
+      "loss": 0.4424,
+      "step": 12824
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7316495562895285e-07,
+      "loss": 0.3807,
+      "step": 12825
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.726213936432964e-07,
+      "loss": 0.3597,
+      "step": 12826
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7207836553070133e-07,
+      "loss": 0.4171,
+      "step": 12827
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.715358713209715e-07,
+      "loss": 0.4323,
+      "step": 12828
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.709939110438808e-07,
+      "loss": 0.4015,
+      "step": 12829
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.704524847291723e-07,
+      "loss": 0.4049,
+      "step": 12830
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.699115924065632e-07,
+      "loss": 0.405,
+      "step": 12831
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6937123410573863e-07,
+      "loss": 0.4173,
+      "step": 12832
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.68831409856356e-07,
+      "loss": 0.4026,
+      "step": 12833
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.682921196880428e-07,
+      "loss": 0.3898,
+      "step": 12834
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.677533636303964e-07,
+      "loss": 0.3603,
+      "step": 12835
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6721514171298647e-07,
+      "loss": 0.4202,
+      "step": 12836
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.666774539653516e-07,
+      "loss": 0.4069,
+      "step": 12837
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.661403004170038e-07,
+      "loss": 0.4009,
+      "step": 12838
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.656036810974216e-07,
+      "loss": 0.3852,
+      "step": 12839
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6506759603605824e-07,
+      "loss": 0.4229,
+      "step": 12840
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.645320452623357e-07,
+      "loss": 0.4014,
+      "step": 12841
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.639970288056459e-07,
+      "loss": 0.4017,
+      "step": 12842
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.634625466953544e-07,
+      "loss": 0.4187,
+      "step": 12843
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6292859896079215e-07,
+      "loss": 0.4336,
+      "step": 12844
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6239518563126785e-07,
+      "loss": 0.4007,
+      "step": 12845
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6186230673605595e-07,
+      "loss": 0.4233,
+      "step": 12846
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6132996230440076e-07,
+      "loss": 0.4154,
+      "step": 12847
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.607981523655212e-07,
+      "loss": 0.3888,
+      "step": 12848
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.60266876948605e-07,
+      "loss": 0.395,
+      "step": 12849
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.597361360828099e-07,
+      "loss": 0.364,
+      "step": 12850
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.592059297972649e-07,
+      "loss": 0.3637,
+      "step": 12851
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5867625812106887e-07,
+      "loss": 0.4061,
+      "step": 12852
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.581471210832931e-07,
+      "loss": 0.4167,
+      "step": 12853
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5761851871297873e-07,
+      "loss": 0.4256,
+      "step": 12854
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5709045103913475e-07,
+      "loss": 0.4084,
+      "step": 12855
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5656291809074694e-07,
+      "loss": 0.4153,
+      "step": 12856
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5603591989676657e-07,
+      "loss": 0.4177,
+      "step": 12857
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5550945648611716e-07,
+      "loss": 0.4087,
+      "step": 12858
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5498352788769334e-07,
+      "loss": 0.4074,
+      "step": 12859
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.544581341303598e-07,
+      "loss": 0.3847,
+      "step": 12860
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.539332752429502e-07,
+      "loss": 0.4124,
+      "step": 12861
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.534089512542737e-07,
+      "loss": 0.411,
+      "step": 12862
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.52885162193105e-07,
+      "loss": 0.39,
+      "step": 12863
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5236190808819117e-07,
+      "loss": 0.415,
+      "step": 12864
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5183918896825255e-07,
+      "loss": 0.4198,
+      "step": 12865
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.513170048619762e-07,
+      "loss": 0.3784,
+      "step": 12866
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.507953557980214e-07,
+      "loss": 0.4272,
+      "step": 12867
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.502742418050186e-07,
+      "loss": 0.4016,
+      "step": 12868
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.497536629115671e-07,
+      "loss": 0.4461,
+      "step": 12869
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4923361914623967e-07,
+      "loss": 0.4319,
+      "step": 12870
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.48714110537579e-07,
+      "loss": 0.3784,
+      "step": 12871
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.481951371140945e-07,
+      "loss": 0.402,
+      "step": 12872
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4767669890427225e-07,
+      "loss": 0.428,
+      "step": 12873
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.47158795936564e-07,
+      "loss": 0.3988,
+      "step": 12874
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4664142823939585e-07,
+      "loss": 0.3407,
+      "step": 12875
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4612459584116066e-07,
+      "loss": 0.426,
+      "step": 12876
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4560829877022575e-07,
+      "loss": 0.3679,
+      "step": 12877
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.450925370549262e-07,
+      "loss": 0.3862,
+      "step": 12878
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.445773107235694e-07,
+      "loss": 0.358,
+      "step": 12879
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.440626198044327e-07,
+      "loss": 0.4052,
+      "step": 12880
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.435484643257646e-07,
+      "loss": 0.4091,
+      "step": 12881
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4303484431578264e-07,
+      "loss": 0.3813,
+      "step": 12882
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4252175980267855e-07,
+      "loss": 0.4286,
+      "step": 12883
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4200921081460993e-07,
+      "loss": 0.3831,
+      "step": 12884
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4149719737970754e-07,
+      "loss": 0.4146,
+      "step": 12885
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.409857195260723e-07,
+      "loss": 0.4041,
+      "step": 12886
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4047477728177726e-07,
+      "loss": 0.3902,
+      "step": 12887
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3996437067486223e-07,
+      "loss": 0.4105,
+      "step": 12888
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3945449973334366e-07,
+      "loss": 0.4229,
+      "step": 12889
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3894516448520365e-07,
+      "loss": 0.4199,
+      "step": 12890
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.384363649583954e-07,
+      "loss": 0.4261,
+      "step": 12891
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.379281011808443e-07,
+      "loss": 0.3763,
+      "step": 12892
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3742037318044586e-07,
+      "loss": 0.4302,
+      "step": 12893
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3691318098506445e-07,
+      "loss": 0.4127,
+      "step": 12894
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3640652462253888e-07,
+      "loss": 0.3857,
+      "step": 12895
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3590040412067472e-07,
+      "loss": 0.3827,
+      "step": 12896
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3539481950725085e-07,
+      "loss": 0.4014,
+      "step": 12897
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3488977081001397e-07,
+      "loss": 0.3841,
+      "step": 12898
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3438525805668522e-07,
+      "loss": 0.3952,
+      "step": 12899
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3388128127495136e-07,
+      "loss": 0.3992,
+      "step": 12900
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3337784049247469e-07,
+      "loss": 0.3905,
+      "step": 12901
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3287493573688425e-07,
+      "loss": 0.4059,
+      "step": 12902
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3237256703578127e-07,
+      "loss": 0.4306,
+      "step": 12903
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3187073441673813e-07,
+      "loss": 0.4075,
+      "step": 12904
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3136943790729726e-07,
+      "loss": 0.4069,
+      "step": 12905
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.308686775349722e-07,
+      "loss": 0.4113,
+      "step": 12906
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3036845332724545e-07,
+      "loss": 0.4309,
+      "step": 12907
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2986876531157165e-07,
+      "loss": 0.4069,
+      "step": 12908
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2936961351537558e-07,
+      "loss": 0.4296,
+      "step": 12909
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2887099796605196e-07,
+      "loss": 0.4065,
+      "step": 12910
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2837291869096556e-07,
+      "loss": 0.4259,
+      "step": 12911
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2787537571745454e-07,
+      "loss": 0.3752,
+      "step": 12912
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2737836907282374e-07,
+      "loss": 0.4028,
+      "step": 12913
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2688189878435352e-07,
+      "loss": 0.439,
+      "step": 12914
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2638596487929098e-07,
+      "loss": 0.3954,
+      "step": 12915
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2589056738485327e-07,
+      "loss": 0.4123,
+      "step": 12916
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2539570632823195e-07,
+      "loss": 0.4167,
+      "step": 12917
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2490138173658415e-07,
+      "loss": 0.3722,
+      "step": 12918
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2440759363704155e-07,
+      "loss": 0.3932,
+      "step": 12919
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.239143420567036e-07,
+      "loss": 0.3767,
+      "step": 12920
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2342162702264304e-07,
+      "loss": 0.3841,
+      "step": 12921
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2292944856190269e-07,
+      "loss": 0.3862,
+      "step": 12922
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.224378067014943e-07,
+      "loss": 0.3762,
+      "step": 12923
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.219467014684007e-07,
+      "loss": 0.4072,
+      "step": 12924
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.214561328895748e-07,
+      "loss": 0.3703,
+      "step": 12925
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2096610099194172e-07,
+      "loss": 0.3561,
+      "step": 12926
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.204766058023955e-07,
+      "loss": 0.393,
+      "step": 12927
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1998764734780132e-07,
+      "loss": 0.3778,
+      "step": 12928
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1949922565499548e-07,
+      "loss": 0.3865,
+      "step": 12929
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1901134075078323e-07,
+      "loss": 0.377,
+      "step": 12930
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 0.3964,
+      "step": 12931
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1803718141522267e-07,
+      "loss": 0.3919,
+      "step": 12932
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1755090703733826e-07,
+      "loss": 0.4035,
+      "step": 12933
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1706516955497858e-07,
+      "loss": 0.4114,
+      "step": 12934
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.165799689948034e-07,
+      "loss": 0.405,
+      "step": 12935
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1609530538344247e-07,
+      "loss": 0.4062,
+      "step": 12936
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.156111787474935e-07,
+      "loss": 0.4175,
+      "step": 12937
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.151275891135296e-07,
+      "loss": 0.4369,
+      "step": 12938
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.146445365080907e-07,
+      "loss": 0.3914,
+      "step": 12939
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1416202095769002e-07,
+      "loss": 0.3722,
+      "step": 12940
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.136800424888086e-07,
+      "loss": 0.387,
+      "step": 12941
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.131986011278986e-07,
+      "loss": 0.4034,
+      "step": 12942
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1271769690138334e-07,
+      "loss": 0.3871,
+      "step": 12943
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1223732983565727e-07,
+      "loss": 0.4263,
+      "step": 12944
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1175749995708372e-07,
+      "loss": 0.3879,
+      "step": 12945
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1127820729199722e-07,
+      "loss": 0.4437,
+      "step": 12946
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1079945186670558e-07,
+      "loss": 0.3964,
+      "step": 12947
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1032123370748336e-07,
+      "loss": 0.3948,
+      "step": 12948
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0984355284057511e-07,
+      "loss": 0.3986,
+      "step": 12949
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0936640929219987e-07,
+      "loss": 0.4321,
+      "step": 12950
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0888980308854333e-07,
+      "loss": 0.3824,
+      "step": 12951
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.084137342557646e-07,
+      "loss": 0.4065,
+      "step": 12952
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0793820281998945e-07,
+      "loss": 0.4094,
+      "step": 12953
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0746320880732028e-07,
+      "loss": 0.4071,
+      "step": 12954
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0698875224382519e-07,
+      "loss": 0.3658,
+      "step": 12955
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0651483315554333e-07,
+      "loss": 0.3798,
+      "step": 12956
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.060414515684861e-07,
+      "loss": 0.4192,
+      "step": 12957
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0556860750863273e-07,
+      "loss": 0.4042,
+      "step": 12958
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.050963010019369e-07,
+      "loss": 0.3795,
+      "step": 12959
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0462453207431787e-07,
+      "loss": 0.4235,
+      "step": 12960
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.041533007516694e-07,
+      "loss": 0.3862,
+      "step": 12961
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.036826070598552e-07,
+      "loss": 0.3963,
+      "step": 12962
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0321245102470578e-07,
+      "loss": 0.4012,
+      "step": 12963
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0274283267202821e-07,
+      "loss": 0.4103,
+      "step": 12964
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0227375202759636e-07,
+      "loss": 0.3774,
+      "step": 12965
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0180520911715296e-07,
+      "loss": 0.4161,
+      "step": 12966
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.013372039664141e-07,
+      "loss": 0.4258,
+      "step": 12967
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0086973660106701e-07,
+      "loss": 0.3821,
+      "step": 12968
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0040280704676563e-07,
+      "loss": 0.4112,
+      "step": 12969
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.999364153291383e-07,
+      "loss": 0.384,
+      "step": 12970
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9947056147378242e-07,
+      "loss": 0.4527,
+      "step": 12971
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.990052455062652e-07,
+      "loss": 0.4259,
+      "step": 12972
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.985404674521241e-07,
+      "loss": 0.4037,
+      "step": 12973
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.980762273368697e-07,
+      "loss": 0.3841,
+      "step": 12974
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.976125251859795e-07,
+      "loss": 0.4016,
+      "step": 12975
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.971493610249031e-07,
+      "loss": 0.3831,
+      "step": 12976
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.966867348790602e-07,
+      "loss": 0.4062,
+      "step": 12977
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9622464677384378e-07,
+      "loss": 0.4102,
+      "step": 12978
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9576309673461358e-07,
+      "loss": 0.3894,
+      "step": 12979
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9530208478669932e-07,
+      "loss": 0.383,
+      "step": 12980
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.948416109554052e-07,
+      "loss": 0.4064,
+      "step": 12981
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9438167526600327e-07,
+      "loss": 0.4138,
+      "step": 12982
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.939222777437355e-07,
+      "loss": 0.3874,
+      "step": 12983
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9346341841381734e-07,
+      "loss": 0.3892,
+      "step": 12984
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9300509730142857e-07,
+      "loss": 0.407,
+      "step": 12985
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9254731443172802e-07,
+      "loss": 0.414,
+      "step": 12986
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.920900698298389e-07,
+      "loss": 0.3882,
+      "step": 12987
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.916333635208556e-07,
+      "loss": 0.4063,
+      "step": 12988
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.911771955298425e-07,
+      "loss": 0.4103,
+      "step": 12989
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9072156588183842e-07,
+      "loss": 0.3798,
+      "step": 12990
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.902664746018501e-07,
+      "loss": 0.3861,
+      "step": 12991
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8981192171485196e-07,
+      "loss": 0.3831,
+      "step": 12992
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8935790724579295e-07,
+      "loss": 0.3901,
+      "step": 12993
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.88904431219592e-07,
+      "loss": 0.3994,
+      "step": 12994
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.884514936611359e-07,
+      "loss": 0.3565,
+      "step": 12995
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8799909459528365e-07,
+      "loss": 0.3763,
+      "step": 12996
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8754723404686425e-07,
+      "loss": 0.383,
+      "step": 12997
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8709591204068013e-07,
+      "loss": 0.4204,
+      "step": 12998
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8664512860149808e-07,
+      "loss": 0.4051,
+      "step": 12999
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8619488375405947e-07,
+      "loss": 0.3711,
+      "step": 13000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8574517752307676e-07,
+      "loss": 0.4132,
+      "step": 13001
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8529600993322906e-07,
+      "loss": 0.4247,
+      "step": 13002
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.848473810091711e-07,
+      "loss": 0.3863,
+      "step": 13003
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.843992907755243e-07,
+      "loss": 0.3827,
+      "step": 13004
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8395173925687903e-07,
+      "loss": 0.4199,
+      "step": 13005
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8350472647780116e-07,
+      "loss": 0.4246,
+      "step": 13006
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8305825246282437e-07,
+      "loss": 0.4215,
+      "step": 13007
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8261231723645246e-07,
+      "loss": 0.4217,
+      "step": 13008
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8216692082315802e-07,
+      "loss": 0.4509,
+      "step": 13009
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8172206324738818e-07,
+      "loss": 0.3937,
+      "step": 13010
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8127774453355674e-07,
+      "loss": 0.402,
+      "step": 13011
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8083396470605195e-07,
+      "loss": 0.3966,
+      "step": 13012
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8039072378922662e-07,
+      "loss": 0.3892,
+      "step": 13013
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.799480218074101e-07,
+      "loss": 0.3563,
+      "step": 13014
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7950585878489856e-07,
+      "loss": 0.3651,
+      "step": 13015
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7906423474595925e-07,
+      "loss": 0.429,
+      "step": 13016
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7862314971483052e-07,
+      "loss": 0.4035,
+      "step": 13017
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.781826037157186e-07,
+      "loss": 0.4155,
+      "step": 13018
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7774259677280636e-07,
+      "loss": 0.3864,
+      "step": 13019
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7730312891023894e-07,
+      "loss": 0.4174,
+      "step": 13020
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7686420015213924e-07,
+      "loss": 0.422,
+      "step": 13021
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7642581052259357e-07,
+      "loss": 0.4223,
+      "step": 13022
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7598796004566488e-07,
+      "loss": 0.3873,
+      "step": 13023
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7555064874538397e-07,
+      "loss": 0.417,
+      "step": 13024
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7511387664575162e-07,
+      "loss": 0.403,
+      "step": 13025
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.746776437707387e-07,
+      "loss": 0.4027,
+      "step": 13026
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7424195014428712e-07,
+      "loss": 0.362,
+      "step": 13027
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7380679579031113e-07,
+      "loss": 0.3831,
+      "step": 13028
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7337218073269156e-07,
+      "loss": 0.4173,
+      "step": 13029
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7293810499528273e-07,
+      "loss": 0.4117,
+      "step": 13030
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.725045686019078e-07,
+      "loss": 0.3999,
+      "step": 13031
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7207157157636101e-07,
+      "loss": 0.3909,
+      "step": 13032
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7163911394240675e-07,
+      "loss": 0.4085,
+      "step": 13033
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.712071957237793e-07,
+      "loss": 0.3979,
+      "step": 13034
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7077581694418423e-07,
+      "loss": 0.4121,
+      "step": 13035
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7034497762729807e-07,
+      "loss": 0.4307,
+      "step": 13036
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.699146777967653e-07,
+      "loss": 0.3763,
+      "step": 13037
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6948491747620365e-07,
+      "loss": 0.4239,
+      "step": 13038
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6905569668919763e-07,
+      "loss": 0.3924,
+      "step": 13039
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6862701545930726e-07,
+      "loss": 0.3989,
+      "step": 13040
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6819887381005927e-07,
+      "loss": 0.4093,
+      "step": 13041
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6777127176495046e-07,
+      "loss": 0.4108,
+      "step": 13042
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6734420934744977e-07,
+      "loss": 0.3668,
+      "step": 13043
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6691768658099627e-07,
+      "loss": 0.4133,
+      "step": 13044
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.664917034889979e-07,
+      "loss": 0.4008,
+      "step": 13045
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6606626009483594e-07,
+      "loss": 0.4135,
+      "step": 13046
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6564135642185953e-07,
+      "loss": 0.4379,
+      "step": 13047
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6521699249338662e-07,
+      "loss": 0.4191,
+      "step": 13048
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6479316833271196e-07,
+      "loss": 0.4076,
+      "step": 13049
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.643698839630925e-07,
+      "loss": 0.3911,
+      "step": 13050
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6394713940776296e-07,
+      "loss": 0.4074,
+      "step": 13051
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6352493468992258e-07,
+      "loss": 0.4068,
+      "step": 13052
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6310326983274506e-07,
+      "loss": 0.39,
+      "step": 13053
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6268214485937294e-07,
+      "loss": 0.4135,
+      "step": 13054
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.622615597929178e-07,
+      "loss": 0.3884,
+      "step": 13055
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6184151465646225e-07,
+      "loss": 0.4046,
+      "step": 13056
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6142200947306119e-07,
+      "loss": 0.4066,
+      "step": 13057
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6100304426573843e-07,
+      "loss": 0.4163,
+      "step": 13058
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.605846190574889e-07,
+      "loss": 0.3921,
+      "step": 13059
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6016673387127645e-07,
+      "loss": 0.3893,
+      "step": 13060
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5974938873003609e-07,
+      "loss": 0.4154,
+      "step": 13061
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5933258365667392e-07,
+      "loss": 0.4247,
+      "step": 13062
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.58916318674065e-07,
+      "loss": 0.3848,
+      "step": 13063
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5850059380505434e-07,
+      "loss": 0.3437,
+      "step": 13064
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.580854090724604e-07,
+      "loss": 0.4355,
+      "step": 13065
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5767076449906827e-07,
+      "loss": 0.385,
+      "step": 13066
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.572566601076364e-07,
+      "loss": 0.3718,
+      "step": 13067
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.568430959208922e-07,
+      "loss": 0.4171,
+      "step": 13068
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5643007196153303e-07,
+      "loss": 0.4158,
+      "step": 13069
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5601758825222746e-07,
+      "loss": 0.4109,
+      "step": 13070
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5560564481561512e-07,
+      "loss": 0.3915,
+      "step": 13071
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5519424167430242e-07,
+      "loss": 0.3938,
+      "step": 13072
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5478337885086904e-07,
+      "loss": 0.405,
+      "step": 13073
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5437305636786693e-07,
+      "loss": 0.4164,
+      "step": 13074
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5396327424781365e-07,
+      "loss": 0.4238,
+      "step": 13075
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5355403251320123e-07,
+      "loss": 0.377,
+      "step": 13076
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5314533118648944e-07,
+      "loss": 0.4331,
+      "step": 13077
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5273717029010927e-07,
+      "loss": 0.3823,
+      "step": 13078
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5232954984646276e-07,
+      "loss": 0.3832,
+      "step": 13079
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.405,
+      "step": 13080
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5151593040682365e-07,
+      "loss": 0.4022,
+      "step": 13081
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5110993145548646e-07,
+      "loss": 0.4451,
+      "step": 13082
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.507044730461915e-07,
+      "loss": 0.4154,
+      "step": 13083
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.50299555201191e-07,
+      "loss": 0.3658,
+      "step": 13084
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4989517794270824e-07,
+      "loss": 0.3577,
+      "step": 13085
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4949134129293773e-07,
+      "loss": 0.4217,
+      "step": 13086
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4908804527404287e-07,
+      "loss": 0.3793,
+      "step": 13087
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4868528990815702e-07,
+      "loss": 0.3813,
+      "step": 13088
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4828307521738694e-07,
+      "loss": 0.3966,
+      "step": 13089
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.47881401223805e-07,
+      "loss": 0.3712,
+      "step": 13090
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4748026794945803e-07,
+      "loss": 0.3815,
+      "step": 13091
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.470796754163628e-07,
+      "loss": 0.3966,
+      "step": 13092
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.466796236465029e-07,
+      "loss": 0.4247,
+      "step": 13093
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4628011266183628e-07,
+      "loss": 0.4122,
+      "step": 13094
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4588114248428875e-07,
+      "loss": 0.4043,
+      "step": 13095
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4548271313575834e-07,
+      "loss": 0.4102,
+      "step": 13096
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.450848246381098e-07,
+      "loss": 0.423,
+      "step": 13097
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.446874770131823e-07,
+      "loss": 0.393,
+      "step": 13098
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4429067028278399e-07,
+      "loss": 0.4136,
+      "step": 13099
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4389440446869185e-07,
+      "loss": 0.4134,
+      "step": 13100
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4349867959265628e-07,
+      "loss": 0.4271,
+      "step": 13101
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4310349567639438e-07,
+      "loss": 0.3974,
+      "step": 13102
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.427088527415965e-07,
+      "loss": 0.3914,
+      "step": 13103
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4231475080992097e-07,
+      "loss": 0.4115,
+      "step": 13104
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.419211899029971e-07,
+      "loss": 0.3844,
+      "step": 13105
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4152817004242648e-07,
+      "loss": 0.4006,
+      "step": 13106
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4113569124977743e-07,
+      "loss": 0.4109,
+      "step": 13107
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4074375354659164e-07,
+      "loss": 0.4206,
+      "step": 13108
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4035235695438077e-07,
+      "loss": 0.3928,
+      "step": 13109
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3996150149462651e-07,
+      "loss": 0.4295,
+      "step": 13110
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3957118718877838e-07,
+      "loss": 0.3836,
+      "step": 13111
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3918141405825924e-07,
+      "loss": 0.4592,
+      "step": 13112
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.387921821244609e-07,
+      "loss": 0.4038,
+      "step": 13113
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.384034914087462e-07,
+      "loss": 0.3821,
+      "step": 13114
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3801534193244704e-07,
+      "loss": 0.444,
+      "step": 13115
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3762773371686855e-07,
+      "loss": 0.4039,
+      "step": 13116
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3724066678328153e-07,
+      "loss": 0.3853,
+      "step": 13117
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3685414115293005e-07,
+      "loss": 0.4418,
+      "step": 13118
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.364681568470294e-07,
+      "loss": 0.4195,
+      "step": 13119
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3608271388676375e-07,
+      "loss": 0.3641,
+      "step": 13120
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3569781229328615e-07,
+      "loss": 0.3807,
+      "step": 13121
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3531345208772195e-07,
+      "loss": 0.3989,
+      "step": 13122
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3492963329116538e-07,
+      "loss": 0.4028,
+      "step": 13123
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3454635592468401e-07,
+      "loss": 0.3999,
+      "step": 13124
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3416362000931104e-07,
+      "loss": 0.3515,
+      "step": 13125
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3378142556605412e-07,
+      "loss": 0.4154,
+      "step": 13126
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3339977261588865e-07,
+      "loss": 0.4019,
+      "step": 13127
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3301866117976125e-07,
+      "loss": 0.4264,
+      "step": 13128
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.326380912785885e-07,
+      "loss": 0.4183,
+      "step": 13129
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3225806293325815e-07,
+      "loss": 0.3961,
+      "step": 13130
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3187857616462573e-07,
+      "loss": 0.4079,
+      "step": 13131
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3149963099352014e-07,
+      "loss": 0.4079,
+      "step": 13132
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.311212274407392e-07,
+      "loss": 0.4194,
+      "step": 13133
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.307433655270518e-07,
+      "loss": 0.4318,
+      "step": 13134
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3036604527319475e-07,
+      "loss": 0.4218,
+      "step": 13135
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.299892666998781e-07,
+      "loss": 0.431,
+      "step": 13136
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.296130298277798e-07,
+      "loss": 0.415,
+      "step": 13137
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2923733467754994e-07,
+      "loss": 0.3804,
+      "step": 13138
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2886218126980654e-07,
+      "loss": 0.4363,
+      "step": 13139
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2848756962514086e-07,
+      "loss": 0.4141,
+      "step": 13140
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2811349976411204e-07,
+      "loss": 0.3579,
+      "step": 13141
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2773997170725028e-07,
+      "loss": 0.3559,
+      "step": 13142
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.27366985475057e-07,
+      "loss": 0.4236,
+      "step": 13143
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2699454108800247e-07,
+      "loss": 0.3669,
+      "step": 13144
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2662263856652813e-07,
+      "loss": 0.3745,
+      "step": 13145
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2625127793104542e-07,
+      "loss": 0.3856,
+      "step": 13146
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.258804592019347e-07,
+      "loss": 0.3661,
+      "step": 13147
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.255101823995475e-07,
+      "loss": 0.3777,
+      "step": 13148
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.251404475442086e-07,
+      "loss": 0.3959,
+      "step": 13149
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2477125465620854e-07,
+      "loss": 0.3742,
+      "step": 13150
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2440260375580993e-07,
+      "loss": 0.43,
+      "step": 13151
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.240344948632466e-07,
+      "loss": 0.4038,
+      "step": 13152
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2366692799872015e-07,
+      "loss": 0.3988,
+      "step": 13153
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.232999031824056e-07,
+      "loss": 0.4195,
+      "step": 13154
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2293342043444568e-07,
+      "loss": 0.3636,
+      "step": 13155
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.225674797749532e-07,
+      "loss": 0.3894,
+      "step": 13156
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2220208122401433e-07,
+      "loss": 0.4111,
+      "step": 13157
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2183722480168304e-07,
+      "loss": 0.3783,
+      "step": 13158
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2147291052798217e-07,
+      "loss": 0.4185,
+      "step": 13159
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2110913842290793e-07,
+      "loss": 0.4026,
+      "step": 13160
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2074590850642665e-07,
+      "loss": 0.4075,
+      "step": 13161
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.203832207984712e-07,
+      "loss": 0.3685,
+      "step": 13162
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2002107531894903e-07,
+      "loss": 0.4251,
+      "step": 13163
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.196594720877342e-07,
+      "loss": 0.3784,
+      "step": 13164
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1929841112467533e-07,
+      "loss": 0.3975,
+      "step": 13165
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1893789244958653e-07,
+      "loss": 0.4123,
+      "step": 13166
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1857791608225533e-07,
+      "loss": 0.4014,
+      "step": 13167
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1821848204243813e-07,
+      "loss": 0.4226,
+      "step": 13168
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1785959034986138e-07,
+      "loss": 0.3859,
+      "step": 13169
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1750124102422378e-07,
+      "loss": 0.3903,
+      "step": 13170
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1714343408519291e-07,
+      "loss": 0.4178,
+      "step": 13171
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1678616955240419e-07,
+      "loss": 0.4014,
+      "step": 13172
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1642944744546747e-07,
+      "loss": 0.3887,
+      "step": 13173
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1607326778396044e-07,
+      "loss": 0.4079,
+      "step": 13174
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.157176305874319e-07,
+      "loss": 0.3573,
+      "step": 13175
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1536253587539847e-07,
+      "loss": 0.4106,
+      "step": 13176
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1500798366735234e-07,
+      "loss": 0.4058,
+      "step": 13177
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1465397398275013e-07,
+      "loss": 0.3878,
+      "step": 13178
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.14300506841023e-07,
+      "loss": 0.3885,
+      "step": 13179
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1394758226156765e-07,
+      "loss": 0.4199,
+      "step": 13180
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1359520026375747e-07,
+      "loss": 0.4093,
+      "step": 13181
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1324336086692922e-07,
+      "loss": 0.3682,
+      "step": 13182
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1289206409039522e-07,
+      "loss": 0.3951,
+      "step": 13183
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1254130995343337e-07,
+      "loss": 0.3979,
+      "step": 13184
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.121910984752983e-07,
+      "loss": 0.4151,
+      "step": 13185
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1184142967520794e-07,
+      "loss": 0.3877,
+      "step": 13186
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1149230357235363e-07,
+      "loss": 0.3619,
+      "step": 13187
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1114372018589559e-07,
+      "loss": 0.4276,
+      "step": 13188
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1079567953496851e-07,
+      "loss": 0.3927,
+      "step": 13189
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1044818163867265e-07,
+      "loss": 0.4129,
+      "step": 13190
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1010122651607836e-07,
+      "loss": 0.3856,
+      "step": 13191
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0975481418622924e-07,
+      "loss": 0.403,
+      "step": 13192
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.094089446681379e-07,
+      "loss": 0.4073,
+      "step": 13193
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.090636179807869e-07,
+      "loss": 0.4287,
+      "step": 13194
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0871883414312778e-07,
+      "loss": 0.3835,
+      "step": 13195
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0837459317408316e-07,
+      "loss": 0.3488,
+      "step": 13196
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0803089509254794e-07,
+      "loss": 0.3778,
+      "step": 13197
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.076877399173859e-07,
+      "loss": 0.3928,
+      "step": 13198
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0734512766742866e-07,
+      "loss": 0.3803,
+      "step": 13199
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0700305836148118e-07,
+      "loss": 0.3956,
+      "step": 13200
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0666153201831619e-07,
+      "loss": 0.3606,
+      "step": 13201
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0632054865667873e-07,
+      "loss": 0.3916,
+      "step": 13202
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0598010829528493e-07,
+      "loss": 0.4054,
+      "step": 13203
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0564021095281652e-07,
+      "loss": 0.3917,
+      "step": 13204
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0530085664792855e-07,
+      "loss": 0.4082,
+      "step": 13205
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0496204539924837e-07,
+      "loss": 0.4355,
+      "step": 13206
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0462377722536887e-07,
+      "loss": 0.3803,
+      "step": 13207
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0428605214485632e-07,
+      "loss": 0.3866,
+      "step": 13208
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0394887017624588e-07,
+      "loss": 0.3964,
+      "step": 13209
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0361223133804388e-07,
+      "loss": 0.4024,
+      "step": 13210
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0327613564872552e-07,
+      "loss": 0.394,
+      "step": 13211
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0294058312673716e-07,
+      "loss": 0.4339,
+      "step": 13212
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0260557379049518e-07,
+      "loss": 0.4056,
+      "step": 13213
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.022711076583871e-07,
+      "loss": 0.3956,
+      "step": 13214
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0193718474876823e-07,
+      "loss": 0.4078,
+      "step": 13215
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0160380507996614e-07,
+      "loss": 0.4245,
+      "step": 13216
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.012709686702773e-07,
+      "loss": 0.4114,
+      "step": 13217
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0093867553796821e-07,
+      "loss": 0.4147,
+      "step": 13218
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0060692570127872e-07,
+      "loss": 0.4295,
+      "step": 13219
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0027571917841538e-07,
+      "loss": 0.3919,
+      "step": 13220
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.994505598755477e-08,
+      "loss": 0.4059,
+      "step": 13221
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.96149361468457e-08,
+      "loss": 0.4205,
+      "step": 13222
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.9285359674407e-08,
+      "loss": 0.396,
+      "step": 13223
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.895632658832643e-08,
+      "loss": 0.3967,
+      "step": 13224
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.862783690666178e-08,
+      "loss": 0.4085,
+      "step": 13225
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.829989064744305e-08,
+      "loss": 0.4379,
+      "step": 13226
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.797248782866809e-08,
+      "loss": 0.4418,
+      "step": 13227
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.764562846830694e-08,
+      "loss": 0.4193,
+      "step": 13228
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.731931258429638e-08,
+      "loss": 0.4321,
+      "step": 13229
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.699354019454876e-08,
+      "loss": 0.3889,
+      "step": 13230
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.66683113169431e-08,
+      "loss": 0.374,
+      "step": 13231
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.634362596932733e-08,
+      "loss": 0.4391,
+      "step": 13232
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.60194841695239e-08,
+      "loss": 0.3991,
+      "step": 13233
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.569588593531965e-08,
+      "loss": 0.4104,
+      "step": 13234
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.537283128447705e-08,
+      "loss": 0.4062,
+      "step": 13235
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.505032023472637e-08,
+      "loss": 0.4082,
+      "step": 13236
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.47283528037668e-08,
+      "loss": 0.4021,
+      "step": 13237
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.440692900927195e-08,
+      "loss": 0.39,
+      "step": 13238
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.408604886887884e-08,
+      "loss": 0.3719,
+      "step": 13239
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.376571240020227e-08,
+      "loss": 0.4141,
+      "step": 13240
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.34459196208204e-08,
+      "loss": 0.4207,
+      "step": 13241
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.312667054828583e-08,
+      "loss": 0.3827,
+      "step": 13242
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.280796520012015e-08,
+      "loss": 0.3593,
+      "step": 13243
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.248980359381487e-08,
+      "loss": 0.3989,
+      "step": 13244
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.217218574683162e-08,
+      "loss": 0.4572,
+      "step": 13245
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.185511167660422e-08,
+      "loss": 0.4051,
+      "step": 13246
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.153858140053206e-08,
+      "loss": 0.4205,
+      "step": 13247
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.122259493598907e-08,
+      "loss": 0.3863,
+      "step": 13248
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.090715230031688e-08,
+      "loss": 0.3816,
+      "step": 13249
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.059225351082945e-08,
+      "loss": 0.4133,
+      "step": 13250
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.02778985848074e-08,
+      "loss": 0.4172,
+      "step": 13251
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.996408753950581e-08,
+      "loss": 0.4113,
+      "step": 13252
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.965082039214535e-08,
+      "loss": 0.4097,
+      "step": 13253
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.933809715992225e-08,
+      "loss": 0.448,
+      "step": 13254
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.902591785999726e-08,
+      "loss": 0.4192,
+      "step": 13255
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.871428250950665e-08,
+      "loss": 0.3961,
+      "step": 13256
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.840319112555119e-08,
+      "loss": 0.3783,
+      "step": 13257
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.80926437252061e-08,
+      "loss": 0.4312,
+      "step": 13258
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.778264032551442e-08,
+      "loss": 0.3969,
+      "step": 13259
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.747318094349144e-08,
+      "loss": 0.3773,
+      "step": 13260
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.716426559612024e-08,
+      "loss": 0.419,
+      "step": 13261
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.685589430035613e-08,
+      "loss": 0.4171,
+      "step": 13262
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.654806707312336e-08,
+      "loss": 0.4089,
+      "step": 13263
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.62407839313173e-08,
+      "loss": 0.3975,
+      "step": 13264
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.593404489180112e-08,
+      "loss": 0.4131,
+      "step": 13265
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.562784997141027e-08,
+      "loss": 0.4269,
+      "step": 13266
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.532219918695128e-08,
+      "loss": 0.3959,
+      "step": 13267
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.50170925551963e-08,
+      "loss": 0.3781,
+      "step": 13268
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.471253009289416e-08,
+      "loss": 0.3947,
+      "step": 13269
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.440851181675813e-08,
+      "loss": 0.3871,
+      "step": 13270
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.410503774347379e-08,
+      "loss": 0.4148,
+      "step": 13271
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.380210788969889e-08,
+      "loss": 0.4395,
+      "step": 13272
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.349972227205683e-08,
+      "loss": 0.4117,
+      "step": 13273
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.319788090714542e-08,
+      "loss": 0.3715,
+      "step": 13274
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.289658381153032e-08,
+      "loss": 0.443,
+      "step": 13275
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.259583100174607e-08,
+      "loss": 0.3919,
+      "step": 13276
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.229562249430056e-08,
+      "loss": 0.3581,
+      "step": 13277
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.199595830567064e-08,
+      "loss": 0.4306,
+      "step": 13278
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.169683845230314e-08,
+      "loss": 0.3711,
+      "step": 13279
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.139826295061381e-08,
+      "loss": 0.3895,
+      "step": 13280
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.110023181699067e-08,
+      "loss": 0.4073,
+      "step": 13281
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.08027450677884e-08,
+      "loss": 0.4114,
+      "step": 13282
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.050580271933728e-08,
+      "loss": 0.3781,
+      "step": 13283
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.020940478793204e-08,
+      "loss": 0.444,
+      "step": 13284
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.99135512898408e-08,
+      "loss": 0.4021,
+      "step": 13285
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.961824224130166e-08,
+      "loss": 0.3821,
+      "step": 13286
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.932347765852166e-08,
+      "loss": 0.4209,
+      "step": 13287
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.902925755767898e-08,
+      "loss": 0.3718,
+      "step": 13288
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.87355819549207e-08,
+      "loss": 0.3793,
+      "step": 13289
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.844245086636614e-08,
+      "loss": 0.4007,
+      "step": 13290
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.814986430810134e-08,
+      "loss": 0.3966,
+      "step": 13291
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.785782229618677e-08,
+      "loss": 0.3769,
+      "step": 13292
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.75663248466496e-08,
+      "loss": 0.377,
+      "step": 13293
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.727537197548707e-08,
+      "loss": 0.3873,
+      "step": 13294
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.69849636986686e-08,
+      "loss": 0.4485,
+      "step": 13295
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.669510003213366e-08,
+      "loss": 0.4359,
+      "step": 13296
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.640578099179175e-08,
+      "loss": 0.3779,
+      "step": 13297
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.611700659351906e-08,
+      "loss": 0.4142,
+      "step": 13298
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.582877685316625e-08,
+      "loss": 0.3566,
+      "step": 13299
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.554109178655178e-08,
+      "loss": 0.3729,
+      "step": 13300
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.525395140946523e-08,
+      "loss": 0.428,
+      "step": 13301
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.496735573766623e-08,
+      "loss": 0.4248,
+      "step": 13302
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.468130478688218e-08,
+      "loss": 0.3415,
+      "step": 13303
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.43957985728161e-08,
+      "loss": 0.4313,
+      "step": 13304
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.411083711113432e-08,
+      "loss": 0.3757,
+      "step": 13305
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.382642041747767e-08,
+      "loss": 0.3869,
+      "step": 13306
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.354254850745702e-08,
+      "loss": 0.4246,
+      "step": 13307
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.325922139665099e-08,
+      "loss": 0.4566,
+      "step": 13308
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.297643910060936e-08,
+      "loss": 0.3907,
+      "step": 13309
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.269420163485308e-08,
+      "loss": 0.3933,
+      "step": 13310
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.241250901487196e-08,
+      "loss": 0.3884,
+      "step": 13311
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.213136125612585e-08,
+      "loss": 0.3995,
+      "step": 13312
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.185075837404576e-08,
+      "loss": 0.4183,
+      "step": 13313
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.157070038403269e-08,
+      "loss": 0.3905,
+      "step": 13314
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.129118730145657e-08,
+      "loss": 0.3797,
+      "step": 13315
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.101221914165845e-08,
+      "loss": 0.4179,
+      "step": 13316
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.073379591994834e-08,
+      "loss": 0.4051,
+      "step": 13317
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.04559176516062e-08,
+      "loss": 0.4205,
+      "step": 13318
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.017858435188541e-08,
+      "loss": 0.3602,
+      "step": 13319
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.990179603600488e-08,
+      "loss": 0.4077,
+      "step": 13320
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.962555271915806e-08,
+      "loss": 0.4109,
+      "step": 13321
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.934985441650388e-08,
+      "loss": 0.3581,
+      "step": 13322
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.907470114317582e-08,
+      "loss": 0.3839,
+      "step": 13323
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.880009291427292e-08,
+      "loss": 0.4433,
+      "step": 13324
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.852602974486866e-08,
+      "loss": 0.3893,
+      "step": 13325
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.825251165000213e-08,
+      "loss": 0.3972,
+      "step": 13326
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.797953864468687e-08,
+      "loss": 0.3917,
+      "step": 13327
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.770711074390535e-08,
+      "loss": 0.3806,
+      "step": 13328
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.743522796260782e-08,
+      "loss": 0.386,
+      "step": 13329
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.716389031571568e-08,
+      "loss": 0.4163,
+      "step": 13330
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.68930978181237e-08,
+      "loss": 0.4195,
+      "step": 13331
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.662285048469108e-08,
+      "loss": 0.4032,
+      "step": 13332
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.635314833025041e-08,
+      "loss": 0.4291,
+      "step": 13333
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.608399136960541e-08,
+      "loss": 0.404,
+      "step": 13334
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.581537961752648e-08,
+      "loss": 0.4384,
+      "step": 13335
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.55473130887585e-08,
+      "loss": 0.3954,
+      "step": 13336
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.52797917980108e-08,
+      "loss": 0.4074,
+      "step": 13337
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.501281575996832e-08,
+      "loss": 0.3711,
+      "step": 13338
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.474638498928265e-08,
+      "loss": 0.4356,
+      "step": 13339
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.448049950057656e-08,
+      "loss": 0.4205,
+      "step": 13340
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.42151593084428e-08,
+      "loss": 0.4396,
+      "step": 13341
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.395036442744418e-08,
+      "loss": 0.4206,
+      "step": 13342
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.36861148721124e-08,
+      "loss": 0.3933,
+      "step": 13343
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.342241065695365e-08,
+      "loss": 0.4344,
+      "step": 13344
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.315925179643745e-08,
+      "loss": 0.412,
+      "step": 13345
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.289663830500892e-08,
+      "loss": 0.3944,
+      "step": 13346
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.263457019708096e-08,
+      "loss": 0.4006,
+      "step": 13347
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.237304748703543e-08,
+      "loss": 0.3984,
+      "step": 13348
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.211207018922749e-08,
+      "loss": 0.4039,
+      "step": 13349
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.185163831798014e-08,
+      "loss": 0.4252,
+      "step": 13350
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.159175188758526e-08,
+      "loss": 0.4105,
+      "step": 13351
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.133241091230813e-08,
+      "loss": 0.3917,
+      "step": 13352
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.107361540638179e-08,
+      "loss": 0.3637,
+      "step": 13353
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.081536538400934e-08,
+      "loss": 0.3952,
+      "step": 13354
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.055766085936498e-08,
+      "loss": 0.4114,
+      "step": 13355
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.030050184659186e-08,
+      "loss": 0.4213,
+      "step": 13356
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.004388835980424e-08,
+      "loss": 0.3778,
+      "step": 13357
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.978782041308639e-08,
+      "loss": 0.4173,
+      "step": 13358
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.953229802049154e-08,
+      "loss": 0.3977,
+      "step": 13359
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.927732119604401e-08,
+      "loss": 0.4231,
+      "step": 13360
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.902288995373706e-08,
+      "loss": 0.4099,
+      "step": 13361
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.876900430753618e-08,
+      "loss": 0.424,
+      "step": 13362
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.851566427137467e-08,
+      "loss": 0.3869,
+      "step": 13363
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.8262869859156965e-08,
+      "loss": 0.3733,
+      "step": 13364
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.801062108475641e-08,
+      "loss": 0.4051,
+      "step": 13365
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.77589179620186e-08,
+      "loss": 0.411,
+      "step": 13366
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.750776050475693e-08,
+      "loss": 0.433,
+      "step": 13367
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.725714872675703e-08,
+      "loss": 0.3835,
+      "step": 13368
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.700708264177235e-08,
+      "loss": 0.3925,
+      "step": 13369
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.6757562263527465e-08,
+      "loss": 0.407,
+      "step": 13370
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.650858760571698e-08,
+      "loss": 0.3979,
+      "step": 13371
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.626015868200663e-08,
+      "loss": 0.3904,
+      "step": 13372
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.6012275506028835e-08,
+      "loss": 0.4127,
+      "step": 13373
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.57649380913905e-08,
+      "loss": 0.4187,
+      "step": 13374
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.55181464516652e-08,
+      "loss": 0.4273,
+      "step": 13375
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.527190060039767e-08,
+      "loss": 0.4064,
+      "step": 13376
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.5026200551103746e-08,
+      "loss": 0.4144,
+      "step": 13377
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4781046317267103e-08,
+      "loss": 0.4135,
+      "step": 13378
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.453643791234253e-08,
+      "loss": 0.4055,
+      "step": 13379
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.429237534975707e-08,
+      "loss": 0.4117,
+      "step": 13380
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4048858642903325e-08,
+      "loss": 0.4024,
+      "step": 13381
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.380588780514839e-08,
+      "loss": 0.4141,
+      "step": 13382
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.356346284982494e-08,
+      "loss": 0.4169,
+      "step": 13383
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.3321583790241217e-08,
+      "loss": 0.3996,
+      "step": 13384
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.308025063966993e-08,
+      "loss": 0.3619,
+      "step": 13385
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.283946341135715e-08,
+      "loss": 0.4208,
+      "step": 13386
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.259922211851787e-08,
+      "loss": 0.4207,
+      "step": 13387
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.235952677433709e-08,
+      "loss": 0.4208,
+      "step": 13388
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.2120377391972064e-08,
+      "loss": 0.3886,
+      "step": 13389
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.1881773984545634e-08,
+      "loss": 0.4068,
+      "step": 13390
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.164371656515399e-08,
+      "loss": 0.3624,
+      "step": 13391
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.1406205146864455e-08,
+      "loss": 0.3903,
+      "step": 13392
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.116923974270993e-08,
+      "loss": 0.4076,
+      "step": 13393
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.093282036569669e-08,
+      "loss": 0.3885,
+      "step": 13394
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.069694702880101e-08,
+      "loss": 0.4553,
+      "step": 13395
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.0461619744968104e-08,
+      "loss": 0.4408,
+      "step": 13396
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.022683852711319e-08,
+      "loss": 0.382,
+      "step": 13397
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.999260338812151e-08,
+      "loss": 0.3836,
+      "step": 13398
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.9758914340849454e-08,
+      "loss": 0.3557,
+      "step": 13399
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.9525771398123424e-08,
+      "loss": 0.4399,
+      "step": 13400
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.929317457273763e-08,
+      "loss": 0.4184,
+      "step": 13401
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.906112387745965e-08,
+      "loss": 0.4244,
+      "step": 13402
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.882961932502262e-08,
+      "loss": 0.3953,
+      "step": 13403
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.859866092813414e-08,
+      "loss": 0.4273,
+      "step": 13404
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.8368248699469656e-08,
+      "loss": 0.3715,
+      "step": 13405
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.81383826516757e-08,
+      "loss": 0.4196,
+      "step": 13406
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7909062797366623e-08,
+      "loss": 0.3703,
+      "step": 13407
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7680289149129035e-08,
+      "loss": 0.3913,
+      "step": 13408
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.745206171951955e-08,
+      "loss": 0.4155,
+      "step": 13409
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.722438052106371e-08,
+      "loss": 0.4063,
+      "step": 13410
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.6997245566257066e-08,
+      "loss": 0.3968,
+      "step": 13411
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.677065686756632e-08,
+      "loss": 0.4274,
+      "step": 13412
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.654461443742597e-08,
+      "loss": 0.4138,
+      "step": 13413
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.631911828824276e-08,
+      "loss": 0.3976,
+      "step": 13414
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.609416843239345e-08,
+      "loss": 0.4354,
+      "step": 13415
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.586976488222372e-08,
+      "loss": 0.3671,
+      "step": 13416
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.564590765004928e-08,
+      "loss": 0.3774,
+      "step": 13417
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.5422596748156964e-08,
+      "loss": 0.4146,
+      "step": 13418
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.519983218880142e-08,
+      "loss": 0.4103,
+      "step": 13419
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4977613984210634e-08,
+      "loss": 0.4165,
+      "step": 13420
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.47559421465793e-08,
+      "loss": 0.408,
+      "step": 13421
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.453481668807325e-08,
+      "loss": 0.447,
+      "step": 13422
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.431423762083054e-08,
+      "loss": 0.391,
+      "step": 13423
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.409420495695593e-08,
+      "loss": 0.4308,
+      "step": 13424
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.387471870852533e-08,
+      "loss": 0.4129,
+      "step": 13425
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.365577888758577e-08,
+      "loss": 0.4232,
+      "step": 13426
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3437385506153174e-08,
+      "loss": 0.4533,
+      "step": 13427
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3219538576214634e-08,
+      "loss": 0.3834,
+      "step": 13428
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3002238109723927e-08,
+      "loss": 0.396,
+      "step": 13429
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.278548411860928e-08,
+      "loss": 0.4035,
+      "step": 13430
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.256927661476673e-08,
+      "loss": 0.3708,
+      "step": 13431
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2353615610061236e-08,
+      "loss": 0.3927,
+      "step": 13432
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.213850111633111e-08,
+      "loss": 0.4119,
+      "step": 13433
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.192393314538135e-08,
+      "loss": 0.3858,
+      "step": 13434
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.170991170898808e-08,
+      "loss": 0.394,
+      "step": 13435
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1496436818898586e-08,
+      "loss": 0.3605,
+      "step": 13436
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.128350848682683e-08,
+      "loss": 0.3894,
+      "step": 13437
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.107112672446123e-08,
+      "loss": 0.394,
+      "step": 13438
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0859291543458025e-08,
+      "loss": 0.3867,
+      "step": 13439
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0648002955442353e-08,
+      "loss": 0.4007,
+      "step": 13440
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0437260972011617e-08,
+      "loss": 0.4248,
+      "step": 13441
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0227065604731e-08,
+      "loss": 0.3943,
+      "step": 13442
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.001741686513683e-08,
+      "loss": 0.3785,
+      "step": 13443
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.9808314764736565e-08,
+      "loss": 0.4339,
+      "step": 13444
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.959975931500548e-08,
+      "loss": 0.4238,
+      "step": 13445
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.939175052738997e-08,
+      "loss": 0.4058,
+      "step": 13446
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.918428841330646e-08,
+      "loss": 0.4156,
+      "step": 13447
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.897737298414028e-08,
+      "loss": 0.3853,
+      "step": 13448
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.8771004251250134e-08,
+      "loss": 0.43,
+      "step": 13449
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.8565182225959174e-08,
+      "loss": 0.4194,
+      "step": 13450
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.835990691956615e-08,
+      "loss": 0.3886,
+      "step": 13451
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.8155178343336484e-08,
+      "loss": 0.3696,
+      "step": 13452
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.795099650850565e-08,
+      "loss": 0.3672,
+      "step": 13453
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7747361426280216e-08,
+      "loss": 0.4164,
+      "step": 13454
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7544273107836816e-08,
+      "loss": 0.4195,
+      "step": 13455
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7341731564322084e-08,
+      "loss": 0.3913,
+      "step": 13456
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7139736806850454e-08,
+      "loss": 0.397,
+      "step": 13457
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.693828884650974e-08,
+      "loss": 0.3806,
+      "step": 13458
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.673738769435553e-08,
+      "loss": 0.4331,
+      "step": 13459
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6537033361414564e-08,
+      "loss": 0.4017,
+      "step": 13460
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.633722585868249e-08,
+      "loss": 0.4019,
+      "step": 13461
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.613796519712609e-08,
+      "loss": 0.4086,
+      "step": 13462
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.593925138767995e-08,
+      "loss": 0.4165,
+      "step": 13463
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5741084441250904e-08,
+      "loss": 0.4141,
+      "step": 13464
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.554346436871581e-08,
+      "loss": 0.3949,
+      "step": 13465
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.534639118092043e-08,
+      "loss": 0.4151,
+      "step": 13466
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.514986488867944e-08,
+      "loss": 0.3871,
+      "step": 13467
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4953885502782005e-08,
+      "loss": 0.4072,
+      "step": 13468
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.475845303398062e-08,
+      "loss": 0.4002,
+      "step": 13469
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.456356749300449e-08,
+      "loss": 0.418,
+      "step": 13470
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.436922889054728e-08,
+      "loss": 0.4079,
+      "step": 13471
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.417543723727601e-08,
+      "loss": 0.4296,
+      "step": 13472
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.398219254382551e-08,
+      "loss": 0.3979,
+      "step": 13473
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.378949482080396e-08,
+      "loss": 0.3901,
+      "step": 13474
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.359734407878512e-08,
+      "loss": 0.3845,
+      "step": 13475
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3405740328316116e-08,
+      "loss": 0.3685,
+      "step": 13476
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.321468357991298e-08,
+      "loss": 0.4071,
+      "step": 13477
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3024173844060646e-08,
+      "loss": 0.3972,
+      "step": 13478
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.283421113121521e-08,
+      "loss": 0.421,
+      "step": 13479
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.264479545180277e-08,
+      "loss": 0.4002,
+      "step": 13480
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.245592681622056e-08,
+      "loss": 0.3694,
+      "step": 13481
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.226760523483141e-08,
+      "loss": 0.3615,
+      "step": 13482
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.207983071797261e-08,
+      "loss": 0.3655,
+      "step": 13483
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.189260327595034e-08,
+      "loss": 0.3598,
+      "step": 13484
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1705922919039735e-08,
+      "loss": 0.3714,
+      "step": 13485
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.151978965748592e-08,
+      "loss": 0.433,
+      "step": 13486
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.133420350150629e-08,
+      "loss": 0.3954,
+      "step": 13487
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.114916446128491e-08,
+      "loss": 0.4098,
+      "step": 13488
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.096467254697699e-08,
+      "loss": 0.4229,
+      "step": 13489
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.078072776870999e-08,
+      "loss": 0.4407,
+      "step": 13490
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0597330136578066e-08,
+      "loss": 0.4213,
+      "step": 13491
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.041447966064648e-08,
+      "loss": 0.4044,
+      "step": 13492
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.023217635095277e-08,
+      "loss": 0.3802,
+      "step": 13493
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0050420217500044e-08,
+      "loss": 0.4156,
+      "step": 13494
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9869211270264767e-08,
+      "loss": 0.4096,
+      "step": 13495
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.968854951919231e-08,
+      "loss": 0.3967,
+      "step": 13496
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9508434974196976e-08,
+      "loss": 0.3856,
+      "step": 13497
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9328867645165294e-08,
+      "loss": 0.4062,
+      "step": 13498
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9149847541952714e-08,
+      "loss": 0.3894,
+      "step": 13499
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.89713746743836e-08,
+      "loss": 0.4003,
+      "step": 13500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8793449052254564e-08,
+      "loss": 0.3918,
+      "step": 13501
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.86160706853289e-08,
+      "loss": 0.4198,
+      "step": 13502
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.843923958334327e-08,
+      "loss": 0.3816,
+      "step": 13503
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.826295575600213e-08,
+      "loss": 0.3768,
+      "step": 13504
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8087219212979965e-08,
+      "loss": 0.3868,
+      "step": 13505
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.7912029963923505e-08,
+      "loss": 0.4072,
+      "step": 13506
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.7737388018446164e-08,
+      "loss": 0.3837,
+      "step": 13507
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7563293386133615e-08,
+      "loss": 0.4152,
+      "step": 13508
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7389746076541544e-08,
+      "loss": 0.393,
+      "step": 13509
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7216746099193448e-08,
+      "loss": 0.432,
+      "step": 13510
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7044293463585057e-08,
+      "loss": 0.3738,
+      "step": 13511
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.687238817917992e-08,
+      "loss": 0.4014,
+      "step": 13512
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6701030255413817e-08,
+      "loss": 0.4247,
+      "step": 13513
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6530219701692563e-08,
+      "loss": 0.4089,
+      "step": 13514
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6359956527388652e-08,
+      "loss": 0.3933,
+      "step": 13515
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.619024074184795e-08,
+      "loss": 0.4084,
+      "step": 13516
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6021072354385223e-08,
+      "loss": 0.4029,
+      "step": 13517
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.585245137428416e-08,
+      "loss": 0.3938,
+      "step": 13518
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5684377810800687e-08,
+      "loss": 0.4053,
+      "step": 13519
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5516851673157427e-08,
+      "loss": 0.4152,
+      "step": 13520
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.534987297055036e-08,
+      "loss": 0.3686,
+      "step": 13521
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5183441712143264e-08,
+      "loss": 0.3717,
+      "step": 13522
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5017557907069945e-08,
+      "loss": 0.3681,
+      "step": 13523
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4852221564436452e-08,
+      "loss": 0.4427,
+      "step": 13524
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4687432693314417e-08,
+      "loss": 0.3939,
+      "step": 13525
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4523191302751048e-08,
+      "loss": 0.4131,
+      "step": 13526
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4359497401758026e-08,
+      "loss": 0.4258,
+      "step": 13527
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4196350999320385e-08,
+      "loss": 0.3571,
+      "step": 13528
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4033752104392072e-08,
+      "loss": 0.412,
+      "step": 13529
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3871700725897062e-08,
+      "loss": 0.3848,
+      "step": 13530
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.371019687272935e-08,
+      "loss": 0.3506,
+      "step": 13531
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3549240553752962e-08,
+      "loss": 0.379,
+      "step": 13532
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3388831777801934e-08,
+      "loss": 0.3823,
+      "step": 13533
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3228970553679232e-08,
+      "loss": 0.4515,
+      "step": 13534
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.3069656890160054e-08,
+      "loss": 0.3981,
+      "step": 13535
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2910890795986295e-08,
+      "loss": 0.3996,
+      "step": 13536
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2752672279873212e-08,
+      "loss": 0.3979,
+      "step": 13537
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.259500135050274e-08,
+      "loss": 0.3868,
+      "step": 13538
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.24378780165313e-08,
+      "loss": 0.4149,
+      "step": 13539
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2281302286578655e-08,
+      "loss": 0.3988,
+      "step": 13540
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2125274169241262e-08,
+      "loss": 0.4068,
+      "step": 13541
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1969793673080052e-08,
+      "loss": 0.391,
+      "step": 13542
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.181486080663042e-08,
+      "loss": 0.3931,
+      "step": 13543
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1660475578395567e-08,
+      "loss": 0.4171,
+      "step": 13544
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1506637996846492e-08,
+      "loss": 0.4221,
+      "step": 13545
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.135334807042866e-08,
+      "loss": 0.3777,
+      "step": 13546
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.120060580755423e-08,
+      "loss": 0.4209,
+      "step": 13547
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1048411216605392e-08,
+      "loss": 0.3968,
+      "step": 13548
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0896764305936567e-08,
+      "loss": 0.3689,
+      "step": 13549
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0745665083871103e-08,
+      "loss": 0.4304,
+      "step": 13550
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0595113558700143e-08,
+      "loss": 0.3345,
+      "step": 13551
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0445109738687074e-08,
+      "loss": 0.391,
+      "step": 13552
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0295653632065316e-08,
+      "loss": 0.3623,
+      "step": 13553
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.014674524703608e-08,
+      "loss": 0.452,
+      "step": 13554
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9998384591773945e-08,
+      "loss": 0.3668,
+      "step": 13555
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9850571674420172e-08,
+      "loss": 0.3924,
+      "step": 13556
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.970330650308827e-08,
+      "loss": 0.4245,
+      "step": 13557
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9556589085859557e-08,
+      "loss": 0.3973,
+      "step": 13558
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9410419430786476e-08,
+      "loss": 0.3972,
+      "step": 13559
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.92647975458915e-08,
+      "loss": 0.3919,
+      "step": 13560
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9119723439167127e-08,
+      "loss": 0.421,
+      "step": 13561
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.8975197118575874e-08,
+      "loss": 0.3652,
+      "step": 13562
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.883121859204917e-08,
+      "loss": 0.4129,
+      "step": 13563
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8687787867489594e-08,
+      "loss": 0.3607,
+      "step": 13564
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8544904952768617e-08,
+      "loss": 0.4129,
+      "step": 13565
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8402569855728856e-08,
+      "loss": 0.3967,
+      "step": 13566
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8260782584181846e-08,
+      "loss": 0.3952,
+      "step": 13567
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8119543145909134e-08,
+      "loss": 0.4208,
+      "step": 13568
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7978851548662303e-08,
+      "loss": 0.4176,
+      "step": 13569
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7838707800162946e-08,
+      "loss": 0.4054,
+      "step": 13570
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7699111908103805e-08,
+      "loss": 0.3938,
+      "step": 13571
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7560063880145417e-08,
+      "loss": 0.4346,
+      "step": 13572
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7421563723919453e-08,
+      "loss": 0.3977,
+      "step": 13573
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7283611447026505e-08,
+      "loss": 0.4586,
+      "step": 13574
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.71462070570394e-08,
+      "loss": 0.3638,
+      "step": 13575
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.700935056149766e-08,
+      "loss": 0.4291,
+      "step": 13576
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6873041967914172e-08,
+      "loss": 0.4159,
+      "step": 13577
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6737281283769613e-08,
+      "loss": 0.393,
+      "step": 13578
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.660206851651358e-08,
+      "loss": 0.4169,
+      "step": 13579
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6467403673569028e-08,
+      "loss": 0.3925,
+      "step": 13580
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6333286762325596e-08,
+      "loss": 0.3933,
+      "step": 13581
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6199717790145174e-08,
+      "loss": 0.4014,
+      "step": 13582
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6066696764356348e-08,
+      "loss": 0.4266,
+      "step": 13583
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5934223692262165e-08,
+      "loss": 0.3973,
+      "step": 13584
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.580229858113236e-08,
+      "loss": 0.397,
+      "step": 13585
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5670921438206698e-08,
+      "loss": 0.3992,
+      "step": 13586
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5540092270696085e-08,
+      "loss": 0.3909,
+      "step": 13587
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5409811085781435e-08,
+      "loss": 0.4213,
+      "step": 13588
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5280077890612587e-08,
+      "loss": 0.4362,
+      "step": 13589
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5150892692310514e-08,
+      "loss": 0.4177,
+      "step": 13590
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.502225549796288e-08,
+      "loss": 0.4245,
+      "step": 13591
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4894166314632919e-08,
+      "loss": 0.4235,
+      "step": 13592
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4766625149349457e-08,
+      "loss": 0.4265,
+      "step": 13593
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4639632009111337e-08,
+      "loss": 0.3956,
+      "step": 13594
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4513186900889653e-08,
+      "loss": 0.4194,
+      "step": 13595
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4387289831624407e-08,
+      "loss": 0.4536,
+      "step": 13596
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4261940808224518e-08,
+      "loss": 0.3985,
+      "step": 13597
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4137139837568925e-08,
+      "loss": 0.388,
+      "step": 13598
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4012886926507707e-08,
+      "loss": 0.3846,
+      "step": 13599
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3889182081860963e-08,
+      "loss": 0.3997,
+      "step": 13600
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3766025310417708e-08,
+      "loss": 0.4049,
+      "step": 13601
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3643416618936977e-08,
+      "loss": 0.3957,
+      "step": 13602
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3521356014148946e-08,
+      "loss": 0.3945,
+      "step": 13603
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3399843502751585e-08,
+      "loss": 0.4113,
+      "step": 13604
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.327887909141401e-08,
+      "loss": 0.42,
+      "step": 13605
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.315846278677535e-08,
+      "loss": 0.379,
+      "step": 13606
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3038594595444764e-08,
+      "loss": 0.4119,
+      "step": 13607
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2919274524000324e-08,
+      "loss": 0.3874,
+      "step": 13608
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2800502578991236e-08,
+      "loss": 0.4126,
+      "step": 13609
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.268227876693673e-08,
+      "loss": 0.412,
+      "step": 13610
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2564603094323835e-08,
+      "loss": 0.3576,
+      "step": 13611
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2447475567612942e-08,
+      "loss": 0.3883,
+      "step": 13612
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2330896193231134e-08,
+      "loss": 0.3837,
+      "step": 13613
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2214864977576623e-08,
+      "loss": 0.3971,
+      "step": 13614
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.209938192701876e-08,
+      "loss": 0.3984,
+      "step": 13615
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1984447047893589e-08,
+      "loss": 0.3945,
+      "step": 13616
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1870060346511614e-08,
+      "loss": 0.3765,
+      "step": 13617
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1756221829148929e-08,
+      "loss": 0.4422,
+      "step": 13618
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1642931502053866e-08,
+      "loss": 0.3844,
+      "step": 13619
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.15301893714459e-08,
+      "loss": 0.4095,
+      "step": 13620
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1417995443510078e-08,
+      "loss": 0.3882,
+      "step": 13621
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1306349724404809e-08,
+      "loss": 0.4125,
+      "step": 13622
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1195252220258523e-08,
+      "loss": 0.411,
+      "step": 13623
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1084702937168568e-08,
+      "loss": 0.3909,
+      "step": 13624
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0974701881200089e-08,
+      "loss": 0.3815,
+      "step": 13625
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0865249058393811e-08,
+      "loss": 0.4442,
+      "step": 13626
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0756344474753822e-08,
+      "loss": 0.4223,
+      "step": 13627
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.064798813625867e-08,
+      "loss": 0.4243,
+      "step": 13628
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.054018004885582e-08,
+      "loss": 0.3748,
+      "step": 13629
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0432920218461651e-08,
+      "loss": 0.4144,
+      "step": 13630
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0326208650962566e-08,
+      "loss": 0.4116,
+      "step": 13631
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0220045352214991e-08,
+      "loss": 0.3677,
+      "step": 13632
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0114430328046487e-08,
+      "loss": 0.3754,
+      "step": 13633
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0009363584252419e-08,
+      "loss": 0.4133,
+      "step": 13634
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.904845126600393e-09,
+      "loss": 0.436,
+      "step": 13635
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.800874960826934e-09,
+      "loss": 0.4153,
+      "step": 13636
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.697453092636367e-09,
+      "loss": 0.3774,
+      "step": 13637
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.594579527707481e-09,
+      "loss": 0.3794,
+      "step": 13638
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.492254271684653e-09,
+      "loss": 0.3699,
+      "step": 13639
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.39047733018339e-09,
+      "loss": 0.3937,
+      "step": 13640
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.289248708791442e-09,
+      "loss": 0.4076,
+      "step": 13641
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.18856841306326e-09,
+      "loss": 0.4011,
+      "step": 13642
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.088436448524418e-09,
+      "loss": 0.4049,
+      "step": 13643
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.988852820671634e-09,
+      "loss": 0.3991,
+      "step": 13644
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.889817534969425e-09,
+      "loss": 0.3981,
+      "step": 13645
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.791330596853442e-09,
+      "loss": 0.4104,
+      "step": 13646
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.693392011728253e-09,
+      "loss": 0.3871,
+      "step": 13647
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.596001784970665e-09,
+      "loss": 0.3974,
+      "step": 13648
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.499159921925293e-09,
+      "loss": 0.3618,
+      "step": 13649
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.402866427906776e-09,
+      "loss": 0.3959,
+      "step": 13650
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.307121308199772e-09,
+      "loss": 0.387,
+      "step": 13651
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.211924568058972e-09,
+      "loss": 0.4011,
+      "step": 13652
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.11727621271019e-09,
+      "loss": 0.383,
+      "step": 13653
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.023176247348164e-09,
+      "loss": 0.4108,
+      "step": 13654
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.929624677135428e-09,
+      "loss": 0.3739,
+      "step": 13655
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.836621507208986e-09,
+      "loss": 0.4136,
+      "step": 13656
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.744166742671422e-09,
+      "loss": 0.381,
+      "step": 13657
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.652260388597565e-09,
+      "loss": 0.4086,
+      "step": 13658
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.56090245003116e-09,
+      "loss": 0.4064,
+      "step": 13659
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.470092931987082e-09,
+      "loss": 0.3956,
+      "step": 13660
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.379831839448015e-09,
+      "loss": 0.4458,
+      "step": 13661
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.290119177368882e-09,
+      "loss": 0.4144,
+      "step": 13662
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.200954950673522e-09,
+      "loss": 0.4103,
+      "step": 13663
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.112339164254689e-09,
+      "loss": 0.4277,
+      "step": 13664
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.0242718229773796e-09,
+      "loss": 0.4096,
+      "step": 13665
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.936752931673285e-09,
+      "loss": 0.3931,
+      "step": 13666
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.84978249514745e-09,
+      "loss": 0.3654,
+      "step": 13667
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.7633605181716136e-09,
+      "loss": 0.4107,
+      "step": 13668
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.6774870054886475e-09,
+      "loss": 0.4286,
+      "step": 13669
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.59216196181367e-09,
+      "loss": 0.4172,
+      "step": 13670
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.5073853918284916e-09,
+      "loss": 0.3818,
+      "step": 13671
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.4231573001849455e-09,
+      "loss": 0.3923,
+      "step": 13672
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.339477691508222e-09,
+      "loss": 0.3942,
+      "step": 13673
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.256346570387983e-09,
+      "loss": 0.4466,
+      "step": 13674
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.1737639413894655e-09,
+      "loss": 0.4301,
+      "step": 13675
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.091729809042379e-09,
+      "loss": 0.3889,
+      "step": 13676
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.010244177852009e-09,
+      "loss": 0.4109,
+      "step": 13677
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.929307052288114e-09,
+      "loss": 0.3812,
+      "step": 13678
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.848918436794915e-09,
+      "loss": 0.4145,
+      "step": 13679
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.769078335782219e-09,
+      "loss": 0.4007,
+      "step": 13680
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.689786753633186e-09,
+      "loss": 0.388,
+      "step": 13681
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.611043694699891e-09,
+      "loss": 0.3823,
+      "step": 13682
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.532849163303322e-09,
+      "loss": 0.3694,
+      "step": 13683
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.4552031637356005e-09,
+      "loss": 0.4107,
+      "step": 13684
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.378105700257763e-09,
+      "loss": 0.4302,
+      "step": 13685
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.301556777101979e-09,
+      "loss": 0.4098,
+      "step": 13686
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.225556398468224e-09,
+      "loss": 0.4063,
+      "step": 13687
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.150104568528713e-09,
+      "loss": 0.4384,
+      "step": 13688
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.075201291423471e-09,
+      "loss": 0.3908,
+      "step": 13689
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.000846571264762e-09,
+      "loss": 0.3803,
+      "step": 13690
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.927040412133766e-09,
+      "loss": 0.395,
+      "step": 13691
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.853782818078357e-09,
+      "loss": 0.3889,
+      "step": 13692
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.781073793123092e-09,
+      "loss": 0.3834,
+      "step": 13693
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.708913341254784e-09,
+      "loss": 0.39,
+      "step": 13694
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.637301466436927e-09,
+      "loss": 0.4265,
+      "step": 13695
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.566238172597492e-09,
+      "loss": 0.3746,
+      "step": 13696
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.495723463638913e-09,
+      "loss": 0.3925,
+      "step": 13697
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.4257573434292046e-09,
+      "loss": 0.383,
+      "step": 13698
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.35633981580974e-09,
+      "loss": 0.4045,
+      "step": 13699
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.287470884588585e-09,
+      "loss": 0.4032,
+      "step": 13700
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.219150553548268e-09,
+      "loss": 0.3821,
+      "step": 13701
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.151378826436903e-09,
+      "loss": 0.3864,
+      "step": 13702
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.084155706973736e-09,
+      "loss": 0.3809,
+      "step": 13703
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.0174811988480385e-09,
+      "loss": 0.4208,
+      "step": 13704
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.951355305720217e-09,
+      "loss": 0.4389,
+      "step": 13705
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.885778031219589e-09,
+      "loss": 0.3779,
+      "step": 13706
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.820749378944388e-09,
+      "loss": 0.4022,
+      "step": 13707
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.756269352462871e-09,
+      "loss": 0.4153,
+      "step": 13708
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.6923379553166495e-09,
+      "loss": 0.3844,
+      "step": 13709
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.628955191011807e-09,
+      "loss": 0.4304,
+      "step": 13710
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.5661210630277832e-09,
+      "loss": 0.384,
+      "step": 13711
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.5038355748140407e-09,
+      "loss": 0.3669,
+      "step": 13712
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.4420987297878462e-09,
+      "loss": 0.4089,
+      "step": 13713
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.38091053133871e-09,
+      "loss": 0.4013,
+      "step": 13714
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.320270982823948e-09,
+      "loss": 0.4105,
+      "step": 13715
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.260180087570897e-09,
+      "loss": 0.447,
+      "step": 13716
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.20063784888025e-09,
+      "loss": 0.4101,
+      "step": 13717
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.1416442700171745e-09,
+      "loss": 0.3706,
+      "step": 13718
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.08319935422019e-09,
+      "loss": 0.4116,
+      "step": 13719
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.025303104697841e-09,
+      "loss": 0.423,
+      "step": 13720
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.967955524627586e-09,
+      "loss": 0.4616,
+      "step": 13721
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.911156617155797e-09,
+      "loss": 0.3999,
+      "step": 13722
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.8549063854010907e-09,
+      "loss": 0.4253,
+      "step": 13723
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.799204832448776e-09,
+      "loss": 0.4325,
+      "step": 13724
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.744051961358629e-09,
+      "loss": 0.3746,
+      "step": 13725
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.6894477751548964e-09,
+      "loss": 0.3623,
+      "step": 13726
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.6353922768362903e-09,
+      "loss": 0.3948,
+      "step": 13727
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5818854693693273e-09,
+      "loss": 0.4109,
+      "step": 13728
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.528927355690547e-09,
+      "loss": 0.3957,
+      "step": 13729
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4765179387054028e-09,
+      "loss": 0.4138,
+      "step": 13730
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.424657221291593e-09,
+      "loss": 0.4022,
+      "step": 13731
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3733452062946195e-09,
+      "loss": 0.4091,
+      "step": 13732
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3225818965311174e-09,
+      "loss": 0.3768,
+      "step": 13733
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2723672947877472e-09,
+      "loss": 0.3926,
+      "step": 13734
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.222701403818972e-09,
+      "loss": 0.4278,
+      "step": 13735
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1735842263514993e-09,
+      "loss": 0.3932,
+      "step": 13736
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.125015765080951e-09,
+      "loss": 0.3953,
+      "step": 13737
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.076996022674083e-09,
+      "loss": 0.4092,
+      "step": 13738
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.029525001764343e-09,
+      "loss": 0.4084,
+      "step": 13739
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.982602704958536e-09,
+      "loss": 0.4693,
+      "step": 13740
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.936229134831269e-09,
+      "loss": 0.3397,
+      "step": 13741
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8904042939271726e-09,
+      "loss": 0.3843,
+      "step": 13742
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.845128184763123e-09,
+      "loss": 0.4119,
+      "step": 13743
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8004008098226889e-09,
+      "loss": 0.4445,
+      "step": 13744
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7562221715605731e-09,
+      "loss": 0.3377,
+      "step": 13745
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7125922724015032e-09,
+      "loss": 0.4173,
+      "step": 13746
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6695111147402298e-09,
+      "loss": 0.3955,
+      "step": 13747
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6269787009415283e-09,
+      "loss": 0.3988,
+      "step": 13748
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5849950333401975e-09,
+      "loss": 0.4208,
+      "step": 13749
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5435601142388402e-09,
+      "loss": 0.4236,
+      "step": 13750
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.502673945912303e-09,
+      "loss": 0.3981,
+      "step": 13751
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4623365306054571e-09,
+      "loss": 0.3674,
+      "step": 13752
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.422547870530977e-09,
+      "loss": 0.4296,
+      "step": 13753
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3833079678726713e-09,
+      "loss": 0.3867,
+      "step": 13754
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3446168247843727e-09,
+      "loss": 0.3974,
+      "step": 13755
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.306474443391048e-09,
+      "loss": 0.4571,
+      "step": 13756
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2688808257832475e-09,
+      "loss": 0.4376,
+      "step": 13757
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2318359740270957e-09,
+      "loss": 0.3688,
+      "step": 13758
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.195339890153191e-09,
+      "loss": 0.4199,
+      "step": 13759
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1593925761665958e-09,
+      "loss": 0.3762,
+      "step": 13760
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1239940340390664e-09,
+      "loss": 0.409,
+      "step": 13761
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0891442657134932e-09,
+      "loss": 0.4045,
+      "step": 13762
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0548432731039005e-09,
+      "loss": 0.3656,
+      "step": 13763
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0210910580910061e-09,
+      "loss": 0.3898,
+      "step": 13764
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.878876225277723e-10,
+      "loss": 0.3999,
+      "step": 13765
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.552329682371853e-10,
+      "loss": 0.3977,
+      "step": 13766
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.23127097011145e-10,
+      "loss": 0.4125,
+      "step": 13767
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.915700106115755e-10,
+      "loss": 0.4382,
+      "step": 13768
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.605617107704245e-10,
+      "loss": 0.4223,
+      "step": 13769
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.301021991885538e-10,
+      "loss": 0.4006,
+      "step": 13770
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.001914775401798e-10,
+      "loss": 0.393,
+      "step": 13771
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.708295474651018e-10,
+      "loss": 0.4047,
+      "step": 13772
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.420164105742533e-10,
+      "loss": 0.4002,
+      "step": 13773
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.137520684508126e-10,
+      "loss": 0.4257,
+      "step": 13774
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.860365226446508e-10,
+      "loss": 0.364,
+      "step": 13775
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.588697746778838e-10,
+      "loss": 0.3965,
+      "step": 13776
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.322518260415411e-10,
+      "loss": 0.3703,
+      "step": 13777
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.061826781944557e-10,
+      "loss": 0.3761,
+      "step": 13778
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.806623325699256e-10,
+      "loss": 0.4153,
+      "step": 13779
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.556907905679421e-10,
+      "loss": 0.3811,
+      "step": 13780
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.312680535574099e-10,
+      "loss": 0.4398,
+      "step": 13781
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.07394122880589e-10,
+      "loss": 0.357,
+      "step": 13782
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.840689998475423e-10,
+      "loss": 0.4424,
+      "step": 13783
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.6129268573835706e-10,
+      "loss": 0.3796,
+      "step": 13784
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.3906518180203416e-10,
+      "loss": 0.4013,
+      "step": 13785
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.173864892587087e-10,
+      "loss": 0.4228,
+      "step": 13786
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.9625660929965004e-10,
+      "loss": 0.4004,
+      "step": 13787
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.7567554308393097e-10,
+      "loss": 0.375,
+      "step": 13788
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.5564329174064826e-10,
+      "loss": 0.3809,
+      "step": 13789
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.361598563689228e-10,
+      "loss": 0.4238,
+      "step": 13790
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.172252380378993e-10,
+      "loss": 0.3895,
+      "step": 13791
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.9883943778896697e-10,
+      "loss": 0.383,
+      "step": 13792
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.810024566290981e-10,
+      "loss": 0.4114,
+      "step": 13793
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.6371429553750937e-10,
+      "loss": 0.3853,
+      "step": 13794
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.469749554634415e-10,
+      "loss": 0.3845,
+      "step": 13795
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.3078443732504897e-10,
+      "loss": 0.4012,
+      "step": 13796
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.1514274201273055e-10,
+      "loss": 0.3843,
+      "step": 13797
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0004987038246827e-10,
+      "loss": 0.3533,
+      "step": 13798
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.855058232647089e-10,
+      "loss": 0.4105,
+      "step": 13799
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.7151060145548237e-10,
+      "loss": 0.4246,
+      "step": 13800
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5806420572528348e-10,
+      "loss": 0.4395,
+      "step": 13801
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4516663681019006e-10,
+      "loss": 0.3983,
+      "step": 13802
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.3281789541963464e-10,
+      "loss": 0.3948,
+      "step": 13803
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2101798222974303e-10,
+      "loss": 0.4021,
+      "step": 13804
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.097668978888855e-10,
+      "loss": 0.3943,
+      "step": 13805
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.906464301545626e-11,
+      "loss": 0.4145,
+      "step": 13806
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.891121819565307e-11,
+      "loss": 0.37,
+      "step": 13807
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.930662398680789e-11,
+      "loss": 0.4309,
+      "step": 13808
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.025086091627665e-11,
+      "loss": 0.3931,
+      "step": 13809
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.174392948143925e-11,
+      "loss": 0.4091,
+      "step": 13810
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.37858301485894e-11,
+      "loss": 0.4108,
+      "step": 13811
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.6376563355154945e-11,
+      "loss": 0.3741,
+      "step": 13812
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.951612950747752e-11,
+      "loss": 0.4038,
+      "step": 13813
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3204528981922727e-11,
+      "loss": 0.4163,
+      "step": 13814
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.7441762124880146e-11,
+      "loss": 0.3964,
+      "step": 13815
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.2227829252763345e-11,
+      "loss": 0.3895,
+      "step": 13816
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.756273065200986e-11,
+      "loss": 0.436,
+      "step": 13817
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.3446466577970996e-11,
+      "loss": 0.4173,
+      "step": 13818
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.87903725713224e-12,
+      "loss": 0.4001,
+      "step": 13819
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.860442884892848e-12,
+      "loss": 0.3915,
+      "step": 13820
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.390683626676051e-12,
+      "loss": 0.3974,
+      "step": 13821
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.4697596190392803e-12,
+      "loss": 0.394,
+      "step": 13822
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0976709674537234e-12,
+      "loss": 0.3977,
+      "step": 13823
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.744177451940999e-13,
+      "loss": 0.3877,
+      "step": 13824
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.4142,
+      "step": 13825
+    },
+    {
+      "epoch": 5.0,
+      "step": 13825,
+      "total_flos": 0.0,
+      "train_loss": 0.47639201751022614,
+      "train_runtime": 63559.3128,
+      "train_samples_per_second": 13.92,
+      "train_steps_per_second": 0.218
+    }
+  ],
+  "max_steps": 13825,
+  "num_train_epochs": 5,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}