diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,21463 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998600811529313,
+  "global_step": 3573,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.259259259259259e-06,
+      "loss": 5.3271,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 2.9834,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7777777777777776e-05,
+      "loss": 2.793,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 1.9531,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6296296296296294e-05,
+      "loss": 1.3994,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.555555555555555e-05,
+      "loss": 1.4438,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.481481481481482e-05,
+      "loss": 1.3291,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.407407407407407e-05,
+      "loss": 1.3271,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-05,
+      "loss": 1.3008,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.259259259259259e-05,
+      "loss": 1.3281,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010185185185185185,
+      "loss": 1.2334,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001111111111111111,
+      "loss": 1.3096,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012037037037037036,
+      "loss": 1.2695,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012962962962962963,
+      "loss": 1.3018,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001388888888888889,
+      "loss": 1.2915,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014814814814814815,
+      "loss": 1.2026,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001574074074074074,
+      "loss": 1.2705,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 1.1392,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017592592592592595,
+      "loss": 1.1023,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018518518518518518,
+      "loss": 1.1401,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019444444444444446,
+      "loss": 1.147,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002037037037037037,
+      "loss": 1.1333,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021296296296296298,
+      "loss": 1.1174,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002222222222222222,
+      "loss": 1.1333,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002314814814814815,
+      "loss": 1.1506,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024074074074074072,
+      "loss": 1.0442,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025,
+      "loss": 1.1055,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025925925925925926,
+      "loss": 1.0588,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002685185185185186,
+      "loss": 1.0542,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002777777777777778,
+      "loss": 1.0547,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028703703703703703,
+      "loss": 1.0312,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002962962962962963,
+      "loss": 1.0566,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003055555555555556,
+      "loss": 1.0449,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003148148148148148,
+      "loss": 1.0083,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032407407407407406,
+      "loss": 0.9839,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 0.9622,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034259259259259263,
+      "loss": 0.9438,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003518518518518519,
+      "loss": 0.9846,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003611111111111111,
+      "loss": 0.9641,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037037037037037035,
+      "loss": 0.9565,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037962962962962966,
+      "loss": 0.9578,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003888888888888889,
+      "loss": 0.9314,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003981481481481481,
+      "loss": 0.8831,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004074074074074074,
+      "loss": 0.8835,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004166666666666667,
+      "loss": 0.9028,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042592592592592595,
+      "loss": 0.8694,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004351851851851852,
+      "loss": 0.8613,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004444444444444444,
+      "loss": 0.8879,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004537037037037037,
+      "loss": 0.8289,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000462962962962963,
+      "loss": 0.823,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047222222222222224,
+      "loss": 0.7856,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00048148148148148144,
+      "loss": 0.8069,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004907407407407408,
+      "loss": 0.8035,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 0.8062,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005092592592592593,
+      "loss": 0.7869,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005185185185185185,
+      "loss": 0.7446,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005277777777777778,
+      "loss": 0.7915,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005370370370370371,
+      "loss": 0.7271,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005462962962962962,
+      "loss": 0.7563,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005555555555555556,
+      "loss": 0.7048,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005648148148148148,
+      "loss": 0.7581,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005740740740740741,
+      "loss": 0.7473,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005833333333333334,
+      "loss": 0.7417,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005925925925925926,
+      "loss": 0.7119,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006018518518518519,
+      "loss": 0.739,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006111111111111112,
+      "loss": 0.7034,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006203703703703704,
+      "loss": 0.7344,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006296296296296296,
+      "loss": 0.7094,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006388888888888888,
+      "loss": 0.6763,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006481481481481481,
+      "loss": 0.6714,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006574074074074074,
+      "loss": 0.6892,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006666666666666666,
+      "loss": 0.672,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000675925925925926,
+      "loss": 0.668,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006851851851851853,
+      "loss": 0.6782,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006944444444444445,
+      "loss": 0.6576,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007037037037037038,
+      "loss": 0.6442,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007129629629629629,
+      "loss": 0.6594,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007222222222222222,
+      "loss": 0.692,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007314814814814815,
+      "loss": 0.6655,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007407407407407407,
+      "loss": 0.6846,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00075,
+      "loss": 0.6907,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007592592592592593,
+      "loss": 0.6665,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007685185185185185,
+      "loss": 0.6848,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007777777777777778,
+      "loss": 0.6711,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007870370370370372,
+      "loss": 0.6494,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007962962962962962,
+      "loss": 0.6973,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008055555555555556,
+      "loss": 0.6796,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008148148148148148,
+      "loss": 0.6638,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008240740740740741,
+      "loss": 0.6653,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008333333333333334,
+      "loss": 0.6637,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008425925925925926,
+      "loss": 0.6724,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008518518518518519,
+      "loss": 0.677,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008611111111111112,
+      "loss": 0.6586,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008703703703703704,
+      "loss": 0.6571,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008796296296296296,
+      "loss": 0.6606,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008888888888888888,
+      "loss": 0.6143,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008981481481481481,
+      "loss": 0.6345,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009074074074074074,
+      "loss": 0.6274,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009166666666666666,
+      "loss": 0.6422,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000925925925925926,
+      "loss": 0.6013,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009351851851851853,
+      "loss": 0.6711,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009444444444444445,
+      "loss": 0.6802,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009537037037037038,
+      "loss": 0.6605,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009629629629629629,
+      "loss": 0.6357,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009722222222222222,
+      "loss": 0.6648,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009814814814814816,
+      "loss": 0.6193,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009907407407407408,
+      "loss": 0.6143,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 0.6262,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999997944898642,
+      "loss": 0.641,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999991779596254,
+      "loss": 0.6389,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999981504097905,
+      "loss": 0.6416,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999967118412046,
+      "loss": 0.6046,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999948622550496,
+      "loss": 0.6272,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999926016528464,
+      "loss": 0.6211,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999899300364532,
+      "loss": 0.6606,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999868474080663,
+      "loss": 0.635,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999833537702191,
+      "loss": 0.6147,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999794491257845,
+      "loss": 0.6005,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999751334779715,
+      "loss": 0.6185,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999970406830328,
+      "loss": 0.626,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999652691867398,
+      "loss": 0.6171,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999597205514296,
+      "loss": 0.6008,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999537609289592,
+      "loss": 0.5988,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999473903242273,
+      "loss": 0.6172,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999406087424709,
+      "loss": 0.6136,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999334161892649,
+      "loss": 0.5858,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999258126705216,
+      "loss": 0.6487,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999177981924916,
+      "loss": 0.5939,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999909372761763,
+      "loss": 0.6079,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999005363852617,
+      "loss": 0.6233,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999891289070252,
+      "loss": 0.6281,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998816308243352,
+      "loss": 0.6317,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999871561655451,
+      "loss": 0.6274,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998610815718763,
+      "loss": 0.6049,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998501905822267,
+      "loss": 0.6027,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998388886954545,
+      "loss": 0.5989,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998271759208508,
+      "loss": 0.5906,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998150522680437,
+      "loss": 0.6233,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999802517746999,
+      "loss": 0.6562,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997895723680213,
+      "loss": 0.5873,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997762161417516,
+      "loss": 0.6492,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997624490791699,
+      "loss": 0.6096,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997482711915926,
+      "loss": 0.6281,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999733682490675,
+      "loss": 0.6162,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997186829884092,
+      "loss": 0.6385,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997032726971257,
+      "loss": 0.6028,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996874516294925,
+      "loss": 0.6238,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996712197985146,
+      "loss": 0.6558,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999654577217536,
+      "loss": 0.6548,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996375239002368,
+      "loss": 0.574,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996200598606359,
+      "loss": 0.6024,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996021851130896,
+      "loss": 0.6282,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995838996722914,
+      "loss": 0.616,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995652035532728,
+      "loss": 0.6565,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995460967714028,
+      "loss": 0.6443,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995265793423878,
+      "loss": 0.6226,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995066512822719,
+      "loss": 0.5945,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999486312607437,
+      "loss": 0.5975,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994655633346022,
+      "loss": 0.621,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994444034808242,
+      "loss": 0.624,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009994228330634973,
+      "loss": 0.634,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009994008521003533,
+      "loss": 0.5649,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993784606094611,
+      "loss": 0.6106,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999355658609228,
+      "loss": 0.6394,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993324461183978,
+      "loss": 0.6099,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993088231560522,
+      "loss": 0.6089,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992847897416102,
+      "loss": 0.6082,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992603458948283,
+      "loss": 0.5898,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992354916358002,
+      "loss": 0.6467,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992102269849572,
+      "loss": 0.6362,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991845519630679,
+      "loss": 0.6071,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999158466591238,
+      "loss": 0.5936,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991319708909112,
+      "loss": 0.6096,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991050648838675,
+      "loss": 0.6129,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999077748592225,
+      "loss": 0.5919,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999050022038439,
+      "loss": 0.5933,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990218852453014,
+      "loss": 0.6377,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989933382359422,
+      "loss": 0.587,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998964381033828,
+      "loss": 0.5983,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989350136627626,
+      "loss": 0.593,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989052361468877,
+      "loss": 0.5579,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998875048510681,
+      "loss": 0.5724,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988444507789582,
+      "loss": 0.5995,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998813442976872,
+      "loss": 0.5957,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987820251299122,
+      "loss": 0.6074,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987501972639052,
+      "loss": 0.5624,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998717959405015,
+      "loss": 0.6122,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986853115797423,
+      "loss": 0.5492,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998652253814925,
+      "loss": 0.6279,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998618786137738,
+      "loss": 0.6,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998584908575693,
+      "loss": 0.5763,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985506211566387,
+      "loss": 0.5757,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985159239087609,
+      "loss": 0.5996,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998480816860582,
+      "loss": 0.5918,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984453000409614,
+      "loss": 0.5902,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984093734790955,
+      "loss": 0.5859,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983730372045171,
+      "loss": 0.6028,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983362912470966,
+      "loss": 0.6124,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982991356370402,
+      "loss": 0.6204,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982615704048916,
+      "loss": 0.6124,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982235955815308,
+      "loss": 0.5625,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981852111981746,
+      "loss": 0.5743,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981464172863768,
+      "loss": 0.5691,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998107213878027,
+      "loss": 0.5675,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980676010053527,
+      "loss": 0.5985,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998027578700917,
+      "loss": 0.624,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979871469976197,
+      "loss": 0.5425,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997946305928697,
+      "loss": 0.5923,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979050555277227,
+      "loss": 0.5851,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978633958286059,
+      "loss": 0.554,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978213268655926,
+      "loss": 0.574,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997778848673265,
+      "loss": 0.5927,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977359612865424,
+      "loss": 0.5826,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976926647406793,
+      "loss": 0.592,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976489590712678,
+      "loss": 0.5873,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976048443142353,
+      "loss": 0.5498,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997560320505846,
+      "loss": 0.5837,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975153876827009,
+      "loss": 0.5466,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974700458817356,
+      "loss": 0.5796,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974242951402235,
+      "loss": 0.5677,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973781354957734,
+      "loss": 0.5925,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973315669863305,
+      "loss": 0.5881,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997284589650176,
+      "loss": 0.5775,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972372035259272,
+      "loss": 0.5596,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971894086525373,
+      "loss": 0.5916,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971412050692953,
+      "loss": 0.5604,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970925928158272,
+      "loss": 0.5925,
+      "step": 227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997043571932094,
+      "loss": 0.5737,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969941424583927,
+      "loss": 0.559,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969443044353563,
+      "loss": 0.5411,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968940579039539,
+      "loss": 0.5934,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968434029054899,
+      "loss": 0.5422,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967923394816048,
+      "loss": 0.5636,
+      "step": 233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967408676742752,
+      "loss": 0.561,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966889875258126,
+      "loss": 0.5898,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966366990788646,
+      "loss": 0.55,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965840023764148,
+      "loss": 0.5483,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965308974617816,
+      "loss": 0.5654,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964773843786196,
+      "loss": 0.5933,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964234631709187,
+      "loss": 0.5439,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963691338830043,
+      "loss": 0.5585,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963143965595373,
+      "loss": 0.5631,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996259251245514,
+      "loss": 0.528,
+      "step": 243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962036979862658,
+      "loss": 0.5804,
+      "step": 244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961477368274603,
+      "loss": 0.5457,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960913678150995,
+      "loss": 0.5438,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996034590995521,
+      "loss": 0.5479,
+      "step": 247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959774064153978,
+      "loss": 0.5571,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959198141217375,
+      "loss": 0.5522,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995861814161884,
+      "loss": 0.609,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995803406583515,
+      "loss": 0.5782,
+      "step": 251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957445914346443,
+      "loss": 0.5387,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956853687636202,
+      "loss": 0.5479,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956257386191263,
+      "loss": 0.5806,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955657010501807,
+      "loss": 0.5626,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955052561061368,
+      "loss": 0.5289,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954444038366827,
+      "loss": 0.5521,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953831442918418,
+      "loss": 0.5551,
+      "step": 258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953214775219716,
+      "loss": 0.559,
+      "step": 259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952594035777648,
+      "loss": 0.6078,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951969225102486,
+      "loss": 0.5685,
+      "step": 261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951340343707852,
+      "loss": 0.5815,
+      "step": 262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950707392110709,
+      "loss": 0.5383,
+      "step": 263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950070370831371,
+      "loss": 0.5865,
+      "step": 264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949429280393495,
+      "loss": 0.5627,
+      "step": 265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948784121324083,
+      "loss": 0.5813,
+      "step": 266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948134894153483,
+      "loss": 0.542,
+      "step": 267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947481599415385,
+      "loss": 0.5345,
+      "step": 268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946824237646824,
+      "loss": 0.5757,
+      "step": 269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946162809388176,
+      "loss": 0.5303,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945497315183164,
+      "loss": 0.5537,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944827755578851,
+      "loss": 0.5823,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944154131125641,
+      "loss": 0.5226,
+      "step": 273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943476442377285,
+      "loss": 0.5403,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942794689890865,
+      "loss": 0.5845,
+      "step": 275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942108874226813,
+      "loss": 0.5536,
+      "step": 276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941418995948895,
+      "loss": 0.59,
+      "step": 277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940725055624217,
+      "loss": 0.5629,
+      "step": 278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940027053823232,
+      "loss": 0.5681,
+      "step": 279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939324991119723,
+      "loss": 0.5544,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993861886809081,
+      "loss": 0.5558,
+      "step": 281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993790868531696,
+      "loss": 0.5417,
+      "step": 282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937194443381972,
+      "loss": 0.5444,
+      "step": 283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993647614287298,
+      "loss": 0.5507,
+      "step": 284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935753784380455,
+      "loss": 0.5348,
+      "step": 285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935027368498206,
+      "loss": 0.5754,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934296895823375,
+      "loss": 0.5563,
+      "step": 287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933562366956443,
+      "loss": 0.5824,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932823782501223,
+      "loss": 0.5697,
+      "step": 289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993208114306486,
+      "loss": 0.5609,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931334449257833,
+      "loss": 0.5275,
+      "step": 291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930583701693951,
+      "loss": 0.5625,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929828900990367,
+      "loss": 0.5382,
+      "step": 293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992907004776755,
+      "loss": 0.5608,
+      "step": 294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928307142649316,
+      "loss": 0.5466,
+      "step": 295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927540186262798,
+      "loss": 0.5638,
+      "step": 296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926769179238466,
+      "loss": 0.5522,
+      "step": 297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992599412221012,
+      "loss": 0.5719,
+      "step": 298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925215015814887,
+      "loss": 0.5178,
+      "step": 299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992443186069323,
+      "loss": 0.5543,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923644657488925,
+      "loss": 0.5487,
+      "step": 301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992285340684909,
+      "loss": 0.5526,
+      "step": 302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922058109424166,
+      "loss": 0.5643,
+      "step": 303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009921258765867918,
+      "loss": 0.5494,
+      "step": 304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000992045537683744,
+      "loss": 0.5636,
+      "step": 305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919647942993148,
+      "loss": 0.5432,
+      "step": 306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918836464998788,
+      "loss": 0.5532,
+      "step": 307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918020943521426,
+      "loss": 0.5657,
+      "step": 308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917201379231455,
+      "loss": 0.5402,
+      "step": 309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991637777280259,
+      "loss": 0.5282,
+      "step": 310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915550124911866,
+      "loss": 0.542,
+      "step": 311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914718436239648,
+      "loss": 0.5114,
+      "step": 312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913882707469613,
+      "loss": 0.569,
+      "step": 313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913042939288768,
+      "loss": 0.559,
+      "step": 314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912199132387433,
+      "loss": 0.5582,
+      "step": 315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911351287459256,
+      "loss": 0.5447,
+      "step": 316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910499405201194,
+      "loss": 0.5363,
+      "step": 317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009909643486313534,
+      "loss": 0.5759,
+      "step": 318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908783531499873,
+      "loss": 0.5757,
+      "step": 319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990791954146713,
+      "loss": 0.5491,
+      "step": 320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009907051516925538,
+      "loss": 0.5635,
+      "step": 321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990617945858865,
+      "loss": 0.5677,
+      "step": 322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905303367173336,
+      "loss": 0.4941,
+      "step": 323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904423243399774,
+      "loss": 0.5519,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903539087991462,
+      "loss": 0.5625,
+      "step": 325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902650901675214,
+      "loss": 0.5487,
+      "step": 326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901758685181153,
+      "loss": 0.5643,
+      "step": 327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900862439242719,
+      "loss": 0.4971,
+      "step": 328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989996216459666,
+      "loss": 0.5499,
+      "step": 329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899057861983043,
+      "loss": 0.5306,
+      "step": 330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898149532145237,
+      "loss": 0.4961,
+      "step": 331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897237175829927,
+      "loss": 0.5378,
+      "step": 332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009896320793787105,
+      "loss": 0.5316,
+      "step": 333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989540038677008,
+      "loss": 0.5586,
+      "step": 334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009894475955535458,
+      "loss": 0.5056,
+      "step": 335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989354750084316,
+      "loss": 0.5375,
+      "step": 336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892615023456416,
+      "loss": 0.4978,
+      "step": 337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989167852414176,
+      "loss": 0.5479,
+      "step": 338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009890738003669028,
+      "loss": 0.5056,
+      "step": 339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988979346281137,
+      "loss": 0.5321,
+      "step": 340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888844902345238,
+      "loss": 0.5464,
+      "step": 341
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009887892323050384,
+      "loss": 0.5424,
+      "step": 342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886935725709868,
+      "loss": 0.5273,
+      "step": 343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885975111110052,
+      "loss": 0.536,
+      "step": 344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885010480040598,
+      "loss": 0.5667,
+      "step": 345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884041833294476,
+      "loss": 0.514,
+      "step": 346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988306917166795,
+      "loss": 0.536,
+      "step": 347
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882092495960588,
+      "loss": 0.5537,
+      "step": 348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009881111806975256,
+      "loss": 0.536,
+      "step": 349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880127105518122,
+      "loss": 0.5471,
+      "step": 350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987913839239865,
+      "loss": 0.5398,
+      "step": 351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009878145668429602,
+      "loss": 0.5176,
+      "step": 352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877148934427035,
+      "loss": 0.5376,
+      "step": 353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987614819121031,
+      "loss": 0.548,
+      "step": 354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875143439602072,
+      "loss": 0.5563,
+      "step": 355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874134680428273,
+      "loss": 0.5785,
+      "step": 356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873121914518153,
+      "loss": 0.5237,
+      "step": 357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872105142704245,
+      "loss": 0.5623,
+      "step": 358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009871084365822376,
+      "loss": 0.5709,
+      "step": 359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870059584711668,
+      "loss": 0.5045,
+      "step": 360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869030800214531,
+      "loss": 0.5264,
+      "step": 361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986799801317667,
+      "loss": 0.4951,
+      "step": 362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866961224447076,
+      "loss": 0.526,
+      "step": 363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986592043487803,
+      "loss": 0.5521,
+      "step": 364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864875645325107,
+      "loss": 0.5271,
+      "step": 365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863826856647163,
+      "loss": 0.554,
+      "step": 366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009862774069706345,
+      "loss": 0.536,
+      "step": 367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986171728536809,
+      "loss": 0.5305,
+      "step": 368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860656504501113,
+      "loss": 0.5724,
+      "step": 369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009859591727977423,
+      "loss": 0.4966,
+      "step": 370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858522956672307,
+      "loss": 0.5087,
+      "step": 371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857450191464337,
+      "loss": 0.5321,
+      "step": 372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009856373433235373,
+      "loss": 0.5228,
+      "step": 373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985529268287055,
+      "loss": 0.5365,
+      "step": 374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009854207941258294,
+      "loss": 0.5869,
+      "step": 375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009853119209290303,
+      "loss": 0.543,
+      "step": 376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852026487861558,
+      "loss": 0.5356,
+      "step": 377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009850929777870322,
+      "loss": 0.5115,
+      "step": 378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009849829080218135,
+      "loss": 0.5446,
+      "step": 379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848724395809816,
+      "loss": 0.5583,
+      "step": 380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847615725553456,
+      "loss": 0.5142,
+      "step": 381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846503070360434,
+      "loss": 0.5237,
+      "step": 382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984538643114539,
+      "loss": 0.5115,
+      "step": 383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844265808826253,
+      "loss": 0.5353,
+      "step": 384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843141204324214,
+      "loss": 0.5417,
+      "step": 385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984201261856375,
+      "loss": 0.511,
+      "step": 386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00098408800524726,
+      "loss": 0.53,
+      "step": 387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839743506981783,
+      "loss": 0.5537,
+      "step": 388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983860298302558,
+      "loss": 0.5569,
+      "step": 389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983745848154155,
+      "loss": 0.5231,
+      "step": 390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009836310003470522,
+      "loss": 0.5458,
+      "step": 391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983515754975659,
+      "loss": 0.526,
+      "step": 392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834001121347118,
+      "loss": 0.561,
+      "step": 393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009832840719192736,
+      "loss": 0.5387,
+      "step": 394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831676344247342,
+      "loss": 0.4817,
+      "step": 395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00098305079974681,
+      "loss": 0.5133,
+      "step": 396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009829335679815438,
+      "loss": 0.5267,
+      "step": 397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009828159392253051,
+      "loss": 0.5728,
+      "step": 398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826979135747892,
+      "loss": 0.5112,
+      "step": 399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009825794911270179,
+      "loss": 0.4948,
+      "step": 400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009824606719793395,
+      "loss": 0.5583,
+      "step": 401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982341456229428,
+      "loss": 0.5142,
+      "step": 402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822218439752835,
+      "loss": 0.5341,
+      "step": 403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009821018353152325,
+      "loss": 0.535,
+      "step": 404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009819814303479266,
+      "loss": 0.5427,
+      "step": 405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009818606291723438,
+      "loss": 0.4922,
+      "step": 406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009817394318877874,
+      "loss": 0.5387,
+      "step": 407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816178385938868,
+      "loss": 0.5262,
+      "step": 408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009814958493905962,
+      "loss": 0.5399,
+      "step": 409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981373464378196,
+      "loss": 0.5223,
+      "step": 410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009812506836572913,
+      "loss": 0.5553,
+      "step": 411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009811275073288131,
+      "loss": 0.545,
+      "step": 412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810039354940171,
+      "loss": 0.5116,
+      "step": 413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009808799682544847,
+      "loss": 0.4861,
+      "step": 414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009807556057121218,
+      "loss": 0.5151,
+      "step": 415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009806308479691594,
+      "loss": 0.5269,
+      "step": 416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009805056951281536,
+      "loss": 0.5688,
+      "step": 417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980380147291985,
+      "loss": 0.5254,
+      "step": 418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802542045638588,
+      "loss": 0.5227,
+      "step": 419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009801278670473054,
+      "loss": 0.5522,
+      "step": 420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009800011348461792,
+      "loss": 0.5447,
+      "step": 421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009798740080646591,
+      "loss": 0.5211,
+      "step": 422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009797464868072487,
+      "loss": 0.5225,
+      "step": 423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796185711787755,
+      "loss": 0.5328,
+      "step": 424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009794902612843912,
+      "loss": 0.5189,
+      "step": 425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793615572295722,
+      "loss": 0.5237,
+      "step": 426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009792324591201178,
+      "loss": 0.5383,
+      "step": 427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791029670621525,
+      "loss": 0.5183,
+      "step": 428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009789730811621236,
+      "loss": 0.5431,
+      "step": 429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009788428015268028,
+      "loss": 0.5499,
+      "step": 430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009787121282632848,
+      "loss": 0.5417,
+      "step": 431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978581061478989,
+      "loss": 0.5264,
+      "step": 432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009784496012816572,
+      "loss": 0.5303,
+      "step": 433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009783177477793552,
+      "loss": 0.5022,
+      "step": 434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978185501080472,
+      "loss": 0.5068,
+      "step": 435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009780528612937192,
+      "loss": 0.5173,
+      "step": 436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779198285281327,
+      "loss": 0.5042,
+      "step": 437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009777864028930705,
+      "loss": 0.5494,
+      "step": 438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009776525844982138,
+      "loss": 0.5214,
+      "step": 439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977518373453567,
+      "loss": 0.5225,
+      "step": 440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977383769869457,
+      "loss": 0.5031,
+      "step": 441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009772487738565333,
+      "loss": 0.5427,
+      "step": 442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009771133855257684,
+      "loss": 0.5754,
+      "step": 443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009769776049884564,
+      "loss": 0.5266,
+      "step": 444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009768414323562146,
+      "loss": 0.5234,
+      "step": 445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000976704867740983,
+      "loss": 0.5044,
+      "step": 446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009765679112550226,
+      "loss": 0.5229,
+      "step": 447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009764305630109173,
+      "loss": 0.5453,
+      "step": 448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762928231215731,
+      "loss": 0.528,
+      "step": 449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009761546917002177,
+      "loss": 0.4828,
+      "step": 450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760161688604008,
+      "loss": 0.5159,
+      "step": 451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758772547159935,
+      "loss": 0.5338,
+      "step": 452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009757379493811892,
+      "loss": 0.5392,
+      "step": 453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009755982529705024,
+      "loss": 0.5332,
+      "step": 454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754581655987692,
+      "loss": 0.5314,
+      "step": 455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753176873811471,
+      "loss": 0.558,
+      "step": 456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009751768184331149,
+      "loss": 0.532,
+      "step": 457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750355588704727,
+      "loss": 0.5221,
+      "step": 458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009748939088093414,
+      "loss": 0.5426,
+      "step": 459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009747518683661631,
+      "loss": 0.5431,
+      "step": 460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009746094376577011,
+      "loss": 0.5206,
+      "step": 461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744666168010387,
+      "loss": 0.5355,
+      "step": 462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000974323405913581,
+      "loss": 0.5923,
+      "step": 463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009741798051130527,
+      "loss": 0.5483,
+      "step": 464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009740358145174998,
+      "loss": 0.5212,
+      "step": 465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009738914342452881,
+      "loss": 0.4742,
+      "step": 466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737466644151044,
+      "loss": 0.473,
+      "step": 467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009736015051459551,
+      "loss": 0.5521,
+      "step": 468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000973455956557167,
+      "loss": 0.5167,
+      "step": 469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009733100187683873,
+      "loss": 0.5374,
+      "step": 470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009731636918995821,
+      "loss": 0.5189,
+      "step": 471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730169760710386,
+      "loss": 0.5409,
+      "step": 472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000972869871403363,
+      "loss": 0.5157,
+      "step": 473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009727223780174813,
+      "loss": 0.5306,
+      "step": 474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000972574496034639,
+      "loss": 0.505,
+      "step": 475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724262255764012,
+      "loss": 0.5393,
+      "step": 476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722775667646519,
+      "loss": 0.5317,
+      "step": 477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009721285197215952,
+      "loss": 0.5289,
+      "step": 478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009719790845697534,
+      "loss": 0.5161,
+      "step": 479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009718292614319684,
+      "loss": 0.5273,
+      "step": 480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009716790504314008,
+      "loss": 0.5114,
+      "step": 481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009715284516915302,
+      "loss": 0.5032,
+      "step": 482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009713774653361549,
+      "loss": 0.4878,
+      "step": 483
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009712260914893917,
+      "loss": 0.5171,
+      "step": 484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009710743302756764,
+      "loss": 0.489,
+      "step": 485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009709221818197624,
+      "loss": 0.5399,
+      "step": 486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009707696462467221,
+      "loss": 0.5262,
+      "step": 487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009706167236819458,
+      "loss": 0.5216,
+      "step": 488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009704634142511424,
+      "loss": 0.5188,
+      "step": 489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000970309718080338,
+      "loss": 0.5352,
+      "step": 490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009701556352958773,
+      "loss": 0.5371,
+      "step": 491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009700011660244228,
+      "loss": 0.5159,
+      "step": 492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009698463103929542,
+      "loss": 0.5332,
+      "step": 493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696910685287692,
+      "loss": 0.5258,
+      "step": 494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969535440559483,
+      "loss": 0.5416,
+      "step": 495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969379426613028,
+      "loss": 0.5187,
+      "step": 496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692230268176541,
+      "loss": 0.4827,
+      "step": 497
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969066241301928,
+      "loss": 0.5173,
+      "step": 498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000968909070194734,
+      "loss": 0.5234,
+      "step": 499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009687515136252732,
+      "loss": 0.4849,
+      "step": 500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009685935717230632,
+      "loss": 0.5176,
+      "step": 501
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009684352446179388,
+      "loss": 0.5071,
+      "step": 502
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682765324400514,
+      "loss": 0.493,
+      "step": 503
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009681174353198686,
+      "loss": 0.4767,
+      "step": 504
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009679579533881749,
+      "loss": 0.5178,
+      "step": 505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967798086776071,
+      "loss": 0.5239,
+      "step": 506
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009676378356149733,
+      "loss": 0.5051,
+      "step": 507
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009674772000366151,
+      "loss": 0.5255,
+      "step": 508
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009673161801730454,
+      "loss": 0.5089,
+      "step": 509
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009671547761566288,
+      "loss": 0.4839,
+      "step": 510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009669929881200461,
+      "loss": 0.5382,
+      "step": 511
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009668308161962937,
+      "loss": 0.5569,
+      "step": 512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009666682605186834,
+      "loss": 0.5171,
+      "step": 513
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009665053212208426,
+      "loss": 0.5184,
+      "step": 514
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009663419984367138,
+      "loss": 0.5306,
+      "step": 515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009661782923005553,
+      "loss": 0.5104,
+      "step": 516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009660142029469399,
+      "loss": 0.5298,
+      "step": 517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009658497305107559,
+      "loss": 0.5397,
+      "step": 518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009656848751272061,
+      "loss": 0.5011,
+      "step": 519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655196369318086,
+      "loss": 0.4955,
+      "step": 520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009653540160603955,
+      "loss": 0.5286,
+      "step": 521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009651880126491142,
+      "loss": 0.5353,
+      "step": 522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009650216268344262,
+      "loss": 0.4919,
+      "step": 523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009648548587531072,
+      "loss": 0.4941,
+      "step": 524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009646877085422476,
+      "loss": 0.5038,
+      "step": 525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009645201763392512,
+      "loss": 0.5261,
+      "step": 526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009643522622818368,
+      "loss": 0.4869,
+      "step": 527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009641839665080363,
+      "loss": 0.5155,
+      "step": 528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009640152891561957,
+      "loss": 0.5667,
+      "step": 529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009638462303649745,
+      "loss": 0.511,
+      "step": 530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009636767902733459,
+      "loss": 0.4794,
+      "step": 531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009635069690205967,
+      "loss": 0.5071,
+      "step": 532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009633367667463267,
+      "loss": 0.4893,
+      "step": 533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963166183590449,
+      "loss": 0.5184,
+      "step": 534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009629952196931902,
+      "loss": 0.5452,
+      "step": 535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009628238751950893,
+      "loss": 0.4708,
+      "step": 536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009626521502369983,
+      "loss": 0.4817,
+      "step": 537
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009624800449600825,
+      "loss": 0.48,
+      "step": 538
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009623075595058191,
+      "loss": 0.5382,
+      "step": 539
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621346940159981,
+      "loss": 0.5378,
+      "step": 540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009619614486327219,
+      "loss": 0.5155,
+      "step": 541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009617878234984055,
+      "loss": 0.5006,
+      "step": 542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009616138187557757,
+      "loss": 0.4973,
+      "step": 543
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009614394345478714,
+      "loss": 0.5353,
+      "step": 544
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009612646710180436,
+      "loss": 0.5009,
+      "step": 545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009610895283099547,
+      "loss": 0.5194,
+      "step": 546
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009609140065675797,
+      "loss": 0.5322,
+      "step": 547
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009607381059352038,
+      "loss": 0.4886,
+      "step": 548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605618265574251,
+      "loss": 0.4922,
+      "step": 549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009603851685791521,
+      "loss": 0.499,
+      "step": 550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009602081321456048,
+      "loss": 0.5127,
+      "step": 551
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009600307174023145,
+      "loss": 0.5286,
+      "step": 552
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009598529244951232,
+      "loss": 0.4872,
+      "step": 553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009596747535701839,
+      "loss": 0.5349,
+      "step": 554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009594962047739603,
+      "loss": 0.5319,
+      "step": 555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009593172782532268,
+      "loss": 0.5082,
+      "step": 556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009591379741550681,
+      "loss": 0.4963,
+      "step": 557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009589582926268797,
+      "loss": 0.5325,
+      "step": 558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009587782338163668,
+      "loss": 0.4984,
+      "step": 559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009585977978715455,
+      "loss": 0.4813,
+      "step": 560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000958416984940741,
+      "loss": 0.5017,
+      "step": 561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000958235795172589,
+      "loss": 0.5233,
+      "step": 562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009580542287160348,
+      "loss": 0.5278,
+      "step": 563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009578722857203336,
+      "loss": 0.4983,
+      "step": 564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009576899663350496,
+      "loss": 0.5233,
+      "step": 565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000957507270710057,
+      "loss": 0.5994,
+      "step": 566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000957324198995539,
+      "loss": 0.5146,
+      "step": 567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009571407513419877,
+      "loss": 0.4987,
+      "step": 568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009569569279002048,
+      "loss": 0.4911,
+      "step": 569
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009567727288213005,
+      "loss": 0.5209,
+      "step": 570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009565881542566939,
+      "loss": 0.5073,
+      "step": 571
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009564032043581128,
+      "loss": 0.4871,
+      "step": 572
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009562178792775935,
+      "loss": 0.5065,
+      "step": 573
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009560321791674807,
+      "loss": 0.476,
+      "step": 574
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009558461041804275,
+      "loss": 0.5188,
+      "step": 575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009556596544693951,
+      "loss": 0.5267,
+      "step": 576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009554728301876525,
+      "loss": 0.5032,
+      "step": 577
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009552856314887772,
+      "loss": 0.4957,
+      "step": 578
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009550980585266537,
+      "loss": 0.5028,
+      "step": 579
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009549101114554751,
+      "loss": 0.5411,
+      "step": 580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954721790429741,
+      "loss": 0.5277,
+      "step": 581
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009545330956042592,
+      "loss": 0.5042,
+      "step": 582
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009543440271341444,
+      "loss": 0.5123,
+      "step": 583
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009541545851748186,
+      "loss": 0.5287,
+      "step": 584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009539647698820108,
+      "loss": 0.5365,
+      "step": 585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009537745814117566,
+      "loss": 0.5686,
+      "step": 586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009535840199203991,
+      "loss": 0.5068,
+      "step": 587
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009533930855645872,
+      "loss": 0.5175,
+      "step": 588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009532017785012768,
+      "loss": 0.5095,
+      "step": 589
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009530100988877302,
+      "loss": 0.4838,
+      "step": 590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009528180468815154,
+      "loss": 0.5104,
+      "step": 591
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009526256226405074,
+      "loss": 0.5177,
+      "step": 592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009524328263228865,
+      "loss": 0.4921,
+      "step": 593
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000952239658087139,
+      "loss": 0.5201,
+      "step": 594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009520461180920574,
+      "loss": 0.5668,
+      "step": 595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009518522064967391,
+      "loss": 0.5284,
+      "step": 596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009516579234605874,
+      "loss": 0.5143,
+      "step": 597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009514632691433108,
+      "loss": 0.5115,
+      "step": 598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009512682437049228,
+      "loss": 0.4838,
+      "step": 599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009510728473057426,
+      "loss": 0.5096,
+      "step": 600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508770801063939,
+      "loss": 0.5177,
+      "step": 601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009506809422678051,
+      "loss": 0.5111,
+      "step": 602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009504844339512095,
+      "loss": 0.4872,
+      "step": 603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009502875553181452,
+      "loss": 0.5188,
+      "step": 604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009500903065304539,
+      "loss": 0.505,
+      "step": 605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009498926877502824,
+      "loss": 0.5128,
+      "step": 606
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009496946991400815,
+      "loss": 0.5055,
+      "step": 607
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009494963408626056,
+      "loss": 0.5063,
+      "step": 608
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009492976130809134,
+      "loss": 0.5017,
+      "step": 609
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009490985159583671,
+      "loss": 0.5265,
+      "step": 610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009488990496586325,
+      "loss": 0.5054,
+      "step": 611
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009486992143456792,
+      "loss": 0.4988,
+      "step": 612
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009484990101837798,
+      "loss": 0.5614,
+      "step": 613
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009482984373375104,
+      "loss": 0.4806,
+      "step": 614
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009480974959717498,
+      "loss": 0.5256,
+      "step": 615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009478961862516801,
+      "loss": 0.5134,
+      "step": 616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009476945083427862,
+      "loss": 0.5168,
+      "step": 617
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009474924624108549,
+      "loss": 0.5334,
+      "step": 618
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009472900486219768,
+      "loss": 0.504,
+      "step": 619
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000947087267142544,
+      "loss": 0.4819,
+      "step": 620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009468841181392511,
+      "loss": 0.5284,
+      "step": 621
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009466806017790947,
+      "loss": 0.5286,
+      "step": 622
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009464767182293739,
+      "loss": 0.5145,
+      "step": 623
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009462724676576887,
+      "loss": 0.5159,
+      "step": 624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009460678502319418,
+      "loss": 0.531,
+      "step": 625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009458628661203367,
+      "loss": 0.5262,
+      "step": 626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009456575154913788,
+      "loss": 0.5026,
+      "step": 627
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009454517985138746,
+      "loss": 0.5277,
+      "step": 628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009452457153569319,
+      "loss": 0.5018,
+      "step": 629
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009450392661899593,
+      "loss": 0.5228,
+      "step": 630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009448324511826665,
+      "loss": 0.52,
+      "step": 631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009446252705050636,
+      "loss": 0.4987,
+      "step": 632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009444177243274617,
+      "loss": 0.5212,
+      "step": 633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009442098128204723,
+      "loss": 0.4879,
+      "step": 634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009440015361550068,
+      "loss": 0.4979,
+      "step": 635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009437928945022771,
+      "loss": 0.5133,
+      "step": 636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009435838880337953,
+      "loss": 0.5133,
+      "step": 637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009433745169213729,
+      "loss": 0.5294,
+      "step": 638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009431647813371218,
+      "loss": 0.4706,
+      "step": 639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009429546814534529,
+      "loss": 0.553,
+      "step": 640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009427442174430768,
+      "loss": 0.5562,
+      "step": 641
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009425333894790036,
+      "loss": 0.5043,
+      "step": 642
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009423221977345424,
+      "loss": 0.582,
+      "step": 643
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009421106423833013,
+      "loss": 0.4949,
+      "step": 644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009418987235991876,
+      "loss": 0.4828,
+      "step": 645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009416864415564068,
+      "loss": 0.5082,
+      "step": 646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009414737964294635,
+      "loss": 0.5028,
+      "step": 647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009412607883931606,
+      "loss": 0.524,
+      "step": 648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009410474176225995,
+      "loss": 0.4802,
+      "step": 649
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009408336842931793,
+      "loss": 0.5112,
+      "step": 650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009406195885805978,
+      "loss": 0.5245,
+      "step": 651
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009404051306608501,
+      "loss": 0.5048,
+      "step": 652
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009401903107102295,
+      "loss": 0.4705,
+      "step": 653
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009399751289053266,
+      "loss": 0.5017,
+      "step": 654
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009397595854230297,
+      "loss": 0.5193,
+      "step": 655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000939543680440524,
+      "loss": 0.4938,
+      "step": 656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009393274141352924,
+      "loss": 0.4768,
+      "step": 657
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009391107866851142,
+      "loss": 0.4832,
+      "step": 658
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009388937982680666,
+      "loss": 0.5044,
+      "step": 659
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009386764490625224,
+      "loss": 0.5403,
+      "step": 660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009384587392471515,
+      "loss": 0.5339,
+      "step": 661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009382406690009203,
+      "loss": 0.5112,
+      "step": 662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009380222385030915,
+      "loss": 0.4614,
+      "step": 663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009378034479332234,
+      "loss": 0.4584,
+      "step": 664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009375842974711713,
+      "loss": 0.4816,
+      "step": 665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009373647872970852,
+      "loss": 0.5176,
+      "step": 666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009371449175914117,
+      "loss": 0.5367,
+      "step": 667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009369246885348925,
+      "loss": 0.5066,
+      "step": 668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009367041003085648,
+      "loss": 0.5175,
+      "step": 669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009364831530937613,
+      "loss": 0.494,
+      "step": 670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009362618470721093,
+      "loss": 0.4819,
+      "step": 671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009360401824255313,
+      "loss": 0.4836,
+      "step": 672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009358181593362449,
+      "loss": 0.5354,
+      "step": 673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009355957779867618,
+      "loss": 0.524,
+      "step": 674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009353730385598887,
+      "loss": 0.5483,
+      "step": 675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009351499412387262,
+      "loss": 0.5061,
+      "step": 676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009349264862066695,
+      "loss": 0.5356,
+      "step": 677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009347026736474076,
+      "loss": 0.5394,
+      "step": 678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009344785037449238,
+      "loss": 0.5406,
+      "step": 679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009342539766834945,
+      "loss": 0.4841,
+      "step": 680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009340290926476901,
+      "loss": 0.5193,
+      "step": 681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009338038518223745,
+      "loss": 0.4961,
+      "step": 682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009335782543927049,
+      "loss": 0.5081,
+      "step": 683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009333523005441313,
+      "loss": 0.512,
+      "step": 684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000933125990462397,
+      "loss": 0.5056,
+      "step": 685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000932899324333538,
+      "loss": 0.5231,
+      "step": 686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009326723023438833,
+      "loss": 0.4917,
+      "step": 687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009324449246800537,
+      "loss": 0.5275,
+      "step": 688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009322171915289634,
+      "loss": 0.4974,
+      "step": 689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000931989103077818,
+      "loss": 0.5347,
+      "step": 690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009317606595141155,
+      "loss": 0.5243,
+      "step": 691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009315318610256457,
+      "loss": 0.5055,
+      "step": 692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009313027078004903,
+      "loss": 0.496,
+      "step": 693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009310732000270226,
+      "loss": 0.5065,
+      "step": 694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009308433378939073,
+      "loss": 0.5179,
+      "step": 695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009306131215901003,
+      "loss": 0.4825,
+      "step": 696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009303825513048488,
+      "loss": 0.4989,
+      "step": 697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009301516272276907,
+      "loss": 0.4905,
+      "step": 698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009299203495484554,
+      "loss": 0.5175,
+      "step": 699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009296887184572621,
+      "loss": 0.5176,
+      "step": 700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009294567341445213,
+      "loss": 0.5135,
+      "step": 701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000929224396800933,
+      "loss": 0.5095,
+      "step": 702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009289917066174886,
+      "loss": 0.5522,
+      "step": 703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009287586637854683,
+      "loss": 0.4863,
+      "step": 704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000928525268496443,
+      "loss": 0.5289,
+      "step": 705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009282915209422731,
+      "loss": 0.4744,
+      "step": 706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009280574213151085,
+      "loss": 0.5474,
+      "step": 707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009278229698073888,
+      "loss": 0.5128,
+      "step": 708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009275881666118422,
+      "loss": 0.4929,
+      "step": 709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009273530119214868,
+      "loss": 0.5396,
+      "step": 710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009271175059296292,
+      "loss": 0.4999,
+      "step": 711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009268816488298648,
+      "loss": 0.463,
+      "step": 712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009266454408160778,
+      "loss": 0.4907,
+      "step": 713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009264088820824408,
+      "loss": 0.4731,
+      "step": 714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009261719728234145,
+      "loss": 0.5319,
+      "step": 715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009259347132337481,
+      "loss": 0.4954,
+      "step": 716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009256971035084784,
+      "loss": 0.51,
+      "step": 717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009254591438429305,
+      "loss": 0.5139,
+      "step": 718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009252208344327166,
+      "loss": 0.4717,
+      "step": 719
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009249821754737369,
+      "loss": 0.5156,
+      "step": 720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009247431671621786,
+      "loss": 0.5034,
+      "step": 721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009245038096945164,
+      "loss": 0.5186,
+      "step": 722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009242641032675118,
+      "loss": 0.4823,
+      "step": 723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009240240480782129,
+      "loss": 0.533,
+      "step": 724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009237836443239554,
+      "loss": 0.5038,
+      "step": 725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009235428922023603,
+      "loss": 0.5499,
+      "step": 726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000923301791911336,
+      "loss": 0.4786,
+      "step": 727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009230603436490763,
+      "loss": 0.5209,
+      "step": 728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000922818547614062,
+      "loss": 0.4565,
+      "step": 729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000922576404005059,
+      "loss": 0.4624,
+      "step": 730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009223339130211192,
+      "loss": 0.5023,
+      "step": 731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009220910748615798,
+      "loss": 0.5398,
+      "step": 732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000921847889726064,
+      "loss": 0.5044,
+      "step": 733
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009216043578144795,
+      "loss": 0.5021,
+      "step": 734
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009213604793270196,
+      "loss": 0.4713,
+      "step": 735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009211162544641622,
+      "loss": 0.5023,
+      "step": 736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009208716834266702,
+      "loss": 0.4987,
+      "step": 737
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009206267664155906,
+      "loss": 0.5237,
+      "step": 738
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009203815036322553,
+      "loss": 0.49,
+      "step": 739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009201358952782804,
+      "loss": 0.4908,
+      "step": 740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009198899415555658,
+      "loss": 0.4862,
+      "step": 741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009196436426662953,
+      "loss": 0.5094,
+      "step": 742
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009193969988129367,
+      "loss": 0.5163,
+      "step": 743
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009191500101982413,
+      "loss": 0.4856,
+      "step": 744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009189026770252437,
+      "loss": 0.4968,
+      "step": 745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009186549994972617,
+      "loss": 0.535,
+      "step": 746
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009184069778178964,
+      "loss": 0.527,
+      "step": 747
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009181586121910317,
+      "loss": 0.5,
+      "step": 748
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000917909902820834,
+      "loss": 0.5314,
+      "step": 749
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009176608499117527,
+      "loss": 0.5297,
+      "step": 750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009174114536685193,
+      "loss": 0.4718,
+      "step": 751
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009171617142961477,
+      "loss": 0.5046,
+      "step": 752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009169116319999336,
+      "loss": 0.4827,
+      "step": 753
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000916661206985455,
+      "loss": 0.5149,
+      "step": 754
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009164104394585713,
+      "loss": 0.4952,
+      "step": 755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009161593296254236,
+      "loss": 0.4723,
+      "step": 756
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009159078776924346,
+      "loss": 0.5171,
+      "step": 757
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009156560838663075,
+      "loss": 0.5049,
+      "step": 758
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009154039483540273,
+      "loss": 0.5093,
+      "step": 759
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009151514713628597,
+      "loss": 0.4307,
+      "step": 760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000914898653100351,
+      "loss": 0.5046,
+      "step": 761
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009146454937743278,
+      "loss": 0.4891,
+      "step": 762
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009143919935928975,
+      "loss": 0.527,
+      "step": 763
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009141381527644477,
+      "loss": 0.5433,
+      "step": 764
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009138839714976456,
+      "loss": 0.5121,
+      "step": 765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009136294500014386,
+      "loss": 0.5166,
+      "step": 766
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009133745884850538,
+      "loss": 0.4821,
+      "step": 767
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009131193871579975,
+      "loss": 0.4951,
+      "step": 768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009128638462300556,
+      "loss": 0.5038,
+      "step": 769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009126079659112932,
+      "loss": 0.5039,
+      "step": 770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009123517464120542,
+      "loss": 0.4974,
+      "step": 771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009120951879429615,
+      "loss": 0.496,
+      "step": 772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009118382907149164,
+      "loss": 0.4863,
+      "step": 773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000911581054939099,
+      "loss": 0.5111,
+      "step": 774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009113234808269675,
+      "loss": 0.4808,
+      "step": 775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009110655685902583,
+      "loss": 0.5006,
+      "step": 776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009108073184409855,
+      "loss": 0.5154,
+      "step": 777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009105487305914415,
+      "loss": 0.488,
+      "step": 778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009102898052541958,
+      "loss": 0.5087,
+      "step": 779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009100305426420956,
+      "loss": 0.5328,
+      "step": 780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009097709429682653,
+      "loss": 0.5316,
+      "step": 781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009095110064461062,
+      "loss": 0.5438,
+      "step": 782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009092507332892967,
+      "loss": 0.5457,
+      "step": 783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009089901237117921,
+      "loss": 0.4967,
+      "step": 784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000908729177927824,
+      "loss": 0.5071,
+      "step": 785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009084678961519,
+      "loss": 0.5355,
+      "step": 786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009082062785988049,
+      "loss": 0.5396,
+      "step": 787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009079443254835987,
+      "loss": 0.5096,
+      "step": 788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009076820370216173,
+      "loss": 0.5004,
+      "step": 789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009074194134284726,
+      "loss": 0.4606,
+      "step": 790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009071564549200517,
+      "loss": 0.5085,
+      "step": 791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009068931617125174,
+      "loss": 0.5325,
+      "step": 792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009066295340223073,
+      "loss": 0.5063,
+      "step": 793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009063655720661341,
+      "loss": 0.5259,
+      "step": 794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000906101276060985,
+      "loss": 0.4908,
+      "step": 795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009058366462241223,
+      "loss": 0.563,
+      "step": 796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009055716827730823,
+      "loss": 0.4847,
+      "step": 797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009053063859256758,
+      "loss": 0.478,
+      "step": 798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009050407558999874,
+      "loss": 0.5096,
+      "step": 799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009047747929143759,
+      "loss": 0.5417,
+      "step": 800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009045084971874737,
+      "loss": 0.521,
+      "step": 801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009042418689381866,
+      "loss": 0.5203,
+      "step": 802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009039749083856937,
+      "loss": 0.5094,
+      "step": 803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009037076157494478,
+      "loss": 0.5243,
+      "step": 804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009034399912491738,
+      "loss": 0.4786,
+      "step": 805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00090317203510487,
+      "loss": 0.4596,
+      "step": 806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009029037475368075,
+      "loss": 0.5023,
+      "step": 807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009026351287655293,
+      "loss": 0.5308,
+      "step": 808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009023661790118511,
+      "loss": 0.5062,
+      "step": 809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009020968984968603,
+      "loss": 0.5249,
+      "step": 810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009018272874419164,
+      "loss": 0.4763,
+      "step": 811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009015573460686509,
+      "loss": 0.4998,
+      "step": 812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009012870745989663,
+      "loss": 0.5088,
+      "step": 813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009010164732550368,
+      "loss": 0.4723,
+      "step": 814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009007455422593077,
+      "loss": 0.5016,
+      "step": 815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000900474281834495,
+      "loss": 0.5598,
+      "step": 816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009002026922035862,
+      "loss": 0.4932,
+      "step": 817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008999307735898387,
+      "loss": 0.5048,
+      "step": 818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008996585262167807,
+      "loss": 0.5074,
+      "step": 819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008993859503082106,
+      "loss": 0.4868,
+      "step": 820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008991130460881968,
+      "loss": 0.5638,
+      "step": 821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008988398137810777,
+      "loss": 0.5032,
+      "step": 822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008985662536114613,
+      "loss": 0.5183,
+      "step": 823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000898292365804225,
+      "loss": 0.5537,
+      "step": 824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000898018150584516,
+      "loss": 0.4928,
+      "step": 825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00089774360817775,
+      "loss": 0.4855,
+      "step": 826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008974687388096122,
+      "loss": 0.474,
+      "step": 827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008971935427060562,
+      "loss": 0.4669,
+      "step": 828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008969180200933046,
+      "loss": 0.5197,
+      "step": 829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008966421711978481,
+      "loss": 0.4879,
+      "step": 830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008963659962464455,
+      "loss": 0.5255,
+      "step": 831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008960894954661238,
+      "loss": 0.4619,
+      "step": 832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000895812669084178,
+      "loss": 0.5043,
+      "step": 833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008955355173281707,
+      "loss": 0.4995,
+      "step": 834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008952580404259316,
+      "loss": 0.4531,
+      "step": 835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008949802386055581,
+      "loss": 0.4982,
+      "step": 836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008947021120954146,
+      "loss": 0.5133,
+      "step": 837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008944236611241323,
+      "loss": 0.5205,
+      "step": 838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008941448859206093,
+      "loss": 0.5464,
+      "step": 839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008938657867140101,
+      "loss": 0.4978,
+      "step": 840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008935863637337654,
+      "loss": 0.5018,
+      "step": 841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008933066172095721,
+      "loss": 0.519,
+      "step": 842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008930265473713938,
+      "loss": 0.5034,
+      "step": 843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008927461544494586,
+      "loss": 0.5127,
+      "step": 844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008924654386742613,
+      "loss": 0.5166,
+      "step": 845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008921844002765613,
+      "loss": 0.4707,
+      "step": 846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008919030394873839,
+      "loss": 0.5234,
+      "step": 847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008916213565380188,
+      "loss": 0.516,
+      "step": 848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008913393516600209,
+      "loss": 0.4706,
+      "step": 849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008910570250852097,
+      "loss": 0.4866,
+      "step": 850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000890774377045669,
+      "loss": 0.5381,
+      "step": 851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008904914077737469,
+      "loss": 0.5015,
+      "step": 852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008902081175020558,
+      "loss": 0.5287,
+      "step": 853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008899245064634718,
+      "loss": 0.4823,
+      "step": 854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008896405748911345,
+      "loss": 0.53,
+      "step": 855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008893563230184471,
+      "loss": 0.4636,
+      "step": 856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008890717510790764,
+      "loss": 0.4846,
+      "step": 857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008887868593069519,
+      "loss": 0.491,
+      "step": 858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008885016479362661,
+      "loss": 0.4826,
+      "step": 859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008882161172014748,
+      "loss": 0.4973,
+      "step": 860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008879302673372953,
+      "loss": 0.4883,
+      "step": 861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000887644098578708,
+      "loss": 0.5154,
+      "step": 862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008873576111609553,
+      "loss": 0.4878,
+      "step": 863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008870708053195413,
+      "loss": 0.4662,
+      "step": 864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008867836812902322,
+      "loss": 0.481,
+      "step": 865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008864962393090552,
+      "loss": 0.4862,
+      "step": 866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008862084796122997,
+      "loss": 0.519,
+      "step": 867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008859204024365159,
+      "loss": 0.5164,
+      "step": 868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008856320080185145,
+      "loss": 0.5292,
+      "step": 869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008853432965953676,
+      "loss": 0.4935,
+      "step": 870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008850542684044079,
+      "loss": 0.5162,
+      "step": 871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000884764923683228,
+      "loss": 0.4888,
+      "step": 872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008844752626696811,
+      "loss": 0.5514,
+      "step": 873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008841852856018802,
+      "loss": 0.5106,
+      "step": 874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008838949927181984,
+      "loss": 0.5087,
+      "step": 875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008836043842572681,
+      "loss": 0.4993,
+      "step": 876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008833134604579814,
+      "loss": 0.4762,
+      "step": 877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000883022221559489,
+      "loss": 0.5065,
+      "step": 878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008827306678012016,
+      "loss": 0.5004,
+      "step": 879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000882438799422788,
+      "loss": 0.5073,
+      "step": 880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008821466166641759,
+      "loss": 0.4916,
+      "step": 881
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008818541197655512,
+      "loss": 0.499,
+      "step": 882
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008815613089673583,
+      "loss": 0.4862,
+      "step": 883
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008812681845102997,
+      "loss": 0.5134,
+      "step": 884
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008809747466353355,
+      "loss": 0.493,
+      "step": 885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008806809955836834,
+      "loss": 0.5613,
+      "step": 886
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008803869315968187,
+      "loss": 0.4722,
+      "step": 887
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008800925549164741,
+      "loss": 0.4572,
+      "step": 888
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008797978657846391,
+      "loss": 0.4894,
+      "step": 889
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00087950286444356,
+      "loss": 0.4853,
+      "step": 890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00087920755113574,
+      "loss": 0.507,
+      "step": 891
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008789119261039385,
+      "loss": 0.512,
+      "step": 892
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008786159895911712,
+      "loss": 0.4849,
+      "step": 893
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008783197418407101,
+      "loss": 0.4757,
+      "step": 894
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008780231830960828,
+      "loss": 0.4694,
+      "step": 895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008777263136010727,
+      "loss": 0.483,
+      "step": 896
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008774291335997182,
+      "loss": 0.5293,
+      "step": 897
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008771316433363138,
+      "loss": 0.4733,
+      "step": 898
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008768338430554083,
+      "loss": 0.4513,
+      "step": 899
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008765357330018055,
+      "loss": 0.4933,
+      "step": 900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008762373134205641,
+      "loss": 0.4897,
+      "step": 901
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008759385845569971,
+      "loss": 0.4989,
+      "step": 902
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008756395466566717,
+      "loss": 0.5148,
+      "step": 903
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008753401999654093,
+      "loss": 0.5078,
+      "step": 904
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008750405447292849,
+      "loss": 0.4539,
+      "step": 905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008747405811946271,
+      "loss": 0.493,
+      "step": 906
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008744403096080182,
+      "loss": 0.53,
+      "step": 907
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008741397302162939,
+      "loss": 0.4922,
+      "step": 908
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008738388432665423,
+      "loss": 0.4808,
+      "step": 909
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008735376490061047,
+      "loss": 0.5074,
+      "step": 910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000873236147682575,
+      "loss": 0.4749,
+      "step": 911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008729343395437999,
+      "loss": 0.5148,
+      "step": 912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008726322248378774,
+      "loss": 0.4784,
+      "step": 913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008723298038131582,
+      "loss": 0.4694,
+      "step": 914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008720270767182448,
+      "loss": 0.4802,
+      "step": 915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008717240438019912,
+      "loss": 0.5068,
+      "step": 916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008714207053135023,
+      "loss": 0.4995,
+      "step": 917
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000871117061502135,
+      "loss": 0.5046,
+      "step": 918
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008708131126174967,
+      "loss": 0.488,
+      "step": 919
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008705088589094458,
+      "loss": 0.4993,
+      "step": 920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008702043006280912,
+      "loss": 0.4933,
+      "step": 921
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000869899438023792,
+      "loss": 0.499,
+      "step": 922
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008695942713471578,
+      "loss": 0.5195,
+      "step": 923
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008692888008490478,
+      "loss": 0.4932,
+      "step": 924
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008689830267805712,
+      "loss": 0.5199,
+      "step": 925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008686769493930868,
+      "loss": 0.4921,
+      "step": 926
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008683705689382025,
+      "loss": 0.4944,
+      "step": 927
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008680638856677754,
+      "loss": 0.4811,
+      "step": 928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008677568998339118,
+      "loss": 0.5107,
+      "step": 929
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008674496116889663,
+      "loss": 0.4811,
+      "step": 930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008671420214855422,
+      "loss": 0.5322,
+      "step": 931
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008668341294764913,
+      "loss": 0.4668,
+      "step": 932
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008665259359149131,
+      "loss": 0.4424,
+      "step": 933
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008662174410541554,
+      "loss": 0.517,
+      "step": 934
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008659086451478135,
+      "loss": 0.4923,
+      "step": 935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008655995484497298,
+      "loss": 0.4624,
+      "step": 936
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008652901512139949,
+      "loss": 0.5204,
+      "step": 937
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008649804536949452,
+      "loss": 0.5112,
+      "step": 938
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008646704561471652,
+      "loss": 0.5164,
+      "step": 939
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008643601588254852,
+      "loss": 0.5337,
+      "step": 940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008640495619849821,
+      "loss": 0.4967,
+      "step": 941
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008637386658809791,
+      "loss": 0.5205,
+      "step": 942
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008634274707690458,
+      "loss": 0.4847,
+      "step": 943
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008631159769049966,
+      "loss": 0.5254,
+      "step": 944
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008628041845448924,
+      "loss": 0.4895,
+      "step": 945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008624920939450392,
+      "loss": 0.5133,
+      "step": 946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008621797053619881,
+      "loss": 0.5038,
+      "step": 947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000861867019052535,
+      "loss": 0.4875,
+      "step": 948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000861554035273721,
+      "loss": 0.535,
+      "step": 949
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008612407542828313,
+      "loss": 0.5127,
+      "step": 950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008609271763373956,
+      "loss": 0.4929,
+      "step": 951
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008606133016951877,
+      "loss": 0.5015,
+      "step": 952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008602991306142252,
+      "loss": 0.4933,
+      "step": 953
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008599846633527696,
+      "loss": 0.4965,
+      "step": 954
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008596699001693256,
+      "loss": 0.4535,
+      "step": 955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008593548413226414,
+      "loss": 0.4884,
+      "step": 956
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008590394870717081,
+      "loss": 0.4989,
+      "step": 957
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008587238376757596,
+      "loss": 0.512,
+      "step": 958
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008584078933942727,
+      "loss": 0.4801,
+      "step": 959
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008580916544869663,
+      "loss": 0.4843,
+      "step": 960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008577751212138017,
+      "loss": 0.4895,
+      "step": 961
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008574582938349817,
+      "loss": 0.4663,
+      "step": 962
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008571411726109519,
+      "loss": 0.5067,
+      "step": 963
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008568237578023981,
+      "loss": 0.4821,
+      "step": 964
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008565060496702489,
+      "loss": 0.4968,
+      "step": 965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008561880484756724,
+      "loss": 0.4934,
+      "step": 966
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008558697544800791,
+      "loss": 0.5052,
+      "step": 967
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008555511679451196,
+      "loss": 0.4976,
+      "step": 968
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008552322891326845,
+      "loss": 0.4779,
+      "step": 969
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008549131183049053,
+      "loss": 0.4958,
+      "step": 970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008545936557241536,
+      "loss": 0.5161,
+      "step": 971
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008542739016530403,
+      "loss": 0.5288,
+      "step": 972
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008539538563544164,
+      "loss": 0.4933,
+      "step": 973
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000853633520091372,
+      "loss": 0.5435,
+      "step": 974
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008533128931272365,
+      "loss": 0.4623,
+      "step": 975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008529919757255782,
+      "loss": 0.5115,
+      "step": 976
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008526707681502044,
+      "loss": 0.4917,
+      "step": 977
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008523492706651607,
+      "loss": 0.4554,
+      "step": 978
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000852027483534731,
+      "loss": 0.5066,
+      "step": 979
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008517054070234373,
+      "loss": 0.4884,
+      "step": 980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008513830413960399,
+      "loss": 0.5237,
+      "step": 981
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008510603869175358,
+      "loss": 0.5055,
+      "step": 982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008507374438531607,
+      "loss": 0.451,
+      "step": 983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008504142124683865,
+      "loss": 0.5087,
+      "step": 984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008500906930289226,
+      "loss": 0.4977,
+      "step": 985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008497668858007151,
+      "loss": 0.5089,
+      "step": 986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008494427910499467,
+      "loss": 0.5071,
+      "step": 987
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008491184090430364,
+      "loss": 0.4897,
+      "step": 988
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008487937400466395,
+      "loss": 0.5243,
+      "step": 989
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008484687843276469,
+      "loss": 0.4588,
+      "step": 990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008481435421531851,
+      "loss": 0.4967,
+      "step": 991
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000847818013790617,
+      "loss": 0.5017,
+      "step": 992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008474921995075398,
+      "loss": 0.5216,
+      "step": 993
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008471660995717858,
+      "loss": 0.5219,
+      "step": 994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008468397142514228,
+      "loss": 0.4627,
+      "step": 995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008465130438147526,
+      "loss": 0.5481,
+      "step": 996
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008461860885303114,
+      "loss": 0.4899,
+      "step": 997
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008458588486668699,
+      "loss": 0.5491,
+      "step": 998
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008455313244934324,
+      "loss": 0.4872,
+      "step": 999
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008452035162792371,
+      "loss": 0.4591,
+      "step": 1000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008448754242937557,
+      "loss": 0.5466,
+      "step": 1001
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008445470488066929,
+      "loss": 0.5194,
+      "step": 1002
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000844218390087987,
+      "loss": 0.4863,
+      "step": 1003
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008438894484078086,
+      "loss": 0.474,
+      "step": 1004
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008435602240365609,
+      "loss": 0.5001,
+      "step": 1005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008432307172448801,
+      "loss": 0.4886,
+      "step": 1006
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008429009283036339,
+      "loss": 0.524,
+      "step": 1007
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008425708574839221,
+      "loss": 0.4951,
+      "step": 1008
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008422405050570765,
+      "loss": 0.4873,
+      "step": 1009
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00084190987129466,
+      "loss": 0.4775,
+      "step": 1010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008415789564684673,
+      "loss": 0.4991,
+      "step": 1011
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008412477608505234,
+      "loss": 0.5079,
+      "step": 1012
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008409162847130846,
+      "loss": 0.4945,
+      "step": 1013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000840584528328638,
+      "loss": 0.5078,
+      "step": 1014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008402524919699005,
+      "loss": 0.4921,
+      "step": 1015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008399201759098197,
+      "loss": 0.5221,
+      "step": 1016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008395875804215725,
+      "loss": 0.4814,
+      "step": 1017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008392547057785661,
+      "loss": 0.4421,
+      "step": 1018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008389215522544371,
+      "loss": 0.4622,
+      "step": 1019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000838588120123051,
+      "loss": 0.4719,
+      "step": 1020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008382544096585026,
+      "loss": 0.5027,
+      "step": 1021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008379204211351154,
+      "loss": 0.5232,
+      "step": 1022
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008375861548274417,
+      "loss": 0.5022,
+      "step": 1023
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008372516110102617,
+      "loss": 0.5168,
+      "step": 1024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008369167899585841,
+      "loss": 0.4543,
+      "step": 1025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008365816919476452,
+      "loss": 0.4672,
+      "step": 1026
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008362463172529093,
+      "loss": 0.4753,
+      "step": 1027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008359106661500682,
+      "loss": 0.4653,
+      "step": 1028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008355747389150403,
+      "loss": 0.4874,
+      "step": 1029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008352385358239718,
+      "loss": 0.4879,
+      "step": 1030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008349020571532349,
+      "loss": 0.4597,
+      "step": 1031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008345653031794292,
+      "loss": 0.5063,
+      "step": 1032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008342282741793796,
+      "loss": 0.4467,
+      "step": 1033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008338909704301378,
+      "loss": 0.516,
+      "step": 1034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008335533922089813,
+      "loss": 0.502,
+      "step": 1035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008332155397934129,
+      "loss": 0.4985,
+      "step": 1036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008328774134611611,
+      "loss": 0.5491,
+      "step": 1037
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008325390134901793,
+      "loss": 0.4766,
+      "step": 1038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008322003401586462,
+      "loss": 0.4912,
+      "step": 1039
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008318613937449648,
+      "loss": 0.5157,
+      "step": 1040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000831522174527763,
+      "loss": 0.472,
+      "step": 1041
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008311826827858926,
+      "loss": 0.4988,
+      "step": 1042
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008308429187984297,
+      "loss": 0.4516,
+      "step": 1043
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000830502882844674,
+      "loss": 0.5123,
+      "step": 1044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008301625752041488,
+      "loss": 0.4602,
+      "step": 1045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008298219961566008,
+      "loss": 0.4879,
+      "step": 1046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008294811459819998,
+      "loss": 0.4949,
+      "step": 1047
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008291400249605385,
+      "loss": 0.5181,
+      "step": 1048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008287986333726322,
+      "loss": 0.4808,
+      "step": 1049
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008284569714989185,
+      "loss": 0.4449,
+      "step": 1050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008281150396202576,
+      "loss": 0.4241,
+      "step": 1051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000827772838017731,
+      "loss": 0.4768,
+      "step": 1052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008274303669726426,
+      "loss": 0.4685,
+      "step": 1053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008270876267665173,
+      "loss": 0.5129,
+      "step": 1054
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008267446176811014,
+      "loss": 0.4961,
+      "step": 1055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008264013399983625,
+      "loss": 0.4769,
+      "step": 1056
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008260577940004888,
+      "loss": 0.4712,
+      "step": 1057
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008257139799698886,
+      "loss": 0.4974,
+      "step": 1058
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008253698981891915,
+      "loss": 0.5021,
+      "step": 1059
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008250255489412463,
+      "loss": 0.498,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008246809325091223,
+      "loss": 0.5109,
+      "step": 1061
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008243360491761078,
+      "loss": 0.496,
+      "step": 1062
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008239908992257115,
+      "loss": 0.5107,
+      "step": 1063
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00082364548294166,
+      "loss": 0.4489,
+      "step": 1064
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008232998006078997,
+      "loss": 0.4929,
+      "step": 1065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008229538525085959,
+      "loss": 0.4926,
+      "step": 1066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008226076389281315,
+      "loss": 0.5262,
+      "step": 1067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008222611601511083,
+      "loss": 0.4526,
+      "step": 1068
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008219144164623458,
+      "loss": 0.4934,
+      "step": 1069
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008215674081468814,
+      "loss": 0.4857,
+      "step": 1070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00082122013548997,
+      "loss": 0.541,
+      "step": 1071
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000820872598777084,
+      "loss": 0.5073,
+      "step": 1072
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008205247982939122,
+      "loss": 0.4828,
+      "step": 1073
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008201767343263612,
+      "loss": 0.5043,
+      "step": 1074
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008198284071605534,
+      "loss": 0.4778,
+      "step": 1075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000819479817082828,
+      "loss": 0.4648,
+      "step": 1076
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00081913096437974,
+      "loss": 0.5195,
+      "step": 1077
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008187818493380607,
+      "loss": 0.5101,
+      "step": 1078
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008184324722447767,
+      "loss": 0.5052,
+      "step": 1079
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00081808283338709,
+      "loss": 0.4672,
+      "step": 1080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008177329330524181,
+      "loss": 0.4702,
+      "step": 1081
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008173827715283932,
+      "loss": 0.5072,
+      "step": 1082
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008170323491028624,
+      "loss": 0.4786,
+      "step": 1083
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000816681666063887,
+      "loss": 0.4792,
+      "step": 1084
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008163307226997425,
+      "loss": 0.5074,
+      "step": 1085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008159795192989189,
+      "loss": 0.4686,
+      "step": 1086
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008156280561501195,
+      "loss": 0.4811,
+      "step": 1087
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008152763335422613,
+      "loss": 0.4771,
+      "step": 1088
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008149243517644744,
+      "loss": 0.5021,
+      "step": 1089
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008145721111061025,
+      "loss": 0.4568,
+      "step": 1090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008142196118567009,
+      "loss": 0.4663,
+      "step": 1091
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008138668543060391,
+      "loss": 0.4961,
+      "step": 1092
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008135138387440977,
+      "loss": 0.4711,
+      "step": 1093
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00081316056546107,
+      "loss": 0.4983,
+      "step": 1094
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008128070347473608,
+      "loss": 0.5367,
+      "step": 1095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008124532468935866,
+      "loss": 0.5013,
+      "step": 1096
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008120992021905756,
+      "loss": 0.4979,
+      "step": 1097
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008117449009293668,
+      "loss": 0.5187,
+      "step": 1098
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008113903434012102,
+      "loss": 0.502,
+      "step": 1099
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008110355298975664,
+      "loss": 0.5024,
+      "step": 1100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008106804607101066,
+      "loss": 0.4815,
+      "step": 1101
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008103251361307119,
+      "loss": 0.5017,
+      "step": 1102
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008099695564514738,
+      "loss": 0.4991,
+      "step": 1103
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008096137219646928,
+      "loss": 0.485,
+      "step": 1104
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008092576329628796,
+      "loss": 0.4703,
+      "step": 1105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008089012897387537,
+      "loss": 0.5005,
+      "step": 1106
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008085446925852437,
+      "loss": 0.5162,
+      "step": 1107
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000808187841795487,
+      "loss": 0.4976,
+      "step": 1108
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008078307376628291,
+      "loss": 0.4581,
+      "step": 1109
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008074733804808245,
+      "loss": 0.4758,
+      "step": 1110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000807115770543235,
+      "loss": 0.4968,
+      "step": 1111
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008067579081440305,
+      "loss": 0.4385,
+      "step": 1112
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008063997935773884,
+      "loss": 0.5114,
+      "step": 1113
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008060414271376937,
+      "loss": 0.5298,
+      "step": 1114
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008056828091195376,
+      "loss": 0.4991,
+      "step": 1115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008053239398177191,
+      "loss": 0.4845,
+      "step": 1116
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008049648195272431,
+      "loss": 0.5125,
+      "step": 1117
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008046054485433212,
+      "loss": 0.4872,
+      "step": 1118
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008042458271613706,
+      "loss": 0.4781,
+      "step": 1119
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000803885955677015,
+      "loss": 0.4896,
+      "step": 1120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008035258343860832,
+      "loss": 0.4534,
+      "step": 1121
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008031654635846095,
+      "loss": 0.4958,
+      "step": 1122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008028048435688333,
+      "loss": 0.491,
+      "step": 1123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008024439746351991,
+      "loss": 0.5508,
+      "step": 1124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008020828570803553,
+      "loss": 0.4507,
+      "step": 1125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008017214912011556,
+      "loss": 0.4928,
+      "step": 1126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008013598772946575,
+      "loss": 0.5275,
+      "step": 1127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008009980156581216,
+      "loss": 0.4905,
+      "step": 1128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008006359065890137,
+      "loss": 0.5291,
+      "step": 1129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008002735503850016,
+      "loss": 0.4781,
+      "step": 1130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007999109473439569,
+      "loss": 0.4701,
+      "step": 1131
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000799548097763954,
+      "loss": 0.5104,
+      "step": 1132
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007991850019432701,
+      "loss": 0.4935,
+      "step": 1133
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007988216601803844,
+      "loss": 0.4874,
+      "step": 1134
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007984580727739789,
+      "loss": 0.5178,
+      "step": 1135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000798094240022937,
+      "loss": 0.5229,
+      "step": 1136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000797730162226344,
+      "loss": 0.4945,
+      "step": 1137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007973658396834867,
+      "loss": 0.5038,
+      "step": 1138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007970012726938527,
+      "loss": 0.4856,
+      "step": 1139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007966364615571314,
+      "loss": 0.5095,
+      "step": 1140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007962714065732116,
+      "loss": 0.5026,
+      "step": 1141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007959061080421839,
+      "loss": 0.4893,
+      "step": 1142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007955405662643383,
+      "loss": 0.4734,
+      "step": 1143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000795174781540165,
+      "loss": 0.4509,
+      "step": 1144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007948087541703537,
+      "loss": 0.4755,
+      "step": 1145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007944424844557941,
+      "loss": 0.467,
+      "step": 1146
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007940759726975744,
+      "loss": 0.4635,
+      "step": 1147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007937092191969822,
+      "loss": 0.4471,
+      "step": 1148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000793342224255504,
+      "loss": 0.485,
+      "step": 1149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007929749881748241,
+      "loss": 0.5028,
+      "step": 1150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007926075112568258,
+      "loss": 0.5056,
+      "step": 1151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00079223979380359,
+      "loss": 0.4906,
+      "step": 1152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000791871836117395,
+      "loss": 0.4888,
+      "step": 1153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007915036385007173,
+      "loss": 0.5214,
+      "step": 1154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007911352012562301,
+      "loss": 0.4603,
+      "step": 1155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007907665246868039,
+      "loss": 0.5265,
+      "step": 1156
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007903976090955058,
+      "loss": 0.4807,
+      "step": 1157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007900284547855992,
+      "loss": 0.494,
+      "step": 1158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007896590620605439,
+      "loss": 0.4996,
+      "step": 1159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007892894312239957,
+      "loss": 0.4885,
+      "step": 1160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007889195625798064,
+      "loss": 0.484,
+      "step": 1161
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007885494564320228,
+      "loss": 0.5054,
+      "step": 1162
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007881791130848872,
+      "loss": 0.4968,
+      "step": 1163
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007878085328428368,
+      "loss": 0.4855,
+      "step": 1164
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007874377160105036,
+      "loss": 0.4828,
+      "step": 1165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007870666628927141,
+      "loss": 0.4889,
+      "step": 1166
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007866953737944891,
+      "loss": 0.4702,
+      "step": 1167
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007863238490210432,
+      "loss": 0.482,
+      "step": 1168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007859520888777847,
+      "loss": 0.4858,
+      "step": 1169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007855800936703157,
+      "loss": 0.4828,
+      "step": 1170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007852078637044312,
+      "loss": 0.5055,
+      "step": 1171
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007848353992861195,
+      "loss": 0.4613,
+      "step": 1172
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007844627007215613,
+      "loss": 0.4874,
+      "step": 1173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007840897683171297,
+      "loss": 0.4984,
+      "step": 1174
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007837166023793909,
+      "loss": 0.4689,
+      "step": 1175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000783343203215102,
+      "loss": 0.4872,
+      "step": 1176
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007829695711312123,
+      "loss": 0.5081,
+      "step": 1177
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007825957064348625,
+      "loss": 0.4917,
+      "step": 1178
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007822216094333848,
+      "loss": 0.5101,
+      "step": 1179
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007818472804343016,
+      "loss": 0.4749,
+      "step": 1180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007814727197453269,
+      "loss": 0.476,
+      "step": 1181
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007810979276743646,
+      "loss": 0.5342,
+      "step": 1182
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000780722904529509,
+      "loss": 0.4758,
+      "step": 1183
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007803476506190445,
+      "loss": 0.517,
+      "step": 1184
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007799721662514448,
+      "loss": 0.4972,
+      "step": 1185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007795964517353734,
+      "loss": 0.5205,
+      "step": 1186
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007792205073796829,
+      "loss": 0.4766,
+      "step": 1187
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007788443334934147,
+      "loss": 0.4779,
+      "step": 1188
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007784679303857991,
+      "loss": 0.4734,
+      "step": 1189
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007780912983662547,
+      "loss": 0.5247,
+      "step": 1190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007777144377443881,
+      "loss": 0.4855,
+      "step": 1191
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007773373488299941,
+      "loss": 0.4872,
+      "step": 1192
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007769600319330552,
+      "loss": 0.4889,
+      "step": 1193
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000776582487363741,
+      "loss": 0.4995,
+      "step": 1194
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007762047154324088,
+      "loss": 0.5004,
+      "step": 1195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000775826716449602,
+      "loss": 0.4525,
+      "step": 1196
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007754484907260512,
+      "loss": 0.4952,
+      "step": 1197
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007750700385726736,
+      "loss": 0.4855,
+      "step": 1198
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007746913603005718,
+      "loss": 0.4961,
+      "step": 1199
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007743124562210351,
+      "loss": 0.5151,
+      "step": 1200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007739333266455375,
+      "loss": 0.4694,
+      "step": 1201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007735539718857393,
+      "loss": 0.4624,
+      "step": 1202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007731743922534853,
+      "loss": 0.5019,
+      "step": 1203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007727945880608056,
+      "loss": 0.4668,
+      "step": 1204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007724145596199142,
+      "loss": 0.4495,
+      "step": 1205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007720343072432104,
+      "loss": 0.4906,
+      "step": 1206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007716538312432765,
+      "loss": 0.5035,
+      "step": 1207
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007712731319328797,
+      "loss": 0.5198,
+      "step": 1208
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007708922096249701,
+      "loss": 0.4851,
+      "step": 1209
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007705110646326811,
+      "loss": 0.5172,
+      "step": 1210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007701296972693297,
+      "loss": 0.5397,
+      "step": 1211
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000769748107848415,
+      "loss": 0.4818,
+      "step": 1212
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007693662966836191,
+      "loss": 0.4956,
+      "step": 1213
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007689842640888063,
+      "loss": 0.4904,
+      "step": 1214
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007686020103780228,
+      "loss": 0.5042,
+      "step": 1215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007682195358654966,
+      "loss": 0.4811,
+      "step": 1216
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007678368408656377,
+      "loss": 0.4739,
+      "step": 1217
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007674539256930363,
+      "loss": 0.4785,
+      "step": 1218
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007670707906624643,
+      "loss": 0.4794,
+      "step": 1219
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007666874360888746,
+      "loss": 0.4994,
+      "step": 1220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007663038622873999,
+      "loss": 0.4706,
+      "step": 1221
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007659200695733536,
+      "loss": 0.4716,
+      "step": 1222
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007655360582622286,
+      "loss": 0.4556,
+      "step": 1223
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007651518286696981,
+      "loss": 0.507,
+      "step": 1224
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007647673811116141,
+      "loss": 0.514,
+      "step": 1225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007643827159040084,
+      "loss": 0.4579,
+      "step": 1226
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007639978333630911,
+      "loss": 0.4736,
+      "step": 1227
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007636127338052513,
+      "loss": 0.5084,
+      "step": 1228
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007632274175470565,
+      "loss": 0.4858,
+      "step": 1229
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007628418849052523,
+      "loss": 0.5061,
+      "step": 1230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007624561361967623,
+      "loss": 0.4557,
+      "step": 1231
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007620701717386873,
+      "loss": 0.4868,
+      "step": 1232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000761683991848306,
+      "loss": 0.5192,
+      "step": 1233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007612975968430737,
+      "loss": 0.4502,
+      "step": 1234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000760910987040623,
+      "loss": 0.5092,
+      "step": 1235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007605241627587627,
+      "loss": 0.5236,
+      "step": 1236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007601371243154779,
+      "loss": 0.4855,
+      "step": 1237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007597498720289301,
+      "loss": 0.455,
+      "step": 1238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007593624062174562,
+      "loss": 0.4882,
+      "step": 1239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000758974727199569,
+      "loss": 0.4677,
+      "step": 1240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007585868352939563,
+      "loss": 0.464,
+      "step": 1241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000758198730819481,
+      "loss": 0.5311,
+      "step": 1242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007578104140951806,
+      "loss": 0.4889,
+      "step": 1243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007574218854402673,
+      "loss": 0.4884,
+      "step": 1244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007570331451741274,
+      "loss": 0.4968,
+      "step": 1245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007566441936163211,
+      "loss": 0.4878,
+      "step": 1246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007562550310865824,
+      "loss": 0.5131,
+      "step": 1247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007558656579048185,
+      "loss": 0.4873,
+      "step": 1248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007554760743911103,
+      "loss": 0.5042,
+      "step": 1249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007550862808657111,
+      "loss": 0.5206,
+      "step": 1250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007546962776490467,
+      "loss": 0.4623,
+      "step": 1251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007543060650617158,
+      "loss": 0.4797,
+      "step": 1252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007539156434244891,
+      "loss": 0.4971,
+      "step": 1253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007535250130583088,
+      "loss": 0.486,
+      "step": 1254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000753134174284289,
+      "loss": 0.4874,
+      "step": 1255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007527431274237149,
+      "loss": 0.4965,
+      "step": 1256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000752351872798043,
+      "loss": 0.4835,
+      "step": 1257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007519604107289003,
+      "loss": 0.4629,
+      "step": 1258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007515687415380847,
+      "loss": 0.4563,
+      "step": 1259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007511768655475642,
+      "loss": 0.4584,
+      "step": 1260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007507847830794765,
+      "loss": 0.4656,
+      "step": 1261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007503924944561294,
+      "loss": 0.498,
+      "step": 1262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00075,
+      "loss": 0.4922,
+      "step": 1263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007496073000337348,
+      "loss": 0.5165,
+      "step": 1264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007492143948801489,
+      "loss": 0.45,
+      "step": 1265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007488212848622266,
+      "loss": 0.4934,
+      "step": 1266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007484279703031198,
+      "loss": 0.4706,
+      "step": 1267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007480344515261494,
+      "loss": 0.4723,
+      "step": 1268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007476407288548036,
+      "loss": 0.5033,
+      "step": 1269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007472468026127384,
+      "loss": 0.5067,
+      "step": 1270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007468526731237772,
+      "loss": 0.4804,
+      "step": 1271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007464583407119106,
+      "loss": 0.4962,
+      "step": 1272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007460638057012955,
+      "loss": 0.4822,
+      "step": 1273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007456690684162558,
+      "loss": 0.5319,
+      "step": 1274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007452741291812814,
+      "loss": 0.4917,
+      "step": 1275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007448789883210288,
+      "loss": 0.4492,
+      "step": 1276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007444836461603195,
+      "loss": 0.4661,
+      "step": 1277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007440881030241406,
+      "loss": 0.4713,
+      "step": 1278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000743692359237645,
+      "loss": 0.5076,
+      "step": 1279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00074329641512615,
+      "loss": 0.4703,
+      "step": 1280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007429002710151375,
+      "loss": 0.5297,
+      "step": 1281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007425039272302543,
+      "loss": 0.4822,
+      "step": 1282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007421073840973109,
+      "loss": 0.4935,
+      "step": 1283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007417106419422819,
+      "loss": 0.511,
+      "step": 1284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007413137010913055,
+      "loss": 0.4758,
+      "step": 1285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000740916561870683,
+      "loss": 0.5406,
+      "step": 1286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000740519224606879,
+      "loss": 0.4576,
+      "step": 1287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007401216896265209,
+      "loss": 0.4573,
+      "step": 1288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007397239572563987,
+      "loss": 0.5239,
+      "step": 1289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007393260278234641,
+      "loss": 0.5234,
+      "step": 1290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007389279016548316,
+      "loss": 0.5117,
+      "step": 1291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007385295790777769,
+      "loss": 0.5028,
+      "step": 1292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007381310604197374,
+      "loss": 0.4956,
+      "step": 1293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007377323460083117,
+      "loss": 0.5012,
+      "step": 1294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007373334361712589,
+      "loss": 0.5533,
+      "step": 1295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007369343312364993,
+      "loss": 0.4481,
+      "step": 1296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007365350315321134,
+      "loss": 0.5078,
+      "step": 1297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007361355373863414,
+      "loss": 0.4661,
+      "step": 1298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000735735849127584,
+      "loss": 0.4647,
+      "step": 1299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007353359670844011,
+      "loss": 0.4951,
+      "step": 1300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007349358915855117,
+      "loss": 0.4888,
+      "step": 1301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007345356229597943,
+      "loss": 0.5094,
+      "step": 1302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000734135161536286,
+      "loss": 0.5104,
+      "step": 1303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007337345076441822,
+      "loss": 0.4924,
+      "step": 1304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007333336616128369,
+      "loss": 0.441,
+      "step": 1305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007329326237717615,
+      "loss": 0.4877,
+      "step": 1306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007325313944506253,
+      "loss": 0.4955,
+      "step": 1307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007321299739792553,
+      "loss": 0.4817,
+      "step": 1308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007317283626876352,
+      "loss": 0.499,
+      "step": 1309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007313265609059061,
+      "loss": 0.5042,
+      "step": 1310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000730924568964365,
+      "loss": 0.4346,
+      "step": 1311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007305223871934656,
+      "loss": 0.4901,
+      "step": 1312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007301200159238179,
+      "loss": 0.501,
+      "step": 1313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007297174554861871,
+      "loss": 0.4669,
+      "step": 1314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007293147062114946,
+      "loss": 0.4766,
+      "step": 1315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007289117684308162,
+      "loss": 0.5098,
+      "step": 1316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007285086424753832,
+      "loss": 0.4637,
+      "step": 1317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007281053286765816,
+      "loss": 0.4843,
+      "step": 1318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007277018273659516,
+      "loss": 0.4867,
+      "step": 1319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007272981388751875,
+      "loss": 0.4814,
+      "step": 1320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000726894263536138,
+      "loss": 0.4436,
+      "step": 1321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007264902016808048,
+      "loss": 0.4578,
+      "step": 1322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007260859536413429,
+      "loss": 0.4678,
+      "step": 1323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000725681519750061,
+      "loss": 0.4888,
+      "step": 1324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007252769003394197,
+      "loss": 0.4962,
+      "step": 1325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007248720957420329,
+      "loss": 0.4965,
+      "step": 1326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007244671062906663,
+      "loss": 0.5167,
+      "step": 1327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007240619323182377,
+      "loss": 0.4615,
+      "step": 1328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007236565741578163,
+      "loss": 0.4845,
+      "step": 1329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007232510321426233,
+      "loss": 0.5183,
+      "step": 1330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007228453066060303,
+      "loss": 0.481,
+      "step": 1331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007224393978815604,
+      "loss": 0.5031,
+      "step": 1332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007220333063028871,
+      "loss": 0.4706,
+      "step": 1333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007216270322038339,
+      "loss": 0.4742,
+      "step": 1334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007212205759183748,
+      "loss": 0.4838,
+      "step": 1335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007208139377806329,
+      "loss": 0.4775,
+      "step": 1336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007204071181248818,
+      "loss": 0.5031,
+      "step": 1337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007200001172855435,
+      "loss": 0.4598,
+      "step": 1338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007195929355971892,
+      "loss": 0.4689,
+      "step": 1339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007191855733945387,
+      "loss": 0.4493,
+      "step": 1340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007187780310124604,
+      "loss": 0.4464,
+      "step": 1341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007183703087859705,
+      "loss": 0.5359,
+      "step": 1342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007179624070502333,
+      "loss": 0.4865,
+      "step": 1343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007175543261405607,
+      "loss": 0.5355,
+      "step": 1344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007171460663924113,
+      "loss": 0.5138,
+      "step": 1345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007167376281413915,
+      "loss": 0.4951,
+      "step": 1346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007163290117232541,
+      "loss": 0.4332,
+      "step": 1347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007159202174738984,
+      "loss": 0.4492,
+      "step": 1348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007155112457293697,
+      "loss": 0.4869,
+      "step": 1349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007151020968258595,
+      "loss": 0.4891,
+      "step": 1350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007146927710997047,
+      "loss": 0.507,
+      "step": 1351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007142832688873875,
+      "loss": 0.4905,
+      "step": 1352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007138735905255354,
+      "loss": 0.4897,
+      "step": 1353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007134637363509209,
+      "loss": 0.5085,
+      "step": 1354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007130537067004606,
+      "loss": 0.4548,
+      "step": 1355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007126435019112152,
+      "loss": 0.489,
+      "step": 1356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007122331223203901,
+      "loss": 0.4902,
+      "step": 1357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007118225682653336,
+      "loss": 0.517,
+      "step": 1358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007114118400835381,
+      "loss": 0.4702,
+      "step": 1359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007110009381126385,
+      "loss": 0.4843,
+      "step": 1360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007105898626904133,
+      "loss": 0.4767,
+      "step": 1361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007101786141547828,
+      "loss": 0.4813,
+      "step": 1362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007097671928438101,
+      "loss": 0.4468,
+      "step": 1363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007093555990957,
+      "loss": 0.4812,
+      "step": 1364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007089438332487997,
+      "loss": 0.4976,
+      "step": 1365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007085318956415969,
+      "loss": 0.4508,
+      "step": 1366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007081197866127214,
+      "loss": 0.4712,
+      "step": 1367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007077075065009433,
+      "loss": 0.4905,
+      "step": 1368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007072950556451736,
+      "loss": 0.5457,
+      "step": 1369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007068824343844638,
+      "loss": 0.4883,
+      "step": 1370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007064696430580051,
+      "loss": 0.4758,
+      "step": 1371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000706056682005129,
+      "loss": 0.4731,
+      "step": 1372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000705643551565306,
+      "loss": 0.4784,
+      "step": 1373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007052302520781458,
+      "loss": 0.5049,
+      "step": 1374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007048167838833977,
+      "loss": 0.4669,
+      "step": 1375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007044031473209493,
+      "loss": 0.4912,
+      "step": 1376
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007039893427308268,
+      "loss": 0.485,
+      "step": 1377
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000703575370453194,
+      "loss": 0.5094,
+      "step": 1378
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007031612308283528,
+      "loss": 0.4845,
+      "step": 1379
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007027469241967432,
+      "loss": 0.4697,
+      "step": 1380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007023324508989417,
+      "loss": 0.4915,
+      "step": 1381
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007019178112756625,
+      "loss": 0.4927,
+      "step": 1382
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007015030056677558,
+      "loss": 0.5092,
+      "step": 1383
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007010880344162087,
+      "loss": 0.5066,
+      "step": 1384
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007006728978621446,
+      "loss": 0.489,
+      "step": 1385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007002575963468225,
+      "loss": 0.4938,
+      "step": 1386
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000699842130211637,
+      "loss": 0.489,
+      "step": 1387
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006994264997981183,
+      "loss": 0.5184,
+      "step": 1388
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006990107054479312,
+      "loss": 0.4869,
+      "step": 1389
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006985947475028755,
+      "loss": 0.5112,
+      "step": 1390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006981786263048858,
+      "loss": 0.5181,
+      "step": 1391
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006977623421960303,
+      "loss": 0.4626,
+      "step": 1392
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006973458955185116,
+      "loss": 0.4937,
+      "step": 1393
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006969292866146656,
+      "loss": 0.4622,
+      "step": 1394
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006965125158269618,
+      "loss": 0.4989,
+      "step": 1395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006960955834980027,
+      "loss": 0.4955,
+      "step": 1396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006956784899705236,
+      "loss": 0.4921,
+      "step": 1397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006952612355873922,
+      "loss": 0.5454,
+      "step": 1398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006948438206916085,
+      "loss": 0.5197,
+      "step": 1399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006944262456263047,
+      "loss": 0.5032,
+      "step": 1400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006940085107347443,
+      "loss": 0.4509,
+      "step": 1401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006935906163603219,
+      "loss": 0.4612,
+      "step": 1402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006931725628465643,
+      "loss": 0.5043,
+      "step": 1403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006927543505371281,
+      "loss": 0.5266,
+      "step": 1404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006923359797758008,
+      "loss": 0.5137,
+      "step": 1405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006919174509065003,
+      "loss": 0.5364,
+      "step": 1406
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006914987642732738,
+      "loss": 0.5225,
+      "step": 1407
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006910799202202993,
+      "loss": 0.4967,
+      "step": 1408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006906609190918832,
+      "loss": 0.4425,
+      "step": 1409
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006902417612324615,
+      "loss": 0.5,
+      "step": 1410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006898224469865991,
+      "loss": 0.4818,
+      "step": 1411
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006894029766989891,
+      "loss": 0.4902,
+      "step": 1412
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006889833507144532,
+      "loss": 0.4628,
+      "step": 1413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006885635693779411,
+      "loss": 0.493,
+      "step": 1414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006881436330345299,
+      "loss": 0.5012,
+      "step": 1415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006877235420294244,
+      "loss": 0.5256,
+      "step": 1416
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006873032967079561,
+      "loss": 0.4828,
+      "step": 1417
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006868828974155841,
+      "loss": 0.5115,
+      "step": 1418
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006864623444978934,
+      "loss": 0.4797,
+      "step": 1419
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006860416383005958,
+      "loss": 0.4933,
+      "step": 1420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006856207791695288,
+      "loss": 0.4963,
+      "step": 1421
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006851997674506556,
+      "loss": 0.475,
+      "step": 1422
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006847786034900648,
+      "loss": 0.4734,
+      "step": 1423
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006843572876339704,
+      "loss": 0.4984,
+      "step": 1424
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000683935820228711,
+      "loss": 0.4636,
+      "step": 1425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00068351420162075,
+      "loss": 0.5083,
+      "step": 1426
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000683092432156675,
+      "loss": 0.4603,
+      "step": 1427
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006826705121831977,
+      "loss": 0.4602,
+      "step": 1428
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000682248442047153,
+      "loss": 0.5175,
+      "step": 1429
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006818262220954999,
+      "loss": 0.4846,
+      "step": 1430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006814038526753205,
+      "loss": 0.4949,
+      "step": 1431
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006809813341338193,
+      "loss": 0.481,
+      "step": 1432
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006805586668183241,
+      "loss": 0.4991,
+      "step": 1433
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006801358510762842,
+      "loss": 0.4659,
+      "step": 1434
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006797128872552713,
+      "loss": 0.5186,
+      "step": 1435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006792897757029788,
+      "loss": 0.5214,
+      "step": 1436
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006788665167672217,
+      "loss": 0.4918,
+      "step": 1437
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006784431107959359,
+      "loss": 0.4929,
+      "step": 1438
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006780195581371784,
+      "loss": 0.4703,
+      "step": 1439
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006775958591391265,
+      "loss": 0.5085,
+      "step": 1440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000677172014150078,
+      "loss": 0.47,
+      "step": 1441
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006767480235184508,
+      "loss": 0.483,
+      "step": 1442
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006763238875927822,
+      "loss": 0.4993,
+      "step": 1443
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006758996067217293,
+      "loss": 0.4744,
+      "step": 1444
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006754751812540679,
+      "loss": 0.4883,
+      "step": 1445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006750506115386932,
+      "loss": 0.4672,
+      "step": 1446
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006746258979246185,
+      "loss": 0.4934,
+      "step": 1447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006742010407609759,
+      "loss": 0.4427,
+      "step": 1448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006737760403970151,
+      "loss": 0.4988,
+      "step": 1449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006733508971821036,
+      "loss": 0.5201,
+      "step": 1450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006729256114657263,
+      "loss": 0.4763,
+      "step": 1451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006725001835974853,
+      "loss": 0.498,
+      "step": 1452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006720746139270996,
+      "loss": 0.4857,
+      "step": 1453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000671648902804405,
+      "loss": 0.4916,
+      "step": 1454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006712230505793529,
+      "loss": 0.5216,
+      "step": 1455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006707970576020112,
+      "loss": 0.4893,
+      "step": 1456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006703709242225633,
+      "loss": 0.5077,
+      "step": 1457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006699446507913083,
+      "loss": 0.4902,
+      "step": 1458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006695182376586602,
+      "loss": 0.4629,
+      "step": 1459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006690916851751481,
+      "loss": 0.4808,
+      "step": 1460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006686649936914151,
+      "loss": 0.5197,
+      "step": 1461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006682381635582189,
+      "loss": 0.4977,
+      "step": 1462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006678111951264313,
+      "loss": 0.5065,
+      "step": 1463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006673840887470377,
+      "loss": 0.4586,
+      "step": 1464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006669568447711367,
+      "loss": 0.4845,
+      "step": 1465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006665294635499404,
+      "loss": 0.4619,
+      "step": 1466
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006661019454347733,
+      "loss": 0.4768,
+      "step": 1467
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006656742907770728,
+      "loss": 0.5198,
+      "step": 1468
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006652464999283881,
+      "loss": 0.4656,
+      "step": 1469
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006648185732403809,
+      "loss": 0.4489,
+      "step": 1470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006643905110648241,
+      "loss": 0.5231,
+      "step": 1471
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006639623137536022,
+      "loss": 0.4761,
+      "step": 1472
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006635339816587109,
+      "loss": 0.4705,
+      "step": 1473
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006631055151322564,
+      "loss": 0.4692,
+      "step": 1474
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006626769145264554,
+      "loss": 0.4949,
+      "step": 1475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006622481801936353,
+      "loss": 0.511,
+      "step": 1476
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006618193124862331,
+      "loss": 0.5095,
+      "step": 1477
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006613903117567951,
+      "loss": 0.4623,
+      "step": 1478
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006609611783579775,
+      "loss": 0.4807,
+      "step": 1479
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006605319126425454,
+      "loss": 0.4904,
+      "step": 1480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006601025149633726,
+      "loss": 0.4551,
+      "step": 1481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006596729856734413,
+      "loss": 0.4861,
+      "step": 1482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006592433251258423,
+      "loss": 0.4805,
+      "step": 1483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006588135336737735,
+      "loss": 0.457,
+      "step": 1484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006583836116705413,
+      "loss": 0.4597,
+      "step": 1485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006579535594695588,
+      "loss": 0.4871,
+      "step": 1486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006575233774243465,
+      "loss": 0.4867,
+      "step": 1487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006570930658885314,
+      "loss": 0.4733,
+      "step": 1488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006566626252158469,
+      "loss": 0.4761,
+      "step": 1489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006562320557601329,
+      "loss": 0.479,
+      "step": 1490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006558013578753347,
+      "loss": 0.4918,
+      "step": 1491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006553705319155039,
+      "loss": 0.4614,
+      "step": 1492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006549395782347963,
+      "loss": 0.4894,
+      "step": 1493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006545084971874737,
+      "loss": 0.5403,
+      "step": 1494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000654077289127902,
+      "loss": 0.5409,
+      "step": 1495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000653645954410552,
+      "loss": 0.509,
+      "step": 1496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006532144933899981,
+      "loss": 0.4724,
+      "step": 1497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006527829064209187,
+      "loss": 0.5101,
+      "step": 1498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006523511938580957,
+      "loss": 0.4579,
+      "step": 1499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006519193560564149,
+      "loss": 0.4866,
+      "step": 1500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006514873933708637,
+      "loss": 0.4869,
+      "step": 1501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006510553061565334,
+      "loss": 0.4949,
+      "step": 1502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006506230947686171,
+      "loss": 0.4642,
+      "step": 1503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006501907595624102,
+      "loss": 0.5062,
+      "step": 1504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006497583008933097,
+      "loss": 0.4651,
+      "step": 1505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000649325719116814,
+      "loss": 0.4789,
+      "step": 1506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000648893014588523,
+      "loss": 0.5105,
+      "step": 1507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006484601876641375,
+      "loss": 0.4489,
+      "step": 1508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006480272386994585,
+      "loss": 0.4645,
+      "step": 1509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006475941680503879,
+      "loss": 0.5349,
+      "step": 1510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000647160976072927,
+      "loss": 0.4578,
+      "step": 1511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006467276631231773,
+      "loss": 0.4731,
+      "step": 1512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006462942295573398,
+      "loss": 0.5052,
+      "step": 1513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006458606757317142,
+      "loss": 0.5029,
+      "step": 1514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006454270020026995,
+      "loss": 0.4951,
+      "step": 1515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006449932087267931,
+      "loss": 0.4512,
+      "step": 1516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006445592962605907,
+      "loss": 0.5242,
+      "step": 1517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006441252649607855,
+      "loss": 0.4872,
+      "step": 1518
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006436911151841693,
+      "loss": 0.4945,
+      "step": 1519
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006432568472876307,
+      "loss": 0.4578,
+      "step": 1520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006428224616281554,
+      "loss": 0.4677,
+      "step": 1521
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006423879585628261,
+      "loss": 0.5184,
+      "step": 1522
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000641953338448822,
+      "loss": 0.4707,
+      "step": 1523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006415186016434185,
+      "loss": 0.4618,
+      "step": 1524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006410837485039866,
+      "loss": 0.4597,
+      "step": 1525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006406487793879936,
+      "loss": 0.4966,
+      "step": 1526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006402136946530014,
+      "loss": 0.4747,
+      "step": 1527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006397784946566676,
+      "loss": 0.4977,
+      "step": 1528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006393431797567439,
+      "loss": 0.4802,
+      "step": 1529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006389077503110769,
+      "loss": 0.4928,
+      "step": 1530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006384722066776074,
+      "loss": 0.4961,
+      "step": 1531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00063803654921437,
+      "loss": 0.4833,
+      "step": 1532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006376007782794926,
+      "loss": 0.4703,
+      "step": 1533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006371648942311966,
+      "loss": 0.4954,
+      "step": 1534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006367288974277964,
+      "loss": 0.4669,
+      "step": 1535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006362927882276989,
+      "loss": 0.463,
+      "step": 1536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006358565669894039,
+      "loss": 0.4846,
+      "step": 1537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006354202340715027,
+      "loss": 0.4717,
+      "step": 1538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006349837898326784,
+      "loss": 0.478,
+      "step": 1539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006345472346317064,
+      "loss": 0.4532,
+      "step": 1540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006341105688274525,
+      "loss": 0.4932,
+      "step": 1541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006336737927788737,
+      "loss": 0.4575,
+      "step": 1542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006332369068450174,
+      "loss": 0.4835,
+      "step": 1543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006327999113850221,
+      "loss": 0.5162,
+      "step": 1544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006323628067581153,
+      "loss": 0.4968,
+      "step": 1545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006319255933236149,
+      "loss": 0.4734,
+      "step": 1546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006314882714409281,
+      "loss": 0.4578,
+      "step": 1547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000631050841469551,
+      "loss": 0.5012,
+      "step": 1548
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006306133037690692,
+      "loss": 0.4662,
+      "step": 1549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006301756586991561,
+      "loss": 0.4584,
+      "step": 1550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006297379066195737,
+      "loss": 0.4647,
+      "step": 1551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000629300047890172,
+      "loss": 0.5023,
+      "step": 1552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006288620828708888,
+      "loss": 0.4885,
+      "step": 1553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006284240119217488,
+      "loss": 0.4698,
+      "step": 1554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006279858354028644,
+      "loss": 0.5222,
+      "step": 1555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006275475536744342,
+      "loss": 0.5001,
+      "step": 1556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006271091670967436,
+      "loss": 0.4818,
+      "step": 1557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000626670676030164,
+      "loss": 0.4884,
+      "step": 1558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006262320808351534,
+      "loss": 0.4998,
+      "step": 1559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006257933818722543,
+      "loss": 0.5085,
+      "step": 1560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006253545795020951,
+      "loss": 0.5249,
+      "step": 1561
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006249156740853894,
+      "loss": 0.4777,
+      "step": 1562
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000624476665982935,
+      "loss": 0.4669,
+      "step": 1563
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006240375555556145,
+      "loss": 0.4968,
+      "step": 1564
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006235983431643943,
+      "loss": 0.4722,
+      "step": 1565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000623159029170325,
+      "loss": 0.4865,
+      "step": 1566
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006227196139345403,
+      "loss": 0.4944,
+      "step": 1567
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006222800978182575,
+      "loss": 0.5038,
+      "step": 1568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006218404811827767,
+      "loss": 0.5016,
+      "step": 1569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006214007643894805,
+      "loss": 0.4979,
+      "step": 1570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006209609477998338,
+      "loss": 0.4855,
+      "step": 1571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006205210317753842,
+      "loss": 0.5099,
+      "step": 1572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006200810166777597,
+      "loss": 0.5134,
+      "step": 1573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006196409028686713,
+      "loss": 0.5195,
+      "step": 1574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006192006907099098,
+      "loss": 0.5045,
+      "step": 1575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006187603805633479,
+      "loss": 0.455,
+      "step": 1576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006183199727909381,
+      "loss": 0.4972,
+      "step": 1577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006178794677547138,
+      "loss": 0.5027,
+      "step": 1578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006174388658167874,
+      "loss": 0.4811,
+      "step": 1579
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006169981673393521,
+      "loss": 0.4768,
+      "step": 1580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006165573726846797,
+      "loss": 0.4775,
+      "step": 1581
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006161164822151212,
+      "loss": 0.4841,
+      "step": 1582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006156754962931069,
+      "loss": 0.5071,
+      "step": 1583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006152344152811444,
+      "loss": 0.4962,
+      "step": 1584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006147932395418205,
+      "loss": 0.4729,
+      "step": 1585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006143519694377997,
+      "loss": 0.4791,
+      "step": 1586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006139106053318239,
+      "loss": 0.5006,
+      "step": 1587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006134691475867121,
+      "loss": 0.4668,
+      "step": 1588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006130275965653604,
+      "loss": 0.5096,
+      "step": 1589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000612585952630742,
+      "loss": 0.4608,
+      "step": 1590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006121442161459058,
+      "loss": 0.452,
+      "step": 1591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006117023874739772,
+      "loss": 0.4903,
+      "step": 1592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006112604669781572,
+      "loss": 0.4976,
+      "step": 1593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006108184550217226,
+      "loss": 0.4813,
+      "step": 1594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006103763519680249,
+      "loss": 0.4861,
+      "step": 1595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006099341581804909,
+      "loss": 0.4752,
+      "step": 1596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006094918740226216,
+      "loss": 0.4509,
+      "step": 1597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006090494998579928,
+      "loss": 0.4918,
+      "step": 1598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006086070360502539,
+      "loss": 0.5277,
+      "step": 1599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000608164482963128,
+      "loss": 0.4651,
+      "step": 1600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006077218409604119,
+      "loss": 0.4946,
+      "step": 1601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006072791104059748,
+      "loss": 0.531,
+      "step": 1602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006068362916637596,
+      "loss": 0.4829,
+      "step": 1603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006063933850977811,
+      "loss": 0.47,
+      "step": 1604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006059503910721266,
+      "loss": 0.4443,
+      "step": 1605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006055073099509549,
+      "loss": 0.469,
+      "step": 1606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006050641420984968,
+      "loss": 0.4829,
+      "step": 1607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006046208878790542,
+      "loss": 0.504,
+      "step": 1608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006041775476570002,
+      "loss": 0.5209,
+      "step": 1609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006037341217967783,
+      "loss": 0.486,
+      "step": 1610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006032906106629024,
+      "loss": 0.4771,
+      "step": 1611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006028470146199567,
+      "loss": 0.467,
+      "step": 1612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006024033340325954,
+      "loss": 0.4535,
+      "step": 1613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006019595692655415,
+      "loss": 0.5122,
+      "step": 1614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006015157206835881,
+      "loss": 0.4854,
+      "step": 1615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006010717886515965,
+      "loss": 0.4762,
+      "step": 1616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006006277735344967,
+      "loss": 0.4896,
+      "step": 1617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006001836756972873,
+      "loss": 0.514,
+      "step": 1618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005997394955050346,
+      "loss": 0.4611,
+      "step": 1619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005992952333228728,
+      "loss": 0.4811,
+      "step": 1620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005988508895160034,
+      "loss": 0.4781,
+      "step": 1621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005984064644496952,
+      "loss": 0.4768,
+      "step": 1622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005979619584892833,
+      "loss": 0.5004,
+      "step": 1623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005975173720001699,
+      "loss": 0.4481,
+      "step": 1624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005970727053478229,
+      "loss": 0.493,
+      "step": 1625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005966279588977767,
+      "loss": 0.4857,
+      "step": 1626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005961831330156305,
+      "loss": 0.491,
+      "step": 1627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005957382280670494,
+      "loss": 0.4635,
+      "step": 1628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005952932444177631,
+      "loss": 0.4695,
+      "step": 1629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005948481824335666,
+      "loss": 0.4429,
+      "step": 1630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005944030424803183,
+      "loss": 0.4607,
+      "step": 1631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005939578249239419,
+      "loss": 0.4716,
+      "step": 1632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005935125301304241,
+      "loss": 0.4995,
+      "step": 1633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005930671584658151,
+      "loss": 0.4974,
+      "step": 1634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005926217102962285,
+      "loss": 0.4891,
+      "step": 1635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005921761859878408,
+      "loss": 0.4944,
+      "step": 1636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005917305859068912,
+      "loss": 0.5066,
+      "step": 1637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005912849104196809,
+      "loss": 0.4962,
+      "step": 1638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005908391598925732,
+      "loss": 0.5021,
+      "step": 1639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000590393334691993,
+      "loss": 0.4958,
+      "step": 1640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000589947435184427,
+      "loss": 0.4539,
+      "step": 1641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005895014617364223,
+      "loss": 0.4894,
+      "step": 1642
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005890554147145874,
+      "loss": 0.4897,
+      "step": 1643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005886092944855912,
+      "loss": 0.4697,
+      "step": 1644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005881631014161622,
+      "loss": 0.4536,
+      "step": 1645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005877168358730894,
+      "loss": 0.4449,
+      "step": 1646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005872704982232212,
+      "loss": 0.4736,
+      "step": 1647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005868240888334653,
+      "loss": 0.4648,
+      "step": 1648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000586377608070788,
+      "loss": 0.4818,
+      "step": 1649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000585931056302215,
+      "loss": 0.4918,
+      "step": 1650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005854844338948296,
+      "loss": 0.5084,
+      "step": 1651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005850377412157738,
+      "loss": 0.4548,
+      "step": 1652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005845909786322469,
+      "loss": 0.5042,
+      "step": 1653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000584144146511506,
+      "loss": 0.49,
+      "step": 1654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005836972452208654,
+      "loss": 0.5029,
+      "step": 1655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005832502751276956,
+      "loss": 0.4236,
+      "step": 1656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005828032365994244,
+      "loss": 0.5011,
+      "step": 1657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005823561300035355,
+      "loss": 0.4582,
+      "step": 1658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005819089557075689,
+      "loss": 0.4795,
+      "step": 1659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005814617140791197,
+      "loss": 0.4528,
+      "step": 1660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005810144054858388,
+      "loss": 0.501,
+      "step": 1661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005805670302954321,
+      "loss": 0.4878,
+      "step": 1662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005801195888756599,
+      "loss": 0.4767,
+      "step": 1663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005796720815943373,
+      "loss": 0.48,
+      "step": 1664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005792245088193335,
+      "loss": 0.4971,
+      "step": 1665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005787768709185714,
+      "loss": 0.4492,
+      "step": 1666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005783291682600275,
+      "loss": 0.4519,
+      "step": 1667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005778814012117315,
+      "loss": 0.4786,
+      "step": 1668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005774335701417662,
+      "loss": 0.4862,
+      "step": 1669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005769856754182667,
+      "loss": 0.5034,
+      "step": 1670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005765377174094208,
+      "loss": 0.5093,
+      "step": 1671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005760896964834683,
+      "loss": 0.5115,
+      "step": 1672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005756416130087001,
+      "loss": 0.4841,
+      "step": 1673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005751934673534592,
+      "loss": 0.4574,
+      "step": 1674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005747452598861397,
+      "loss": 0.4991,
+      "step": 1675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005742969909751859,
+      "loss": 0.4552,
+      "step": 1676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005738486609890934,
+      "loss": 0.4512,
+      "step": 1677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005734002702964075,
+      "loss": 0.5079,
+      "step": 1678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005729518192657234,
+      "loss": 0.4775,
+      "step": 1679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005725033082656862,
+      "loss": 0.4897,
+      "step": 1680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005720547376649901,
+      "loss": 0.4817,
+      "step": 1681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005716061078323782,
+      "loss": 0.4517,
+      "step": 1682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005711574191366427,
+      "loss": 0.472,
+      "step": 1683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005707086719466234,
+      "loss": 0.4576,
+      "step": 1684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005702598666312092,
+      "loss": 0.4661,
+      "step": 1685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005698110035593359,
+      "loss": 0.4623,
+      "step": 1686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005693620830999873,
+      "loss": 0.4858,
+      "step": 1687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005689131056221944,
+      "loss": 0.4987,
+      "step": 1688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005684640714950346,
+      "loss": 0.4796,
+      "step": 1689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005680149810876322,
+      "loss": 0.5021,
+      "step": 1690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005675658347691577,
+      "loss": 0.5018,
+      "step": 1691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005671166329088278,
+      "loss": 0.4558,
+      "step": 1692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005666673758759045,
+      "loss": 0.4753,
+      "step": 1693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005662180640396953,
+      "loss": 0.4907,
+      "step": 1694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005657686977695526,
+      "loss": 0.4551,
+      "step": 1695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005653192774348739,
+      "loss": 0.5184,
+      "step": 1696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005648698034051009,
+      "loss": 0.5093,
+      "step": 1697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005644202760497195,
+      "loss": 0.4943,
+      "step": 1698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005639706957382593,
+      "loss": 0.4678,
+      "step": 1699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005635210628402937,
+      "loss": 0.4609,
+      "step": 1700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005630713777254389,
+      "loss": 0.4822,
+      "step": 1701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005626216407633546,
+      "loss": 0.4822,
+      "step": 1702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005621718523237427,
+      "loss": 0.4581,
+      "step": 1703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005617220127763474,
+      "loss": 0.4933,
+      "step": 1704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005612721224909552,
+      "loss": 0.4383,
+      "step": 1705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005608221818373942,
+      "loss": 0.4889,
+      "step": 1706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005603721911855337,
+      "loss": 0.4653,
+      "step": 1707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005599221509052844,
+      "loss": 0.5186,
+      "step": 1708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005594720613665977,
+      "loss": 0.4908,
+      "step": 1709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005590219229394652,
+      "loss": 0.4854,
+      "step": 1710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005585717359939192,
+      "loss": 0.4977,
+      "step": 1711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005581215009000315,
+      "loss": 0.4862,
+      "step": 1712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005576712180279134,
+      "loss": 0.4762,
+      "step": 1713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005572208877477159,
+      "loss": 0.4852,
+      "step": 1714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005567705104296288,
+      "loss": 0.4563,
+      "step": 1715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005563200864438808,
+      "loss": 0.4899,
+      "step": 1716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000555869616160738,
+      "loss": 0.5011,
+      "step": 1717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005554190999505056,
+      "loss": 0.4669,
+      "step": 1718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005549685381835261,
+      "loss": 0.4736,
+      "step": 1719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005545179312301797,
+      "loss": 0.4916,
+      "step": 1720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005540672794608835,
+      "loss": 0.5045,
+      "step": 1721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005536165832460913,
+      "loss": 0.5038,
+      "step": 1722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005531658429562941,
+      "loss": 0.4593,
+      "step": 1723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005527150589620183,
+      "loss": 0.4852,
+      "step": 1724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005522642316338268,
+      "loss": 0.4655,
+      "step": 1725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005518133613423179,
+      "loss": 0.499,
+      "step": 1726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005513624484581255,
+      "loss": 0.467,
+      "step": 1727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005509114933519178,
+      "loss": 0.5238,
+      "step": 1728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005504604963943985,
+      "loss": 0.5029,
+      "step": 1729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005500094579563053,
+      "loss": 0.5056,
+      "step": 1730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005495583784084101,
+      "loss": 0.4987,
+      "step": 1731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005491072581215186,
+      "loss": 0.4778,
+      "step": 1732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005486560974664699,
+      "loss": 0.4841,
+      "step": 1733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005482048968141365,
+      "loss": 0.4854,
+      "step": 1734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005477536565354235,
+      "loss": 0.4694,
+      "step": 1735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005473023770012686,
+      "loss": 0.4443,
+      "step": 1736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005468510585826421,
+      "loss": 0.4965,
+      "step": 1737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005463997016505459,
+      "loss": 0.5031,
+      "step": 1738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005459483065760138,
+      "loss": 0.4883,
+      "step": 1739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005454968737301107,
+      "loss": 0.4905,
+      "step": 1740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005450454034839329,
+      "loss": 0.4764,
+      "step": 1741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005445938962086069,
+      "loss": 0.485,
+      "step": 1742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005441423522752904,
+      "loss": 0.5101,
+      "step": 1743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005436907720551707,
+      "loss": 0.4829,
+      "step": 1744
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005432391559194649,
+      "loss": 0.4772,
+      "step": 1745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005427875042394199,
+      "loss": 0.4656,
+      "step": 1746
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005423358173863116,
+      "loss": 0.5087,
+      "step": 1747
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005418840957314451,
+      "loss": 0.4752,
+      "step": 1748
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005414323396461537,
+      "loss": 0.4825,
+      "step": 1749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005409805495017992,
+      "loss": 0.4769,
+      "step": 1750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005405287256697716,
+      "loss": 0.4778,
+      "step": 1751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005400768685214882,
+      "loss": 0.4631,
+      "step": 1752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005396249784283942,
+      "loss": 0.4861,
+      "step": 1753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005391730557619615,
+      "loss": 0.4768,
+      "step": 1754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005387211008936885,
+      "loss": 0.4745,
+      "step": 1755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005382691141951008,
+      "loss": 0.4127,
+      "step": 1756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005378170960377495,
+      "loss": 0.4911,
+      "step": 1757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005373650467932121,
+      "loss": 0.5166,
+      "step": 1758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005369129668330913,
+      "loss": 0.4718,
+      "step": 1759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005364608565290155,
+      "loss": 0.4559,
+      "step": 1760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000536008716252637,
+      "loss": 0.5151,
+      "step": 1761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005355565463756339,
+      "loss": 0.501,
+      "step": 1762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005351043472697081,
+      "loss": 0.4735,
+      "step": 1763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005346521193065856,
+      "loss": 0.5051,
+      "step": 1764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005341998628580161,
+      "loss": 0.4624,
+      "step": 1765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005337475782957729,
+      "loss": 0.5227,
+      "step": 1766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000533295265991652,
+      "loss": 0.4799,
+      "step": 1767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005328429263174724,
+      "loss": 0.4669,
+      "step": 1768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005323905596450759,
+      "loss": 0.5037,
+      "step": 1769
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005319381663463262,
+      "loss": 0.4786,
+      "step": 1770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000531485746793109,
+      "loss": 0.4952,
+      "step": 1771
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005310333013573311,
+      "loss": 0.5065,
+      "step": 1772
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005305808304109215,
+      "loss": 0.4674,
+      "step": 1773
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005301283343258293,
+      "loss": 0.4979,
+      "step": 1774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005296758134740249,
+      "loss": 0.5002,
+      "step": 1775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005292232682274985,
+      "loss": 0.4905,
+      "step": 1776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005287706989582609,
+      "loss": 0.4999,
+      "step": 1777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005283181060383423,
+      "loss": 0.469,
+      "step": 1778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005278654898397923,
+      "loss": 0.4595,
+      "step": 1779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005274128507346801,
+      "loss": 0.5166,
+      "step": 1780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000526960189095093,
+      "loss": 0.5278,
+      "step": 1781
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005265075052931374,
+      "loss": 0.4646,
+      "step": 1782
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005260547997009379,
+      "loss": 0.5063,
+      "step": 1783
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005256020726906366,
+      "loss": 0.4764,
+      "step": 1784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005251493246343936,
+      "loss": 0.4935,
+      "step": 1785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000524696555904386,
+      "loss": 0.4696,
+      "step": 1786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005242437668728082,
+      "loss": 0.5244,
+      "step": 1787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005237909579118712,
+      "loss": 0.4713,
+      "step": 1788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005233381293938022,
+      "loss": 0.4949,
+      "step": 1789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005228852816908444,
+      "loss": 0.5278,
+      "step": 1790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005224324151752575,
+      "loss": 0.4878,
+      "step": 1791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005219795302193157,
+      "loss": 0.4818,
+      "step": 1792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000521526627195309,
+      "loss": 0.4558,
+      "step": 1793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000521073706475542,
+      "loss": 0.4977,
+      "step": 1794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005206207684323337,
+      "loss": 0.4633,
+      "step": 1795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005201678134380178,
+      "loss": 0.5043,
+      "step": 1796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005197148418649416,
+      "loss": 0.4862,
+      "step": 1797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000519261854085466,
+      "loss": 0.4744,
+      "step": 1798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005188088504719656,
+      "loss": 0.4709,
+      "step": 1799
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005183558313968274,
+      "loss": 0.489,
+      "step": 1800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005179027972324516,
+      "loss": 0.4812,
+      "step": 1801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005174497483512506,
+      "loss": 0.438,
+      "step": 1802
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005169966851256488,
+      "loss": 0.473,
+      "step": 1803
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000516543607928083,
+      "loss": 0.4845,
+      "step": 1804
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005160905171310008,
+      "loss": 0.4562,
+      "step": 1805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005156374131068609,
+      "loss": 0.4742,
+      "step": 1806
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005151842962281335,
+      "loss": 0.5048,
+      "step": 1807
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000514731166867299,
+      "loss": 0.4698,
+      "step": 1808
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005142780253968481,
+      "loss": 0.4966,
+      "step": 1809
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005138248721892814,
+      "loss": 0.491,
+      "step": 1810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005133717076171095,
+      "loss": 0.4847,
+      "step": 1811
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005129185320528515,
+      "loss": 0.5344,
+      "step": 1812
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005124653458690365,
+      "loss": 0.5027,
+      "step": 1813
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005120121494382018,
+      "loss": 0.4807,
+      "step": 1814
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005115589431328931,
+      "loss": 0.47,
+      "step": 1815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005111057273256647,
+      "loss": 0.5126,
+      "step": 1816
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005106525023890783,
+      "loss": 0.4624,
+      "step": 1817
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005101992686957028,
+      "loss": 0.5035,
+      "step": 1818
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005097460266181149,
+      "loss": 0.516,
+      "step": 1819
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005092927765288981,
+      "loss": 0.4503,
+      "step": 1820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005088395188006422,
+      "loss": 0.4619,
+      "step": 1821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005083862538059435,
+      "loss": 0.4832,
+      "step": 1822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000507932981917404,
+      "loss": 0.5028,
+      "step": 1823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005074797035076318,
+      "loss": 0.4938,
+      "step": 1824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00050702641894924,
+      "loss": 0.4498,
+      "step": 1825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005065731286148471,
+      "loss": 0.5156,
+      "step": 1826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005061198328770761,
+      "loss": 0.5112,
+      "step": 1827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005056665321085542,
+      "loss": 0.4969,
+      "step": 1828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005052132266819132,
+      "loss": 0.4773,
+      "step": 1829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005047599169697884,
+      "loss": 0.4545,
+      "step": 1830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005043066033448189,
+      "loss": 0.4667,
+      "step": 1831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005038532861796469,
+      "loss": 0.5109,
+      "step": 1832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005033999658469174,
+      "loss": 0.5139,
+      "step": 1833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005029466427192782,
+      "loss": 0.4783,
+      "step": 1834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005024933171693791,
+      "loss": 0.4613,
+      "step": 1835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005020399895698721,
+      "loss": 0.4628,
+      "step": 1836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005015866602934111,
+      "loss": 0.4913,
+      "step": 1837
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005011333297126512,
+      "loss": 0.4883,
+      "step": 1838
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005006799982002482,
+      "loss": 0.4901,
+      "step": 1839
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000500226666128859,
+      "loss": 0.5221,
+      "step": 1840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004997733338711411,
+      "loss": 0.4733,
+      "step": 1841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004993200017997519,
+      "loss": 0.5063,
+      "step": 1842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000498866670287349,
+      "loss": 0.4663,
+      "step": 1843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004984133397065889,
+      "loss": 0.4955,
+      "step": 1844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000497960010430128,
+      "loss": 0.5049,
+      "step": 1845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000497506682830621,
+      "loss": 0.5032,
+      "step": 1846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004970533572807218,
+      "loss": 0.5139,
+      "step": 1847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004966000341530827,
+      "loss": 0.467,
+      "step": 1848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004961467138203532,
+      "loss": 0.4752,
+      "step": 1849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004956933966551813,
+      "loss": 0.4965,
+      "step": 1850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004952400830302117,
+      "loss": 0.4584,
+      "step": 1851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004947867733180871,
+      "loss": 0.4756,
+      "step": 1852
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004943334678914459,
+      "loss": 0.4779,
+      "step": 1853
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004938801671229242,
+      "loss": 0.5088,
+      "step": 1854
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004934268713851529,
+      "loss": 0.5264,
+      "step": 1855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004929735810507599,
+      "loss": 0.4747,
+      "step": 1856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004925202964923683,
+      "loss": 0.4857,
+      "step": 1857
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000492067018082596,
+      "loss": 0.4815,
+      "step": 1858
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004916137461940566,
+      "loss": 0.4749,
+      "step": 1859
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000491160481199358,
+      "loss": 0.4784,
+      "step": 1860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004907072234711021,
+      "loss": 0.4875,
+      "step": 1861
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004902539733818852,
+      "loss": 0.481,
+      "step": 1862
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004898007313042975,
+      "loss": 0.48,
+      "step": 1863
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004893474976109219,
+      "loss": 0.4614,
+      "step": 1864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004888942726743353,
+      "loss": 0.4822,
+      "step": 1865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004884410568671069,
+      "loss": 0.4751,
+      "step": 1866
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004879878505617983,
+      "loss": 0.4636,
+      "step": 1867
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048753465413096367,
+      "loss": 0.4739,
+      "step": 1868
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004870814679471485,
+      "loss": 0.4445,
+      "step": 1869
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048662829238289076,
+      "loss": 0.4835,
+      "step": 1870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004861751278107186,
+      "loss": 0.5212,
+      "step": 1871
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004857219746031519,
+      "loss": 0.4618,
+      "step": 1872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048526883313270103,
+      "loss": 0.5173,
+      "step": 1873
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048481570377186645,
+      "loss": 0.4535,
+      "step": 1874
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004843625868931392,
+      "loss": 0.4818,
+      "step": 1875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004839094828689994,
+      "loss": 0.502,
+      "step": 1876
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048345639207191703,
+      "loss": 0.4824,
+      "step": 1877
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004830033148743511,
+      "loss": 0.5103,
+      "step": 1878
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004825502516487497,
+      "loss": 0.5094,
+      "step": 1879
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048209720276754854,
+      "loss": 0.4882,
+      "step": 1880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048164416860317286,
+      "loss": 0.5178,
+      "step": 1881
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004811911495280345,
+      "loss": 0.4503,
+      "step": 1882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048073814591453396,
+      "loss": 0.4451,
+      "step": 1883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048028515813505854,
+      "loss": 0.4375,
+      "step": 1884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004798321865619822,
+      "loss": 0.5111,
+      "step": 1885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047937923156766646,
+      "loss": 0.5156,
+      "step": 1886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004789262935244581,
+      "loss": 0.506,
+      "step": 1887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047847337280469113,
+      "loss": 0.4958,
+      "step": 1888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047802046978068436,
+      "loss": 0.4495,
+      "step": 1889
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004775675848247427,
+      "loss": 0.4777,
+      "step": 1890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047711471830915554,
+      "loss": 0.4684,
+      "step": 1891
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047666187060619783,
+      "loss": 0.4546,
+      "step": 1892
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004762090420881289,
+      "loss": 0.421,
+      "step": 1893
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004757562331271918,
+      "loss": 0.4835,
+      "step": 1894
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004753034440956141,
+      "loss": 0.483,
+      "step": 1895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047485067536560645,
+      "loss": 0.5078,
+      "step": 1896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047439792730936357,
+      "loss": 0.4988,
+      "step": 1897
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047394520029906214,
+      "loss": 0.5275,
+      "step": 1898
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004734924947068626,
+      "loss": 0.4806,
+      "step": 1899
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004730398109049071,
+      "loss": 0.4205,
+      "step": 1900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047258714926531995,
+      "loss": 0.5123,
+      "step": 1901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047213451016020773,
+      "loss": 0.4572,
+      "step": 1902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047168189396165774,
+      "loss": 0.443,
+      "step": 1903
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004712293010417392,
+      "loss": 0.5244,
+      "step": 1904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004707767317725016,
+      "loss": 0.4834,
+      "step": 1905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047032418652597543,
+      "loss": 0.535,
+      "step": 1906
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046987166567417086,
+      "loss": 0.4854,
+      "step": 1907
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004694191695890788,
+      "loss": 0.5066,
+      "step": 1908
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046896669864266897,
+      "loss": 0.4464,
+      "step": 1909
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046851425320689117,
+      "loss": 0.5,
+      "step": 1910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046806183365367396,
+      "loss": 0.4866,
+      "step": 1911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046760944035492404,
+      "loss": 0.4592,
+      "step": 1912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004671570736825277,
+      "loss": 0.4905,
+      "step": 1913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046670473400834805,
+      "loss": 0.4473,
+      "step": 1914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004662524217042272,
+      "loss": 0.5134,
+      "step": 1915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004658001371419839,
+      "loss": 0.4968,
+      "step": 1916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046534788069341453,
+      "loss": 0.4895,
+      "step": 1917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046489565273029196,
+      "loss": 0.4969,
+      "step": 1918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046444345362436603,
+      "loss": 0.4652,
+      "step": 1919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004639912837473631,
+      "loss": 0.48,
+      "step": 1920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046353914347098467,
+      "loss": 0.4882,
+      "step": 1921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004630870331669088,
+      "loss": 0.4877,
+      "step": 1922
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004626349532067879,
+      "loss": 0.4487,
+      "step": 1923
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004621829039622506,
+      "loss": 0.4548,
+      "step": 1924
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004617308858048993,
+      "loss": 0.4955,
+      "step": 1925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046127889910631167,
+      "loss": 0.4609,
+      "step": 1926
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046082694423803873,
+      "loss": 0.4999,
+      "step": 1927
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046037502157160573,
+      "loss": 0.5026,
+      "step": 1928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004599231314785118,
+      "loss": 0.4987,
+      "step": 1929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004594712743302284,
+      "loss": 0.448,
+      "step": 1930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004590194504982009,
+      "loss": 0.4302,
+      "step": 1931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004585676603538465,
+      "loss": 0.4952,
+      "step": 1932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045811590426855515,
+      "loss": 0.4977,
+      "step": 1933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004576641826136884,
+      "loss": 0.4855,
+      "step": 1934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004572124957605803,
+      "loss": 0.4843,
+      "step": 1935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045676084408053513,
+      "loss": 0.4801,
+      "step": 1936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004563092279448293,
+      "loss": 0.495,
+      "step": 1937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004558576477247097,
+      "loss": 0.4625,
+      "step": 1938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000455406103791393,
+      "loss": 0.4669,
+      "step": 1939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004549545965160673,
+      "loss": 0.4974,
+      "step": 1940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045450312626988933,
+      "loss": 0.5033,
+      "step": 1941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045405169342398633,
+      "loss": 0.4872,
+      "step": 1942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004536002983494542,
+      "loss": 0.4362,
+      "step": 1943
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045314894141735803,
+      "loss": 0.5145,
+      "step": 1944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045269762299873144,
+      "loss": 0.4766,
+      "step": 1945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045224634346457656,
+      "loss": 0.4839,
+      "step": 1946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045179510318586356,
+      "loss": 0.4839,
+      "step": 1947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045134390253353016,
+      "loss": 0.4957,
+      "step": 1948
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045089274187848144,
+      "loss": 0.4785,
+      "step": 1949
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045044162159159,
+      "loss": 0.4217,
+      "step": 1950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044999054204369486,
+      "loss": 0.4683,
+      "step": 1951
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004495395036056016,
+      "loss": 0.4463,
+      "step": 1952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004490885066480824,
+      "loss": 0.496,
+      "step": 1953
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044863755154187467,
+      "loss": 0.4949,
+      "step": 1954
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004481866386576821,
+      "loss": 0.4469,
+      "step": 1955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044773576836617336,
+      "loss": 0.495,
+      "step": 1956
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004472849410379818,
+      "loss": 0.4742,
+      "step": 1957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004468341570437061,
+      "loss": 0.4786,
+      "step": 1958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004463834167539087,
+      "loss": 0.5016,
+      "step": 1959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004459327205391167,
+      "loss": 0.4718,
+      "step": 1960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044548206876982043,
+      "loss": 0.5017,
+      "step": 1961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004450314618164741,
+      "loss": 0.4336,
+      "step": 1962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044458090004949454,
+      "loss": 0.5107,
+      "step": 1963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000444130383839262,
+      "loss": 0.4508,
+      "step": 1964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044367991355611936,
+      "loss": 0.5021,
+      "step": 1965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000443229489570371,
+      "loss": 0.4927,
+      "step": 1966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004427791122522841,
+      "loss": 0.4894,
+      "step": 1967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004423287819720866,
+      "loss": 0.4447,
+      "step": 1968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044187849909996875,
+      "loss": 0.4955,
+      "step": 1969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004414282640060809,
+      "loss": 0.4557,
+      "step": 1970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004409780770605349,
+      "loss": 0.4745,
+      "step": 1971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044052793863340247,
+      "loss": 0.4689,
+      "step": 1972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004400778490947157,
+      "loss": 0.4417,
+      "step": 1973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004396278088144663,
+      "loss": 0.4562,
+      "step": 1974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043917781816260577,
+      "loss": 0.4426,
+      "step": 1975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004387278775090448,
+      "loss": 0.5065,
+      "step": 1976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043827798722365264,
+      "loss": 0.4696,
+      "step": 1977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004378281476762576,
+      "loss": 0.5078,
+      "step": 1978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043737835923664546,
+      "loss": 0.48,
+      "step": 1979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043692862227456125,
+      "loss": 0.4751,
+      "step": 1980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004364789371597064,
+      "loss": 0.4794,
+      "step": 1981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043602930426174085,
+      "loss": 0.5381,
+      "step": 1982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043557972395028065,
+      "loss": 0.4958,
+      "step": 1983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004351301965948991,
+      "loss": 0.4851,
+      "step": 1984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004346807225651262,
+      "loss": 0.4651,
+      "step": 1985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004342313022304474,
+      "loss": 0.4958,
+      "step": 1986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004337819359603049,
+      "loss": 0.4995,
+      "step": 1987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004333326241240957,
+      "loss": 0.5347,
+      "step": 1988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004328833670911724,
+      "loss": 0.4625,
+      "step": 1989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043243416523084234,
+      "loss": 0.458,
+      "step": 1990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000431985018912368,
+      "loss": 0.4674,
+      "step": 1991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043153592850496555,
+      "loss": 0.4801,
+      "step": 1992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043108689437780566,
+      "loss": 0.4553,
+      "step": 1993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004306379169000128,
+      "loss": 0.4665,
+      "step": 1994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004301889964406641,
+      "loss": 0.4397,
+      "step": 1995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042974013336879096,
+      "loss": 0.472,
+      "step": 1996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042929132805337656,
+      "loss": 0.4941,
+      "step": 1997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004288425808633575,
+      "loss": 0.4839,
+      "step": 1998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042839389216762185,
+      "loss": 0.4871,
+      "step": 1999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004279452623350101,
+      "loss": 0.469,
+      "step": 2000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042749669173431385,
+      "loss": 0.4578,
+      "step": 2001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004270481807342765,
+      "loss": 0.4993,
+      "step": 2002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042659972970359263,
+      "loss": 0.4735,
+      "step": 2003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042615133901090663,
+      "loss": 0.4663,
+      "step": 2004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042570300902481425,
+      "loss": 0.4841,
+      "step": 2005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042525474011386046,
+      "loss": 0.4846,
+      "step": 2006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004248065326465409,
+      "loss": 0.5074,
+      "step": 2007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042435838699129996,
+      "loss": 0.4709,
+      "step": 2008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042391030351653195,
+      "loss": 0.5013,
+      "step": 2009
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004234622825905792,
+      "loss": 0.4572,
+      "step": 2010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042301432458173317,
+      "loss": 0.4496,
+      "step": 2011
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004225664298582339,
+      "loss": 0.4604,
+      "step": 2012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004221185987882684,
+      "loss": 0.4492,
+      "step": 2013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042167083173997254,
+      "loss": 0.4895,
+      "step": 2014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004212231290814287,
+      "loss": 0.495,
+      "step": 2015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042077549118066653,
+      "loss": 0.4849,
+      "step": 2016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042032791840566273,
+      "loss": 0.4507,
+      "step": 2017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041988041112434027,
+      "loss": 0.4827,
+      "step": 2018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000419432969704568,
+      "loss": 0.4844,
+      "step": 2019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004189855945141612,
+      "loss": 0.4167,
+      "step": 2020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004185382859208803,
+      "loss": 0.425,
+      "step": 2021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041809104429243115,
+      "loss": 0.5103,
+      "step": 2022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004176438699964646,
+      "loss": 0.5165,
+      "step": 2023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041719676340057566,
+      "loss": 0.4901,
+      "step": 2024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041674972487230455,
+      "loss": 0.4982,
+      "step": 2025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004163027547791347,
+      "loss": 0.5012,
+      "step": 2026
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000415855853488494,
+      "loss": 0.5243,
+      "step": 2027
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004154090213677531,
+      "loss": 0.4647,
+      "step": 2028
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041496225878422615,
+      "loss": 0.4788,
+      "step": 2029
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041451556610517046,
+      "loss": 0.475,
+      "step": 2030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041406894369778503,
+      "loss": 0.4977,
+      "step": 2031
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000413622391929212,
+      "loss": 0.5126,
+      "step": 2032
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041317591116653486,
+      "loss": 0.5209,
+      "step": 2033
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041272950177677895,
+      "loss": 0.4751,
+      "step": 2034
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041228316412691057,
+      "loss": 0.5063,
+      "step": 2035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041183689858383793,
+      "loss": 0.4662,
+      "step": 2036
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041139070551440886,
+      "loss": 0.473,
+      "step": 2037
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041094458528541253,
+      "loss": 0.4485,
+      "step": 2038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004104985382635777,
+      "loss": 0.4795,
+      "step": 2039
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041005256481557305,
+      "loss": 0.4635,
+      "step": 2040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040960666530800706,
+      "loss": 0.5077,
+      "step": 2041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004091608401074268,
+      "loss": 0.4623,
+      "step": 2042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004087150895803192,
+      "loss": 0.5164,
+      "step": 2043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004082694140931089,
+      "loss": 0.493,
+      "step": 2044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040782381401215936,
+      "loss": 0.4926,
+      "step": 2045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004073782897037716,
+      "loss": 0.472,
+      "step": 2046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000406932841534185,
+      "loss": 0.5059,
+      "step": 2047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040648746986957605,
+      "loss": 0.4613,
+      "step": 2048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004060421750760581,
+      "loss": 0.4976,
+      "step": 2049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004055969575196818,
+      "loss": 0.4711,
+      "step": 2050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004051518175664336,
+      "loss": 0.4722,
+      "step": 2051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004047067555822371,
+      "loss": 0.5012,
+      "step": 2052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004042617719329507,
+      "loss": 0.4849,
+      "step": 2053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004038168669843697,
+      "loss": 0.4871,
+      "step": 2054
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004033720411022235,
+      "loss": 0.4745,
+      "step": 2055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000402927294652177,
+      "loss": 0.4346,
+      "step": 2056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040248262799983016,
+      "loss": 0.5341,
+      "step": 2057
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040203804151071665,
+      "loss": 0.4756,
+      "step": 2058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004015935355503049,
+      "loss": 0.4911,
+      "step": 2059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004011491104839967,
+      "loss": 0.5089,
+      "step": 2060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040070476667712743,
+      "loss": 0.4773,
+      "step": 2061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040026050449496554,
+      "loss": 0.4977,
+      "step": 2062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000399816324302713,
+      "loss": 0.4861,
+      "step": 2063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003993722264655034,
+      "loss": 0.4734,
+      "step": 2064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003989282113484036,
+      "loss": 0.4359,
+      "step": 2065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000398484279316412,
+      "loss": 0.4917,
+      "step": 2066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003980404307344584,
+      "loss": 0.4625,
+      "step": 2067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003975966659674047,
+      "loss": 0.5084,
+      "step": 2068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039715298538004325,
+      "loss": 0.4575,
+      "step": 2069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039670938933709774,
+      "loss": 0.5054,
+      "step": 2070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003962658782032219,
+      "loss": 0.4563,
+      "step": 2071
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000395822452343,
+      "loss": 0.4249,
+      "step": 2072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003953791121209458,
+      "loss": 0.4708,
+      "step": 2073
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039493585790150316,
+      "loss": 0.4871,
+      "step": 2074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003944926900490452,
+      "loss": 0.4958,
+      "step": 2075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003940496089278735,
+      "loss": 0.4736,
+      "step": 2076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039360661490221905,
+      "loss": 0.4796,
+      "step": 2077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003931637083362405,
+      "loss": 0.4384,
+      "step": 2078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039272088959402534,
+      "loss": 0.4739,
+      "step": 2079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003922781590395883,
+      "loss": 0.4779,
+      "step": 2080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039183551703687213,
+      "loss": 0.4529,
+      "step": 2081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003913929639497462,
+      "loss": 0.4684,
+      "step": 2082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039095050014200717,
+      "loss": 0.4832,
+      "step": 2083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003905081259773784,
+      "loss": 0.5115,
+      "step": 2084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003900658418195091,
+      "loss": 0.4618,
+      "step": 2085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003896236480319752,
+      "loss": 0.4485,
+      "step": 2086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003891815449782775,
+      "loss": 0.464,
+      "step": 2087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038873953302184284,
+      "loss": 0.4233,
+      "step": 2088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003882976125260229,
+      "loss": 0.4817,
+      "step": 2089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003878557838540944,
+      "loss": 0.4829,
+      "step": 2090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038741404736925813,
+      "loss": 0.4871,
+      "step": 2091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038697240343463956,
+      "loss": 0.5183,
+      "step": 2092
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000386530852413288,
+      "loss": 0.516,
+      "step": 2093
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003860893946681762,
+      "loss": 0.5054,
+      "step": 2094
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038564803056220033,
+      "loss": 0.4952,
+      "step": 2095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003852067604581794,
+      "loss": 0.5059,
+      "step": 2096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038476558471885577,
+      "loss": 0.4803,
+      "step": 2097
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003843245037068932,
+      "loss": 0.4641,
+      "step": 2098
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003838835177848788,
+      "loss": 0.4978,
+      "step": 2099
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003834426273153204,
+      "loss": 0.515,
+      "step": 2100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038300183266064806,
+      "loss": 0.4648,
+      "step": 2101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003825611341832126,
+      "loss": 0.488,
+      "step": 2102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003821205322452863,
+      "loss": 0.4678,
+      "step": 2103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003816800272090619,
+      "loss": 0.4709,
+      "step": 2104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003812396194366522,
+      "loss": 0.4952,
+      "step": 2105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003807993092900903,
+      "loss": 0.4984,
+      "step": 2106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003803590971313289,
+      "loss": 0.5062,
+      "step": 2107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003799189833222404,
+      "loss": 0.4481,
+      "step": 2108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000379478968224616,
+      "loss": 0.462,
+      "step": 2109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003790390522001662,
+      "loss": 0.4615,
+      "step": 2110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037859923561051967,
+      "loss": 0.4541,
+      "step": 2111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003781595188172233,
+      "loss": 0.472,
+      "step": 2112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037771990218174255,
+      "loss": 0.4545,
+      "step": 2113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003772803860654597,
+      "loss": 0.4701,
+      "step": 2114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037684097082967515,
+      "loss": 0.4839,
+      "step": 2115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037640165683560577,
+      "loss": 0.4585,
+      "step": 2116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003759624444443858,
+      "loss": 0.5077,
+      "step": 2117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037552333401706504,
+      "loss": 0.4955,
+      "step": 2118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037508432591461076,
+      "loss": 0.479,
+      "step": 2119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037464542049790484,
+      "loss": 0.4688,
+      "step": 2120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037420661812774574,
+      "loss": 0.5177,
+      "step": 2121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003737679191648468,
+      "loss": 0.4512,
+      "step": 2122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037332932396983587,
+      "loss": 0.4513,
+      "step": 2123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037289083290325663,
+      "loss": 0.5237,
+      "step": 2124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000372452446325566,
+      "loss": 0.492,
+      "step": 2125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003720141645971358,
+      "loss": 0.4951,
+      "step": 2126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037157598807825134,
+      "loss": 0.4761,
+      "step": 2127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003711379171291115,
+      "loss": 0.4819,
+      "step": 2128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037069995210982804,
+      "loss": 0.4996,
+      "step": 2129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003702620933804263,
+      "loss": 0.4478,
+      "step": 2130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036982434130084397,
+      "loss": 0.4332,
+      "step": 2131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036938669623093085,
+      "loss": 0.4531,
+      "step": 2132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036894915853044906,
+      "loss": 0.5129,
+      "step": 2133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000368511728559072,
+      "loss": 0.4841,
+      "step": 2134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036807440667638525,
+      "loss": 0.44,
+      "step": 2135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003676371932418847,
+      "loss": 0.5038,
+      "step": 2136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000367200088614978,
+      "loss": 0.4698,
+      "step": 2137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003667630931549826,
+      "loss": 0.4519,
+      "step": 2138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036632620722112637,
+      "loss": 0.4067,
+      "step": 2139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003658894311725476,
+      "loss": 0.5225,
+      "step": 2140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036545276536829353,
+      "loss": 0.5144,
+      "step": 2141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036501621016732155,
+      "loss": 0.5034,
+      "step": 2142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003645797659284975,
+      "loss": 0.4875,
+      "step": 2143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036414343301059633,
+      "loss": 0.4332,
+      "step": 2144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036370721177230114,
+      "loss": 0.4717,
+      "step": 2145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036327110257220385,
+      "loss": 0.475,
+      "step": 2146
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036283510576880354,
+      "loss": 0.4644,
+      "step": 2147
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036239922172050747,
+      "loss": 0.4351,
+      "step": 2148
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036196345078563015,
+      "loss": 0.4663,
+      "step": 2149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003615277933223925,
+      "loss": 0.4946,
+      "step": 2150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036109224968892316,
+      "loss": 0.4801,
+      "step": 2151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003606568202432562,
+      "loss": 0.4858,
+      "step": 2152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036022150534333264,
+      "loss": 0.4569,
+      "step": 2153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003597863053469987,
+      "loss": 0.4896,
+      "step": 2154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003593512206120065,
+      "loss": 0.4464,
+      "step": 2155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035891625149601344,
+      "loss": 0.4623,
+      "step": 2156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003584813983565815,
+      "loss": 0.5032,
+      "step": 2157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035804666155117804,
+      "loss": 0.5389,
+      "step": 2158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035761204143717383,
+      "loss": 0.5115,
+      "step": 2159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003571775383718447,
+      "loss": 0.4781,
+      "step": 2160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035674315271236947,
+      "loss": 0.4444,
+      "step": 2161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003563088848158309,
+      "loss": 0.5002,
+      "step": 2162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035587473503921455,
+      "loss": 0.469,
+      "step": 2163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003554407037394096,
+      "loss": 0.4399,
+      "step": 2164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003550067912732069,
+      "loss": 0.4614,
+      "step": 2165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035457299799730046,
+      "loss": 0.4558,
+      "step": 2166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003541393242682859,
+      "loss": 0.4539,
+      "step": 2167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003537057704426602,
+      "loss": 0.4805,
+      "step": 2168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035327233687682273,
+      "loss": 0.4906,
+      "step": 2169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000352839023927073,
+      "loss": 0.5132,
+      "step": 2170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003524058319496123,
+      "loss": 0.4874,
+      "step": 2171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003519727613005416,
+      "loss": 0.473,
+      "step": 2172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035153981233586274,
+      "loss": 0.5034,
+      "step": 2173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035110698541147704,
+      "loss": 0.473,
+      "step": 2174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000350674280883186,
+      "loss": 0.4719,
+      "step": 2175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003502416991066904,
+      "loss": 0.4594,
+      "step": 2176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034980924043758984,
+      "loss": 0.4459,
+      "step": 2177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000349376905231383,
+      "loss": 0.4692,
+      "step": 2178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034894469384346663,
+      "loss": 0.4929,
+      "step": 2179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003485126066291364,
+      "loss": 0.4962,
+      "step": 2180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034808064394358526,
+      "loss": 0.4641,
+      "step": 2181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003476488061419043,
+      "loss": 0.4771,
+      "step": 2182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003472170935790814,
+      "loss": 0.507,
+      "step": 2183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034678550661000195,
+      "loss": 0.4445,
+      "step": 2184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034635404558944807,
+      "loss": 0.4988,
+      "step": 2185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003459227108720978,
+      "loss": 0.4745,
+      "step": 2186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034549150281252633,
+      "loss": 0.4644,
+      "step": 2187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034506042176520377,
+      "loss": 0.48,
+      "step": 2188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034462946808449634,
+      "loss": 0.476,
+      "step": 2189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003441986421246653,
+      "loss": 0.4875,
+      "step": 2190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034376794423986737,
+      "loss": 0.4938,
+      "step": 2191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003433373747841532,
+      "loss": 0.4698,
+      "step": 2192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034290693411146876,
+      "loss": 0.4445,
+      "step": 2193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003424766225756537,
+      "loss": 0.4792,
+      "step": 2194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003420464405304412,
+      "loss": 0.4631,
+      "step": 2195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034161638832945886,
+      "loss": 0.5057,
+      "step": 2196
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003411864663262265,
+      "loss": 0.535,
+      "step": 2197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034075667487415786,
+      "loss": 0.4711,
+      "step": 2198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003403270143265587,
+      "loss": 0.4454,
+      "step": 2199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033989748503662754,
+      "loss": 0.5455,
+      "step": 2200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003394680873574546,
+      "loss": 0.4631,
+      "step": 2201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033903882164202243,
+      "loss": 0.5046,
+      "step": 2202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000338609688243205,
+      "loss": 0.4989,
+      "step": 2203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033818068751376705,
+      "loss": 0.507,
+      "step": 2204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033775181980636484,
+      "loss": 0.5179,
+      "step": 2205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033732308547354463,
+      "loss": 0.4899,
+      "step": 2206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033689448486774385,
+      "loss": 0.5142,
+      "step": 2207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003364660183412892,
+      "loss": 0.4536,
+      "step": 2208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033603768624639784,
+      "loss": 0.4602,
+      "step": 2209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033560948893517604,
+      "loss": 0.491,
+      "step": 2210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033518142675961917,
+      "loss": 0.4579,
+      "step": 2211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033475350007161197,
+      "loss": 0.4899,
+      "step": 2212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033432570922292725,
+      "loss": 0.4838,
+      "step": 2213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003338980545652267,
+      "loss": 0.4841,
+      "step": 2214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033347053645005966,
+      "loss": 0.5143,
+      "step": 2215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003330431552288633,
+      "loss": 0.514,
+      "step": 2216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003326159112529624,
+      "loss": 0.4991,
+      "step": 2217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033218880487356886,
+      "loss": 0.4799,
+      "step": 2218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033176183644178115,
+      "loss": 0.4541,
+      "step": 2219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033133500630858504,
+      "loss": 0.4432,
+      "step": 2220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033090831482485194,
+      "loss": 0.4955,
+      "step": 2221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033048176234133967,
+      "loss": 0.4768,
+      "step": 2222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033005534920869175,
+      "loss": 0.4799,
+      "step": 2223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003296290757774367,
+      "loss": 0.495,
+      "step": 2224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000329202942397989,
+      "loss": 0.4644,
+      "step": 2225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032877694942064716,
+      "loss": 0.4772,
+      "step": 2226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032835109719559516,
+      "loss": 0.475,
+      "step": 2227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032792538607290034,
+      "loss": 0.5007,
+      "step": 2228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003274998164025148,
+      "loss": 0.4893,
+      "step": 2229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003270743885342737,
+      "loss": 0.4464,
+      "step": 2230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003266491028178964,
+      "loss": 0.4365,
+      "step": 2231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003262239596029849,
+      "loss": 0.484,
+      "step": 2232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032579895923902407,
+      "loss": 0.4755,
+      "step": 2233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032537410207538153,
+      "loss": 0.5052,
+      "step": 2234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003249493884613069,
+      "loss": 0.4677,
+      "step": 2235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003245248187459323,
+      "loss": 0.4973,
+      "step": 2236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003241003932782709,
+      "loss": 0.4977,
+      "step": 2237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032367611240721793,
+      "loss": 0.4744,
+      "step": 2238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032325197648154936,
+      "loss": 0.4637,
+      "step": 2239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032282798584992197,
+      "loss": 0.5272,
+      "step": 2240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032240414086087357,
+      "loss": 0.5045,
+      "step": 2241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003219804418628216,
+      "loss": 0.4662,
+      "step": 2242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032155688920406414,
+      "loss": 0.4738,
+      "step": 2243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032113348323277837,
+      "loss": 0.4899,
+      "step": 2244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032071022429702137,
+      "loss": 0.4963,
+      "step": 2245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003202871127447288,
+      "loss": 0.4546,
+      "step": 2246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000319864148923716,
+      "loss": 0.4918,
+      "step": 2247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003194413331816759,
+      "loss": 0.5132,
+      "step": 2248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031901866586618057,
+      "loss": 0.5153,
+      "step": 2249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031859614732467957,
+      "loss": 0.4636,
+      "step": 2250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003181737779045001,
+      "loss": 0.4777,
+      "step": 2251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031775155795284713,
+      "loss": 0.4921,
+      "step": 2252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003173294878168025,
+      "loss": 0.4391,
+      "step": 2253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003169075678433251,
+      "loss": 0.4364,
+      "step": 2254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031648579837925006,
+      "loss": 0.4613,
+      "step": 2255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003160641797712891,
+      "loss": 0.4825,
+      "step": 2256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003156427123660297,
+      "loss": 0.4459,
+      "step": 2257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003152213965099352,
+      "loss": 0.4451,
+      "step": 2258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003148002325493445,
+      "loss": 0.549,
+      "step": 2259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003143792208304712,
+      "loss": 0.4551,
+      "step": 2260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003139583616994043,
+      "loss": 0.5081,
+      "step": 2261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031353765550210655,
+      "loss": 0.4283,
+      "step": 2262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003131171025844161,
+      "loss": 0.4866,
+      "step": 2263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031269670329204396,
+      "loss": 0.5071,
+      "step": 2264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003122764579705759,
+      "loss": 0.5081,
+      "step": 2265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003118563669654703,
+      "loss": 0.4436,
+      "step": 2266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003114364306220589,
+      "loss": 0.4949,
+      "step": 2267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003110166492855468,
+      "loss": 0.4678,
+      "step": 2268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031059702330101083,
+      "loss": 0.4746,
+      "step": 2269
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000310177553013401,
+      "loss": 0.4628,
+      "step": 2270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003097582387675385,
+      "loss": 0.4927,
+      "step": 2271
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030933908090811696,
+      "loss": 0.4799,
+      "step": 2272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003089200797797008,
+      "loss": 0.4635,
+      "step": 2273
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030850123572672626,
+      "loss": 0.4674,
+      "step": 2274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003080825490934999,
+      "loss": 0.463,
+      "step": 2275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003076640202241992,
+      "loss": 0.4808,
+      "step": 2276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030724564946287204,
+      "loss": 0.4677,
+      "step": 2277
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030682743715343565,
+      "loss": 0.4996,
+      "step": 2278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030640938363967813,
+      "loss": 0.5001,
+      "step": 2279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003059914892652559,
+      "loss": 0.4509,
+      "step": 2280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003055737543736954,
+      "loss": 0.4637,
+      "step": 2281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003051561793083915,
+      "loss": 0.5131,
+      "step": 2282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030473876441260785,
+      "loss": 0.4666,
+      "step": 2283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003043215100294765,
+      "loss": 0.5039,
+      "step": 2284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030390441650199725,
+      "loss": 0.4991,
+      "step": 2285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003034874841730382,
+      "loss": 0.4583,
+      "step": 2286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030307071338533443,
+      "loss": 0.4652,
+      "step": 2287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003026541044814885,
+      "loss": 0.5029,
+      "step": 2288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030223765780396973,
+      "loss": 0.4827,
+      "step": 2289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003018213736951144,
+      "loss": 0.4696,
+      "step": 2290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003014052524971245,
+      "loss": 0.483,
+      "step": 2291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030098929455206903,
+      "loss": 0.431,
+      "step": 2292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003005735002018818,
+      "loss": 0.4696,
+      "step": 2293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000300157869788363,
+      "loss": 0.5117,
+      "step": 2294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029974240365317754,
+      "loss": 0.4787,
+      "step": 2295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002993271021378553,
+      "loss": 0.469,
+      "step": 2296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002989119655837913,
+      "loss": 0.4863,
+      "step": 2297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002984969943322442,
+      "loss": 0.4865,
+      "step": 2298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029808218872433767,
+      "loss": 0.4694,
+      "step": 2299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002976675491010583,
+      "loss": 0.4808,
+      "step": 2300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000297253075803257,
+      "loss": 0.4945,
+      "step": 2301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002968387691716472,
+      "loss": 0.4641,
+      "step": 2302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029642462954680606,
+      "loss": 0.4705,
+      "step": 2303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002960106572691733,
+      "loss": 0.4802,
+      "step": 2304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002955968526790506,
+      "loss": 0.479,
+      "step": 2305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002951832161166024,
+      "loss": 0.4559,
+      "step": 2306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002947697479218543,
+      "loss": 0.4401,
+      "step": 2307
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029435644843469434,
+      "loss": 0.4203,
+      "step": 2308
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029394331799487106,
+      "loss": 0.4584,
+      "step": 2309
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002935303569419949,
+      "loss": 0.4696,
+      "step": 2310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029311756561553625,
+      "loss": 0.4951,
+      "step": 2311
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029270494435482635,
+      "loss": 0.4844,
+      "step": 2312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002922924934990568,
+      "loss": 0.478,
+      "step": 2313
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029188021338727877,
+      "loss": 0.4734,
+      "step": 2314
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029146810435840334,
+      "loss": 0.4299,
+      "step": 2315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002910561667512005,
+      "loss": 0.4645,
+      "step": 2316
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002906444009043001,
+      "loss": 0.484,
+      "step": 2317
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029023280715619005,
+      "loss": 0.4828,
+      "step": 2318
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002898213858452173,
+      "loss": 0.4694,
+      "step": 2319
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002894101373095867,
+      "loss": 0.4597,
+      "step": 2320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028899906188736125,
+      "loss": 0.4854,
+      "step": 2321
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028858815991646185,
+      "loss": 0.4284,
+      "step": 2322
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002881774317346664,
+      "loss": 0.4973,
+      "step": 2323
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002877668776796099,
+      "loss": 0.4377,
+      "step": 2324
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028735649808878485,
+      "loss": 0.4412,
+      "step": 2325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028694629329953966,
+      "loss": 0.483,
+      "step": 2326
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002865362636490791,
+      "loss": 0.4559,
+      "step": 2327
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002861264094744647,
+      "loss": 0.4856,
+      "step": 2328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028571673111261266,
+      "loss": 0.4539,
+      "step": 2329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028530722890029536,
+      "loss": 0.4543,
+      "step": 2330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002848979031741406,
+      "loss": 0.449,
+      "step": 2331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002844887542706303,
+      "loss": 0.4264,
+      "step": 2332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028407978252610166,
+      "loss": 0.4734,
+      "step": 2333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028367098827674573,
+      "loss": 0.4564,
+      "step": 2334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028326237185860847,
+      "loss": 0.4802,
+      "step": 2335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002828539336075888,
+      "loss": 0.4379,
+      "step": 2336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028244567385943964,
+      "loss": 0.459,
+      "step": 2337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002820375929497668,
+      "loss": 0.4635,
+      "step": 2338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028162969121402946,
+      "loss": 0.5398,
+      "step": 2339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002812219689875397,
+      "loss": 0.5032,
+      "step": 2340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028081442660546124,
+      "loss": 0.5277,
+      "step": 2341
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028040706440281087,
+      "loss": 0.4535,
+      "step": 2342
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027999988271445645,
+      "loss": 0.4097,
+      "step": 2343
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027959288187511823,
+      "loss": 0.4562,
+      "step": 2344
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000279186062219367,
+      "loss": 0.4423,
+      "step": 2345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002787794240816254,
+      "loss": 0.4814,
+      "step": 2346
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027837296779616627,
+      "loss": 0.4729,
+      "step": 2347
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002779666936971129,
+      "loss": 0.4796,
+      "step": 2348
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027756060211843957,
+      "loss": 0.4838,
+      "step": 2349
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002771546933939696,
+      "loss": 0.5179,
+      "step": 2350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027674896785737687,
+      "loss": 0.4927,
+      "step": 2351
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027634342584218364,
+      "loss": 0.4694,
+      "step": 2352
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002759380676817624,
+      "loss": 0.4938,
+      "step": 2353
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027553289370933363,
+      "loss": 0.4945,
+      "step": 2354
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002751279042579672,
+      "loss": 0.5028,
+      "step": 2355
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002747230996605803,
+      "loss": 0.504,
+      "step": 2356
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002743184802499392,
+      "loss": 0.4982,
+      "step": 2357
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027391404635865725,
+      "loss": 0.4368,
+      "step": 2358
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002735097983191954,
+      "loss": 0.4794,
+      "step": 2359
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027310573646386217,
+      "loss": 0.5167,
+      "step": 2360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002727018611248125,
+      "loss": 0.4673,
+      "step": 2361
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027229817263404863,
+      "loss": 0.4176,
+      "step": 2362
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002718946713234185,
+      "loss": 0.4625,
+      "step": 2363
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027149135752461696,
+      "loss": 0.452,
+      "step": 2364
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027108823156918386,
+      "loss": 0.4625,
+      "step": 2365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027068529378850557,
+      "loss": 0.4777,
+      "step": 2366
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002702825445138127,
+      "loss": 0.4771,
+      "step": 2367
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026987998407618214,
+      "loss": 0.4806,
+      "step": 2368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002694776128065345,
+      "loss": 0.5129,
+      "step": 2369
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026907543103563516,
+      "loss": 0.4351,
+      "step": 2370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002686734390940941,
+      "loss": 0.4613,
+      "step": 2371
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002682716373123648,
+      "loss": 0.493,
+      "step": 2372
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002678700260207449,
+      "loss": 0.4517,
+      "step": 2373
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002674686055493748,
+      "loss": 0.4655,
+      "step": 2374
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026706737622823873,
+      "loss": 0.4408,
+      "step": 2375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026666633838716316,
+      "loss": 0.4637,
+      "step": 2376
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002662654923558176,
+      "loss": 0.4528,
+      "step": 2377
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000265864838463714,
+      "loss": 0.4726,
+      "step": 2378
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002654643770402058,
+      "loss": 0.4729,
+      "step": 2379
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002650641084144885,
+      "loss": 0.4934,
+      "step": 2380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026466403291559913,
+      "loss": 0.4573,
+      "step": 2381
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026426415087241623,
+      "loss": 0.4698,
+      "step": 2382
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002638644626136587,
+      "loss": 0.5118,
+      "step": 2383
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026346496846788687,
+      "loss": 0.4669,
+      "step": 2384
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002630656687635007,
+      "loss": 0.4529,
+      "step": 2385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000262666563828741,
+      "loss": 0.493,
+      "step": 2386
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026226765399168833,
+      "loss": 0.4943,
+      "step": 2387
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002618689395802624,
+      "loss": 0.4814,
+      "step": 2388
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026147042092222305,
+      "loss": 0.4897,
+      "step": 2389
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026107209834516854,
+      "loss": 0.479,
+      "step": 2390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000260673972176536,
+      "loss": 0.4883,
+      "step": 2391
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026027604274360156,
+      "loss": 0.4313,
+      "step": 2392
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002598783103734793,
+      "loss": 0.4692,
+      "step": 2393
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002594807753931211,
+      "loss": 0.4811,
+      "step": 2394
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000259083438129317,
+      "loss": 0.4636,
+      "step": 2395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025868629890869467,
+      "loss": 0.4006,
+      "step": 2396
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000258289358057718,
+      "loss": 0.4824,
+      "step": 2397
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002578926159026891,
+      "loss": 0.4309,
+      "step": 2398
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025749607276974565,
+      "loss": 0.4792,
+      "step": 2399
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002570997289848625,
+      "loss": 0.4415,
+      "step": 2400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025670358487385,
+      "loss": 0.4904,
+      "step": 2401
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000256307640762355,
+      "loss": 0.4619,
+      "step": 2402
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025591189697585947,
+      "loss": 0.4369,
+      "step": 2403
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025551635383968066,
+      "loss": 0.457,
+      "step": 2404
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025512101167897135,
+      "loss": 0.4811,
+      "step": 2405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025472587081871854,
+      "loss": 0.5002,
+      "step": 2406
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002543309315837444,
+      "loss": 0.5015,
+      "step": 2407
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002539361942987046,
+      "loss": 0.4795,
+      "step": 2408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002535416592880896,
+      "loss": 0.4888,
+      "step": 2409
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002531473268762227,
+      "loss": 0.4836,
+      "step": 2410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002527531973872617,
+      "loss": 0.4464,
+      "step": 2411
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002523592711451964,
+      "loss": 0.4789,
+      "step": 2412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002519655484738507,
+      "loss": 0.4832,
+      "step": 2413
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002515720296968803,
+      "loss": 0.4639,
+      "step": 2414
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002511787151377735,
+      "loss": 0.4489,
+      "step": 2415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025078560511985115,
+      "loss": 0.472,
+      "step": 2416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025039269996626523,
+      "loss": 0.5028,
+      "step": 2417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002500000000000001,
+      "loss": 0.4596,
+      "step": 2418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002496075055438707,
+      "loss": 0.4479,
+      "step": 2419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002492152169205237,
+      "loss": 0.4827,
+      "step": 2420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024882313445243585,
+      "loss": 0.4932,
+      "step": 2421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024843125846191514,
+      "loss": 0.4485,
+      "step": 2422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024803958927109966,
+      "loss": 0.4515,
+      "step": 2423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024764812720195714,
+      "loss": 0.4952,
+      "step": 2424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002472568725762853,
+      "loss": 0.4408,
+      "step": 2425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024686582571571117,
+      "loss": 0.4897,
+      "step": 2426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002464749869416914,
+      "loss": 0.5193,
+      "step": 2427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024608435657551096,
+      "loss": 0.4955,
+      "step": 2428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002456939349382843,
+      "loss": 0.4539,
+      "step": 2429
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002453037223509534,
+      "loss": 0.4829,
+      "step": 2430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000244913719134289,
+      "loss": 0.4927,
+      "step": 2431
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024452392560888974,
+      "loss": 0.4862,
+      "step": 2432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024413434209518138,
+      "loss": 0.5088,
+      "step": 2433
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024374496891341768,
+      "loss": 0.4829,
+      "step": 2434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024335580638367905,
+      "loss": 0.4828,
+      "step": 2435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024296685482587282,
+      "loss": 0.4547,
+      "step": 2436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002425781145597328,
+      "loss": 0.4741,
+      "step": 2437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002421895859048196,
+      "loss": 0.465,
+      "step": 2438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024180126918051909,
+      "loss": 0.4874,
+      "step": 2439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024141316470604362,
+      "loss": 0.4829,
+      "step": 2440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024102527280043103,
+      "loss": 0.4644,
+      "step": 2441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024063759378254373,
+      "loss": 0.4567,
+      "step": 2442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024025012797106998,
+      "loss": 0.4846,
+      "step": 2443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023986287568452208,
+      "loss": 0.4774,
+      "step": 2444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023947583724123745,
+      "loss": 0.4901,
+      "step": 2445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023908901295937712,
+      "loss": 0.4728,
+      "step": 2446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023870240315692644,
+      "loss": 0.4576,
+      "step": 2447
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023831600815169408,
+      "loss": 0.5087,
+      "step": 2448
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023792982826131265,
+      "loss": 0.4753,
+      "step": 2449
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023754386380323779,
+      "loss": 0.4764,
+      "step": 2450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002371581150947476,
+      "loss": 0.4802,
+      "step": 2451
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002367725824529436,
+      "loss": 0.5125,
+      "step": 2452
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023638726619474876,
+      "loss": 0.4697,
+      "step": 2453
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002360021666369091,
+      "loss": 0.4399,
+      "step": 2454
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002356172840959917,
+      "loss": 0.4903,
+      "step": 2455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023523261888838592,
+      "loss": 0.473,
+      "step": 2456
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023484817133030206,
+      "loss": 0.4725,
+      "step": 2457
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023446394173777142,
+      "loss": 0.4816,
+      "step": 2458
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023407993042664656,
+      "loss": 0.5209,
+      "step": 2459
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023369613771260007,
+      "loss": 0.4667,
+      "step": 2460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023331256391112548,
+      "loss": 0.4805,
+      "step": 2461
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023292920933753563,
+      "loss": 0.4939,
+      "step": 2462
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002325460743069639,
+      "loss": 0.4664,
+      "step": 2463
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023216315913436242,
+      "loss": 0.4543,
+      "step": 2464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023178046413450333,
+      "loss": 0.4664,
+      "step": 2465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023139798962197716,
+      "loss": 0.4933,
+      "step": 2466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002310157359111938,
+      "loss": 0.5087,
+      "step": 2467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023063370331638083,
+      "loss": 0.5139,
+      "step": 2468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023025189215158509,
+      "loss": 0.4772,
+      "step": 2469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002298703027306705,
+      "loss": 0.4795,
+      "step": 2470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022948893536731891,
+      "loss": 0.472,
+      "step": 2471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022910779037503003,
+      "loss": 0.4591,
+      "step": 2472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022872686806712033,
+      "loss": 0.4382,
+      "step": 2473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002283461687567236,
+      "loss": 0.4713,
+      "step": 2474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022796569275678973,
+      "loss": 0.4457,
+      "step": 2475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002275854403800857,
+      "loss": 0.4619,
+      "step": 2476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022720541193919447,
+      "loss": 0.5037,
+      "step": 2477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022682560774651457,
+      "loss": 0.4636,
+      "step": 2478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002264460281142607,
+      "loss": 0.4385,
+      "step": 2479
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022606667335446257,
+      "loss": 0.4564,
+      "step": 2480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022568754377896516,
+      "loss": 0.4685,
+      "step": 2481
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002253086396994282,
+      "loss": 0.4717,
+      "step": 2482
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022492996142732657,
+      "loss": 0.4402,
+      "step": 2483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002245515092739488,
+      "loss": 0.4839,
+      "step": 2484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022417328355039818,
+      "loss": 0.4597,
+      "step": 2485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022379528456759135,
+      "loss": 0.4648,
+      "step": 2486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022341751263625886,
+      "loss": 0.4448,
+      "step": 2487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022303996806694487,
+      "loss": 0.479,
+      "step": 2488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002226626511700059,
+      "loss": 0.4972,
+      "step": 2489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022228556225561204,
+      "loss": 0.486,
+      "step": 2490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022190870163374556,
+      "loss": 0.4564,
+      "step": 2491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022153206961420113,
+      "loss": 0.4873,
+      "step": 2492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022115566650658536,
+      "loss": 0.4642,
+      "step": 2493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022077949262031728,
+      "loss": 0.4429,
+      "step": 2494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022040354826462666,
+      "loss": 0.4752,
+      "step": 2495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022002783374855517,
+      "loss": 0.4653,
+      "step": 2496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021965234938095562,
+      "loss": 0.4741,
+      "step": 2497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021927709547049095,
+      "loss": 0.4567,
+      "step": 2498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021890207232563554,
+      "loss": 0.4678,
+      "step": 2499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021852728025467316,
+      "loss": 0.4547,
+      "step": 2500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021815271956569855,
+      "loss": 0.4528,
+      "step": 2501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021777839056661552,
+      "loss": 0.4515,
+      "step": 2502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002174042935651377,
+      "loss": 0.4724,
+      "step": 2503
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021703042886878781,
+      "loss": 0.4618,
+      "step": 2504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021665679678489804,
+      "loss": 0.467,
+      "step": 2505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021628339762060916,
+      "loss": 0.5095,
+      "step": 2506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021591023168287017,
+      "loss": 0.4806,
+      "step": 2507
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002155372992784389,
+      "loss": 0.4686,
+      "step": 2508
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002151646007138806,
+      "loss": 0.4633,
+      "step": 2509
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021479213629556893,
+      "loss": 0.4965,
+      "step": 2510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021441990632968438,
+      "loss": 0.4894,
+      "step": 2511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021404791112221544,
+      "loss": 0.4969,
+      "step": 2512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021367615097895703,
+      "loss": 0.5089,
+      "step": 2513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021330462620551094,
+      "loss": 0.4589,
+      "step": 2514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021293333710728596,
+      "loss": 0.4956,
+      "step": 2515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002125622839894964,
+      "loss": 0.4993,
+      "step": 2516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002121914671571633,
+      "loss": 0.4203,
+      "step": 2517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021182088691511287,
+      "loss": 0.4485,
+      "step": 2518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021145054356797727,
+      "loss": 0.4297,
+      "step": 2519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021108043742019356,
+      "loss": 0.4167,
+      "step": 2520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021071056877600432,
+      "loss": 0.4733,
+      "step": 2521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021034093793945613,
+      "loss": 0.4861,
+      "step": 2522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020997154521440098,
+      "loss": 0.4669,
+      "step": 2523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002096023909044944,
+      "loss": 0.4542,
+      "step": 2524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020923347531319613,
+      "loss": 0.531,
+      "step": 2525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020886479874377,
+      "loss": 0.4666,
+      "step": 2526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002084963614992828,
+      "loss": 0.4563,
+      "step": 2527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020812816388260518,
+      "loss": 0.4867,
+      "step": 2528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020776020619641024,
+      "loss": 0.5198,
+      "step": 2529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002073924887431744,
+      "loss": 0.4877,
+      "step": 2530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020702501182517596,
+      "loss": 0.4706,
+      "step": 2531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020665777574449608,
+      "loss": 0.4409,
+      "step": 2532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020629078080301784,
+      "loss": 0.4565,
+      "step": 2533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020592402730242577,
+      "loss": 0.4442,
+      "step": 2534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000205557515544206,
+      "loss": 0.4641,
+      "step": 2535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020519124582964632,
+      "loss": 0.5002,
+      "step": 2536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020482521845983521,
+      "loss": 0.4717,
+      "step": 2537
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020445943373566177,
+      "loss": 0.4568,
+      "step": 2538
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020409389195781625,
+      "loss": 0.4525,
+      "step": 2539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020372859342678845,
+      "loss": 0.495,
+      "step": 2540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020336353844286877,
+      "loss": 0.501,
+      "step": 2541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020299872730614733,
+      "loss": 0.4615,
+      "step": 2542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020263416031651333,
+      "loss": 0.5059,
+      "step": 2543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020226983777365604,
+      "loss": 0.4828,
+      "step": 2544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002019057599770629,
+      "loss": 0.4922,
+      "step": 2545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002015419272260211,
+      "loss": 0.4913,
+      "step": 2546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020117833981961563,
+      "loss": 0.4945,
+      "step": 2547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020081499805673015,
+      "loss": 0.4853,
+      "step": 2548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020045190223604608,
+      "loss": 0.4543,
+      "step": 2549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020008905265604316,
+      "loss": 0.4745,
+      "step": 2550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019972644961499853,
+      "loss": 0.4197,
+      "step": 2551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019936409341098633,
+      "loss": 0.5165,
+      "step": 2552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019900198434187838,
+      "loss": 0.4816,
+      "step": 2553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019864012270534259,
+      "loss": 0.4744,
+      "step": 2554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001982785087988444,
+      "loss": 0.4827,
+      "step": 2555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019791714291964463,
+      "loss": 0.4507,
+      "step": 2556
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019755602536480104,
+      "loss": 0.462,
+      "step": 2557
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019719515643116677,
+      "loss": 0.516,
+      "step": 2558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019683453641539052,
+      "loss": 0.5269,
+      "step": 2559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001964741656139169,
+      "loss": 0.4487,
+      "step": 2560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000196114044322985,
+      "loss": 0.4626,
+      "step": 2561
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001957541728386295,
+      "loss": 0.4615,
+      "step": 2562
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001953945514566789,
+      "loss": 0.4733,
+      "step": 2563
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019503518047275697,
+      "loss": 0.5354,
+      "step": 2564
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001946760601822809,
+      "loss": 0.4703,
+      "step": 2565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019431719088046245,
+      "loss": 0.4907,
+      "step": 2566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019395857286230638,
+      "loss": 0.4479,
+      "step": 2567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019360020642261155,
+      "loss": 0.4774,
+      "step": 2568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019324209185596963,
+      "loss": 0.4729,
+      "step": 2569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001928842294567651,
+      "loss": 0.4336,
+      "step": 2570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019252661951917567,
+      "loss": 0.4998,
+      "step": 2571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019216926233717085,
+      "loss": 0.4751,
+      "step": 2572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001918121582045132,
+      "loss": 0.5071,
+      "step": 2573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019145530741475632,
+      "loss": 0.4836,
+      "step": 2574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019109871026124636,
+      "loss": 0.4475,
+      "step": 2575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019074236703712038,
+      "loss": 0.4553,
+      "step": 2576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019038627803530712,
+      "loss": 0.4633,
+      "step": 2577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019003044354852633,
+      "loss": 0.4496,
+      "step": 2578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018967486386928817,
+      "loss": 0.4829,
+      "step": 2579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018931953928989366,
+      "loss": 0.4618,
+      "step": 2580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018896447010243367,
+      "loss": 0.5024,
+      "step": 2581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018860965659878997,
+      "loss": 0.4867,
+      "step": 2582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018825509907063325,
+      "loss": 0.4521,
+      "step": 2583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018790079780942455,
+      "loss": 0.4835,
+      "step": 2584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018754675310641344,
+      "loss": 0.4663,
+      "step": 2585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018719296525263924,
+      "loss": 0.4923,
+      "step": 2586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001868394345389301,
+      "loss": 0.4648,
+      "step": 2587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001864861612559022,
+      "loss": 0.5145,
+      "step": 2588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018613314569396088,
+      "loss": 0.4779,
+      "step": 2589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001857803881432991,
+      "loss": 0.4756,
+      "step": 2590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018542788889389789,
+      "loss": 0.4993,
+      "step": 2591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018507564823552563,
+      "loss": 0.5128,
+      "step": 2592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001847236664577389,
+      "loss": 0.4766,
+      "step": 2593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001843719438498806,
+      "loss": 0.429,
+      "step": 2594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018402048070108106,
+      "loss": 0.5406,
+      "step": 2595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018366927730025756,
+      "loss": 0.5288,
+      "step": 2596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018331833393611308,
+      "loss": 0.4896,
+      "step": 2597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018296765089713767,
+      "loss": 0.4856,
+      "step": 2598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018261722847160672,
+      "loss": 0.4779,
+      "step": 2599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018226706694758193,
+      "loss": 0.4492,
+      "step": 2600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018191716661291013,
+      "loss": 0.4395,
+      "step": 2601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018156752775522338,
+      "loss": 0.4635,
+      "step": 2602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001812181506619394,
+      "loss": 0.4624,
+      "step": 2603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018086903562025998,
+      "loss": 0.476,
+      "step": 2604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018052018291717214,
+      "loss": 0.4838,
+      "step": 2605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018017159283944655,
+      "loss": 0.504,
+      "step": 2606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001798232656736389,
+      "loss": 0.4974,
+      "step": 2607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017947520170608772,
+      "loss": 0.4412,
+      "step": 2608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017912740122291615,
+      "loss": 0.4717,
+      "step": 2609
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017877986451002992,
+      "loss": 0.4966,
+      "step": 2610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017843259185311867,
+      "loss": 0.4564,
+      "step": 2611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001780855835376542,
+      "loss": 0.4775,
+      "step": 2612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017773883984889178,
+      "loss": 0.4946,
+      "step": 2613
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017739236107186857,
+      "loss": 0.4486,
+      "step": 2614
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017704614749140414,
+      "loss": 0.4631,
+      "step": 2615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017670019939210026,
+      "loss": 0.4542,
+      "step": 2616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017635451705834006,
+      "loss": 0.4403,
+      "step": 2617
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001760091007742888,
+      "loss": 0.4933,
+      "step": 2618
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001756639508238922,
+      "loss": 0.489,
+      "step": 2619
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017531906749087794,
+      "loss": 0.4543,
+      "step": 2620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017497445105875377,
+      "loss": 0.472,
+      "step": 2621
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017463010181080867,
+      "loss": 0.4539,
+      "step": 2622
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017428602003011134,
+      "loss": 0.4911,
+      "step": 2623
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017394220599951138,
+      "loss": 0.4924,
+      "step": 2624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017359866000163758,
+      "loss": 0.5021,
+      "step": 2625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017325538231889858,
+      "loss": 0.4648,
+      "step": 2626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017291237323348286,
+      "loss": 0.5139,
+      "step": 2627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001725696330273575,
+      "loss": 0.4663,
+      "step": 2628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017222716198226913,
+      "loss": 0.4941,
+      "step": 2629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001718849603797425,
+      "loss": 0.4692,
+      "step": 2630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017154302850108156,
+      "loss": 0.4597,
+      "step": 2631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017120136662736784,
+      "loss": 0.4984,
+      "step": 2632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017085997503946143,
+      "loss": 0.5018,
+      "step": 2633
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017051885401800015,
+      "loss": 0.4802,
+      "step": 2634
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017017800384339925,
+      "loss": 0.4722,
+      "step": 2635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001698374247958514,
+      "loss": 0.4694,
+      "step": 2636
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016949711715532606,
+      "loss": 0.4781,
+      "step": 2637
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016915708120157042,
+      "loss": 0.4769,
+      "step": 2638
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016881731721410736,
+      "loss": 0.466,
+      "step": 2639
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001684778254722371,
+      "loss": 0.4583,
+      "step": 2640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016813860625503518,
+      "loss": 0.4955,
+      "step": 2641
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016779965984135375,
+      "loss": 0.495,
+      "step": 2642
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016746098650982072,
+      "loss": 0.4733,
+      "step": 2643
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001671225865388389,
+      "loss": 0.4677,
+      "step": 2644
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016678446020658712,
+      "loss": 0.5129,
+      "step": 2645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016644660779101885,
+      "loss": 0.5072,
+      "step": 2646
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016610902956986235,
+      "loss": 0.4256,
+      "step": 2647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001657717258206205,
+      "loss": 0.4879,
+      "step": 2648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016543469682057105,
+      "loss": 0.4441,
+      "step": 2649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016509794284676505,
+      "loss": 0.4474,
+      "step": 2650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001647614641760282,
+      "loss": 0.4979,
+      "step": 2651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001644252610849597,
+      "loss": 0.4924,
+      "step": 2652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016408933384993185,
+      "loss": 0.4718,
+      "step": 2653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016375368274709068,
+      "loss": 0.4938,
+      "step": 2654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001634183080523548,
+      "loss": 0.4825,
+      "step": 2655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016308321004141607,
+      "loss": 0.4824,
+      "step": 2656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016274838898973847,
+      "loss": 0.4718,
+      "step": 2657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016241384517255852,
+      "loss": 0.4541,
+      "step": 2658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001620795788648846,
+      "loss": 0.4217,
+      "step": 2659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001617455903414974,
+      "loss": 0.452,
+      "step": 2660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001614118798769491,
+      "loss": 0.4604,
+      "step": 2661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016107844774556295,
+      "loss": 0.4797,
+      "step": 2662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016074529422143398,
+      "loss": 0.4414,
+      "step": 2663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001604124195784276,
+      "loss": 0.4683,
+      "step": 2664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016007982409018057,
+      "loss": 0.4901,
+      "step": 2665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015974750803009953,
+      "loss": 0.4948,
+      "step": 2666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015941547167136211,
+      "loss": 0.4695,
+      "step": 2667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015908371528691552,
+      "loss": 0.4528,
+      "step": 2668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015875223914947674,
+      "loss": 0.4862,
+      "step": 2669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015842104353153285,
+      "loss": 0.5079,
+      "step": 2670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015809012870533995,
+      "loss": 0.4778,
+      "step": 2671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015775949494292363,
+      "loss": 0.4678,
+      "step": 2672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015742914251607793,
+      "loss": 0.4955,
+      "step": 2673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015709907169636627,
+      "loss": 0.4641,
+      "step": 2674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015676928275511998,
+      "loss": 0.4679,
+      "step": 2675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015643977596343917,
+      "loss": 0.4492,
+      "step": 2676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001561105515921915,
+      "loss": 0.5144,
+      "step": 2677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001557816099120131,
+      "loss": 0.4602,
+      "step": 2678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001554529511933071,
+      "loss": 0.4564,
+      "step": 2679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015512457570624445,
+      "loss": 0.5002,
+      "step": 2680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001547964837207631,
+      "loss": 0.4967,
+      "step": 2681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015446867550656767,
+      "loss": 0.4863,
+      "step": 2682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015414115133313027,
+      "loss": 0.4733,
+      "step": 2683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015381391146968864,
+      "loss": 0.4707,
+      "step": 2684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015348695618524756,
+      "loss": 0.4848,
+      "step": 2685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001531602857485772,
+      "loss": 0.511,
+      "step": 2686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015283390042821404,
+      "loss": 0.4873,
+      "step": 2687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015250780049246026,
+      "loss": 0.5001,
+      "step": 2688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001521819862093829,
+      "loss": 0.4574,
+      "step": 2689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015185645784681484,
+      "loss": 0.4625,
+      "step": 2690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015153121567235335,
+      "loss": 0.4341,
+      "step": 2691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015120625995336073,
+      "loss": 0.4688,
+      "step": 2692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015088159095696363,
+      "loss": 0.4456,
+      "step": 2693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015055720895005342,
+      "loss": 0.4942,
+      "step": 2694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015023311419928493,
+      "loss": 0.4648,
+      "step": 2695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001499093069710774,
+      "loss": 0.4393,
+      "step": 2696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001495857875316136,
+      "loss": 0.4728,
+      "step": 2697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014926255614683932,
+      "loss": 0.4357,
+      "step": 2698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014893961308246417,
+      "loss": 0.4771,
+      "step": 2699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001486169586039602,
+      "loss": 0.4949,
+      "step": 2700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014829459297656267,
+      "loss": 0.4746,
+      "step": 2701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001479725164652691,
+      "loss": 0.4657,
+      "step": 2702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014765072933483947,
+      "loss": 0.4946,
+      "step": 2703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014732923184979563,
+      "loss": 0.4465,
+      "step": 2704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001470080242744218,
+      "loss": 0.4641,
+      "step": 2705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014668710687276365,
+      "loss": 0.5059,
+      "step": 2706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014636647990862807,
+      "loss": 0.4783,
+      "step": 2707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014604614364558373,
+      "loss": 0.473,
+      "step": 2708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014572609834695972,
+      "loss": 0.4335,
+      "step": 2709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014540634427584654,
+      "loss": 0.4192,
+      "step": 2710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014508688169509465,
+      "loss": 0.4597,
+      "step": 2711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014476771086731566,
+      "loss": 0.4678,
+      "step": 2712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014444883205488068,
+      "loss": 0.4697,
+      "step": 2713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014413024551992087,
+      "loss": 0.4701,
+      "step": 2714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014381195152432768,
+      "loss": 0.4883,
+      "step": 2715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014349395032975132,
+      "loss": 0.4559,
+      "step": 2716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014317624219760188,
+      "loss": 0.4719,
+      "step": 2717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014285882738904822,
+      "loss": 0.4539,
+      "step": 2718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014254170616501827,
+      "loss": 0.431,
+      "step": 2719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001422248787861984,
+      "loss": 0.4847,
+      "step": 2720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001419083455130337,
+      "loss": 0.4916,
+      "step": 2721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014159210660572724,
+      "loss": 0.4889,
+      "step": 2722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001412761623242404,
+      "loss": 0.4773,
+      "step": 2723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014096051292829205,
+      "loss": 0.4777,
+      "step": 2724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014064515867735867,
+      "loss": 0.4888,
+      "step": 2725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014033009983067452,
+      "loss": 0.4476,
+      "step": 2726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014001533664723047,
+      "loss": 0.5007,
+      "step": 2727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001397008693857749,
+      "loss": 0.4545,
+      "step": 2728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013938669830481233,
+      "loss": 0.4542,
+      "step": 2729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001390728236626045,
+      "loss": 0.5139,
+      "step": 2730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001387592457171687,
+      "loss": 0.469,
+      "step": 2731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013844596472627885,
+      "loss": 0.4685,
+      "step": 2732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001381329809474649,
+      "loss": 0.4908,
+      "step": 2733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000137820294638012,
+      "loss": 0.4783,
+      "step": 2734
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013750790605496089,
+      "loss": 0.4674,
+      "step": 2735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013719581545510763,
+      "loss": 0.4408,
+      "step": 2736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013688402309500352,
+      "loss": 0.4873,
+      "step": 2737
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013657252923095437,
+      "loss": 0.4515,
+      "step": 2738
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013626133411902093,
+      "loss": 0.5026,
+      "step": 2739
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013595043801501794,
+      "loss": 0.4943,
+      "step": 2740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013563984117451495,
+      "loss": 0.4613,
+      "step": 2741
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013532954385283481,
+      "loss": 0.4423,
+      "step": 2742
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013501954630505464,
+      "loss": 0.4645,
+      "step": 2743
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013470984878600519,
+      "loss": 0.4971,
+      "step": 2744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013440045155027014,
+      "loss": 0.4597,
+      "step": 2745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013409135485218654,
+      "loss": 0.4558,
+      "step": 2746
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013378255894584462,
+      "loss": 0.4669,
+      "step": 2747
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013347406408508694,
+      "loss": 0.4895,
+      "step": 2748
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013316587052350875,
+      "loss": 0.4832,
+      "step": 2749
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013285797851445792,
+      "loss": 0.4475,
+      "step": 2750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013255038831103382,
+      "loss": 0.4692,
+      "step": 2751
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001322431001660882,
+      "loss": 0.4772,
+      "step": 2752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013193611433222463,
+      "loss": 0.472,
+      "step": 2753
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013162943106179747,
+      "loss": 0.4861,
+      "step": 2754
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013132305060691325,
+      "loss": 0.4629,
+      "step": 2755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013101697321942873,
+      "loss": 0.4412,
+      "step": 2756
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013071119915095226,
+      "loss": 0.4385,
+      "step": 2757
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013040572865284235,
+      "loss": 0.4492,
+      "step": 2758
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013010056197620813,
+      "loss": 0.4689,
+      "step": 2759
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012979569937190887,
+      "loss": 0.4189,
+      "step": 2760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012949114109055414,
+      "loss": 0.4618,
+      "step": 2761
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012918688738250333,
+      "loss": 0.439,
+      "step": 2762
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012888293849786504,
+      "loss": 0.4466,
+      "step": 2763
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001285792946864978,
+      "loss": 0.4424,
+      "step": 2764
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012827595619800892,
+      "loss": 0.4714,
+      "step": 2765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012797292328175524,
+      "loss": 0.4991,
+      "step": 2766
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012767019618684173,
+      "loss": 0.4812,
+      "step": 2767
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012736777516212267,
+      "loss": 0.5397,
+      "step": 2768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012706566045620027,
+      "loss": 0.4395,
+      "step": 2769
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012676385231742492,
+      "loss": 0.4601,
+      "step": 2770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012646235099389542,
+      "loss": 0.4647,
+      "step": 2771
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012616115673345785,
+      "loss": 0.4504,
+      "step": 2772
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001258602697837063,
+      "loss": 0.472,
+      "step": 2773
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012555969039198178,
+      "loss": 0.4757,
+      "step": 2774
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012525941880537307,
+      "loss": 0.4628,
+      "step": 2775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012495945527071523,
+      "loss": 0.4655,
+      "step": 2776
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012465980003459082,
+      "loss": 0.486,
+      "step": 2777
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012436045334332825,
+      "loss": 0.4686,
+      "step": 2778
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012406141544300293,
+      "loss": 0.5144,
+      "step": 2779
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012376268657943602,
+      "loss": 0.4243,
+      "step": 2780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012346426699819457,
+      "loss": 0.4934,
+      "step": 2781
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001231661569445919,
+      "loss": 0.4648,
+      "step": 2782
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012286835666368623,
+      "loss": 0.4524,
+      "step": 2783
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001225708664002818,
+      "loss": 0.4723,
+      "step": 2784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001222736863989274,
+      "loss": 0.4644,
+      "step": 2785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012197681690391726,
+      "loss": 0.4548,
+      "step": 2786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001216802581592899,
+      "loss": 0.4717,
+      "step": 2787
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012138401040882874,
+      "loss": 0.4791,
+      "step": 2788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012108807389606158,
+      "loss": 0.4334,
+      "step": 2789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012079244886426016,
+      "loss": 0.4745,
+      "step": 2790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012049713555644015,
+      "loss": 0.4561,
+      "step": 2791
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012020213421536102,
+      "loss": 0.4414,
+      "step": 2792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011990744508352603,
+      "loss": 0.484,
+      "step": 2793
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011961306840318131,
+      "loss": 0.5029,
+      "step": 2794
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011931900441631677,
+      "loss": 0.4219,
+      "step": 2795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011902525336466464,
+      "loss": 0.4325,
+      "step": 2796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011873181548970025,
+      "loss": 0.4698,
+      "step": 2797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001184386910326417,
+      "loss": 0.46,
+      "step": 2798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011814588023444878,
+      "loss": 0.4652,
+      "step": 2799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011785338333582424,
+      "loss": 0.4667,
+      "step": 2800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011756120057721209,
+      "loss": 0.4686,
+      "step": 2801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011726933219879854,
+      "loss": 0.473,
+      "step": 2802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011697777844051105,
+      "loss": 0.4507,
+      "step": 2803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001166865395420189,
+      "loss": 0.4808,
+      "step": 2804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000116395615742732,
+      "loss": 0.5081,
+      "step": 2805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011610500728180163,
+      "loss": 0.4899,
+      "step": 2806
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011581471439811986,
+      "loss": 0.4829,
+      "step": 2807
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011552473733031893,
+      "loss": 0.4833,
+      "step": 2808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001152350763167721,
+      "loss": 0.4409,
+      "step": 2809
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011494573159559212,
+      "loss": 0.4664,
+      "step": 2810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011465670340463241,
+      "loss": 0.4719,
+      "step": 2811
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011436799198148561,
+      "loss": 0.4784,
+      "step": 2812
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011407959756348424,
+      "loss": 0.4792,
+      "step": 2813
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011379152038770029,
+      "loss": 0.4655,
+      "step": 2814
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011350376069094482,
+      "loss": 0.5127,
+      "step": 2815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011321631870976806,
+      "loss": 0.4777,
+      "step": 2816
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011292919468045875,
+      "loss": 0.4509,
+      "step": 2817
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011264238883904482,
+      "loss": 0.4581,
+      "step": 2818
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011235590142129198,
+      "loss": 0.4775,
+      "step": 2819
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011206973266270476,
+      "loss": 0.4451,
+      "step": 2820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001117838827985252,
+      "loss": 0.4832,
+      "step": 2821
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011149835206373382,
+      "loss": 0.4413,
+      "step": 2822
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011121314069304811,
+      "loss": 0.5076,
+      "step": 2823
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011092824892092374,
+      "loss": 0.4794,
+      "step": 2824
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011064367698155303,
+      "loss": 0.4396,
+      "step": 2825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011035942510886565,
+      "loss": 0.4423,
+      "step": 2826
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011007549353652834,
+      "loss": 0.4847,
+      "step": 2827
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001097918824979442,
+      "loss": 0.524,
+      "step": 2828
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010950859222625314,
+      "loss": 0.4731,
+      "step": 2829
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010922562295433108,
+      "loss": 0.4817,
+      "step": 2830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010894297491479043,
+      "loss": 0.4482,
+      "step": 2831
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010866064833997914,
+      "loss": 0.4934,
+      "step": 2832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010837864346198118,
+      "loss": 0.4728,
+      "step": 2833
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010809696051261614,
+      "loss": 0.468,
+      "step": 2834
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010781559972343869,
+      "loss": 0.4677,
+      "step": 2835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010753456132573885,
+      "loss": 0.4792,
+      "step": 2836
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001072538455505414,
+      "loss": 0.4539,
+      "step": 2837
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010697345262860636,
+      "loss": 0.4764,
+      "step": 2838
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010669338279042784,
+      "loss": 0.4644,
+      "step": 2839
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001064136362662348,
+      "loss": 0.4392,
+      "step": 2840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010613421328599005,
+      "loss": 0.47,
+      "step": 2841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010585511407939064,
+      "loss": 0.4781,
+      "step": 2842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010557633887586765,
+      "loss": 0.4423,
+      "step": 2843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010529788790458534,
+      "loss": 0.512,
+      "step": 2844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010501976139444191,
+      "loss": 0.4625,
+      "step": 2845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010474195957406851,
+      "loss": 0.4623,
+      "step": 2846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010446448267182951,
+      "loss": 0.453,
+      "step": 2847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000104187330915822,
+      "loss": 0.4625,
+      "step": 2848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010391050453387629,
+      "loss": 0.5054,
+      "step": 2849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010363400375355464,
+      "loss": 0.4609,
+      "step": 2850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010335782880215194,
+      "loss": 0.4475,
+      "step": 2851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010308197990669538,
+      "loss": 0.4665,
+      "step": 2852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010280645729394367,
+      "loss": 0.4581,
+      "step": 2853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001025312611903878,
+      "loss": 0.4849,
+      "step": 2854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010225639182224994,
+      "loss": 0.4766,
+      "step": 2855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010198184941548405,
+      "loss": 0.4639,
+      "step": 2856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010170763419577506,
+      "loss": 0.4792,
+      "step": 2857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010143374638853892,
+      "loss": 0.4803,
+      "step": 2858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010116018621892236,
+      "loss": 0.4774,
+      "step": 2859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010088695391180314,
+      "loss": 0.4554,
+      "step": 2860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010061404969178944,
+      "loss": 0.4733,
+      "step": 2861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010034147378321923,
+      "loss": 0.4454,
+      "step": 2862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010006922641016131,
+      "loss": 0.4656,
+      "step": 2863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.979730779641372e-05,
+      "loss": 0.4788,
+      "step": 2864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.952571816550494e-05,
+      "loss": 0.4619,
+      "step": 2865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.925445774069231e-05,
+      "loss": 0.4503,
+      "step": 2866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.898352674496325e-05,
+      "loss": 0.478,
+      "step": 2867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.871292540103377e-05,
+      "loss": 0.4595,
+      "step": 2868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.844265393134927e-05,
+      "loss": 0.4631,
+      "step": 2869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.817271255808363e-05,
+      "loss": 0.4849,
+      "step": 2870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.790310150313974e-05,
+      "loss": 0.4357,
+      "step": 2871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.763382098814905e-05,
+      "loss": 0.4622,
+      "step": 2872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.736487123447069e-05,
+      "loss": 0.4923,
+      "step": 2873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.70962524631926e-05,
+      "loss": 0.4556,
+      "step": 2874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.682796489512995e-05,
+      "loss": 0.469,
+      "step": 2875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.656000875082638e-05,
+      "loss": 0.483,
+      "step": 2876
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.629238425055231e-05,
+      "loss": 0.4735,
+      "step": 2877
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.602509161430629e-05,
+      "loss": 0.4659,
+      "step": 2878
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.575813106181353e-05,
+      "loss": 0.4756,
+      "step": 2879
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.549150281252633e-05,
+      "loss": 0.4755,
+      "step": 2880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.522520708562416e-05,
+      "loss": 0.4946,
+      "step": 2881
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.49592441000126e-05,
+      "loss": 0.4373,
+      "step": 2882
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.46936140743243e-05,
+      "loss": 0.4716,
+      "step": 2883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.442831722691775e-05,
+      "loss": 0.4951,
+      "step": 2884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.416335377587781e-05,
+      "loss": 0.4604,
+      "step": 2885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.389872393901499e-05,
+      "loss": 0.4625,
+      "step": 2886
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.363442793386607e-05,
+      "loss": 0.4431,
+      "step": 2887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.337046597769272e-05,
+      "loss": 0.5138,
+      "step": 2888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.31068382874825e-05,
+      "loss": 0.449,
+      "step": 2889
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.28435450799483e-05,
+      "loss": 0.5156,
+      "step": 2890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.258058657152762e-05,
+      "loss": 0.4851,
+      "step": 2891
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.231796297838297e-05,
+      "loss": 0.4689,
+      "step": 2892
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.205567451640151e-05,
+      "loss": 0.4246,
+      "step": 2893
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.179372140119524e-05,
+      "loss": 0.4499,
+      "step": 2894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.153210384809995e-05,
+      "loss": 0.4733,
+      "step": 2895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.127082207217619e-05,
+      "loss": 0.4224,
+      "step": 2896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.10098762882079e-05,
+      "loss": 0.4784,
+      "step": 2897
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.074926671070321e-05,
+      "loss": 0.4034,
+      "step": 2898
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.04889935538939e-05,
+      "loss": 0.4786,
+      "step": 2899
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.022905703173478e-05,
+      "loss": 0.483,
+      "step": 2900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.996945735790446e-05,
+      "loss": 0.502,
+      "step": 2901
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.971019474580427e-05,
+      "loss": 0.4551,
+      "step": 2902
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.945126940855863e-05,
+      "loss": 0.4871,
+      "step": 2903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.919268155901456e-05,
+      "loss": 0.4868,
+      "step": 2904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.893443140974189e-05,
+      "loss": 0.4781,
+      "step": 2905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.867651917303254e-05,
+      "loss": 0.4703,
+      "step": 2906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.841894506090098e-05,
+      "loss": 0.4866,
+      "step": 2907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.816170928508365e-05,
+      "loss": 0.4421,
+      "step": 2908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.790481205703855e-05,
+      "loss": 0.4911,
+      "step": 2909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.764825358794587e-05,
+      "loss": 0.4723,
+      "step": 2910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.739203408870677e-05,
+      "loss": 0.4624,
+      "step": 2911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.713615376994444e-05,
+      "loss": 0.4476,
+      "step": 2912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.688061284200266e-05,
+      "loss": 0.4957,
+      "step": 2913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.66254115149464e-05,
+      "loss": 0.4941,
+      "step": 2914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.637054999856148e-05,
+      "loss": 0.4414,
+      "step": 2915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.611602850235445e-05,
+      "loss": 0.436,
+      "step": 2916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.586184723555241e-05,
+      "loss": 0.498,
+      "step": 2917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.560800640710248e-05,
+      "loss": 0.5109,
+      "step": 2918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.535450622567231e-05,
+      "loss": 0.5162,
+      "step": 2919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.510134689964915e-05,
+      "loss": 0.4423,
+      "step": 2920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.484852863714038e-05,
+      "loss": 0.4777,
+      "step": 2921
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.459605164597267e-05,
+      "loss": 0.4783,
+      "step": 2922
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.434391613369258e-05,
+      "loss": 0.4901,
+      "step": 2923
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.409212230756563e-05,
+      "loss": 0.4485,
+      "step": 2924
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.384067037457638e-05,
+      "loss": 0.4473,
+      "step": 2925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.358956054142875e-05,
+      "loss": 0.4976,
+      "step": 2926
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.3338793014545e-05,
+      "loss": 0.4266,
+      "step": 2927
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.308836800006647e-05,
+      "loss": 0.4763,
+      "step": 2928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.283828570385238e-05,
+      "loss": 0.4514,
+      "step": 2929
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.258854633148083e-05,
+      "loss": 0.4885,
+      "step": 2930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.233915008824733e-05,
+      "loss": 0.4983,
+      "step": 2931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.20900971791661e-05,
+      "loss": 0.4666,
+      "step": 2932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.184138780896839e-05,
+      "loss": 0.496,
+      "step": 2933
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.159302218210368e-05,
+      "loss": 0.4418,
+      "step": 2934
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.134500050273841e-05,
+      "loss": 0.4767,
+      "step": 2935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.109732297475635e-05,
+      "loss": 0.4661,
+      "step": 2936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.084998980175878e-05,
+      "loss": 0.4541,
+      "step": 2937
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.060300118706327e-05,
+      "loss": 0.4438,
+      "step": 2938
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.035635733370478e-05,
+      "loss": 0.4625,
+      "step": 2939
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.011005844443425e-05,
+      "loss": 0.4336,
+      "step": 2940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.986410472171962e-05,
+      "loss": 0.4557,
+      "step": 2941
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.961849636774459e-05,
+      "loss": 0.5051,
+      "step": 2942
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.937323358440934e-05,
+      "loss": 0.4974,
+      "step": 2943
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.912831657332992e-05,
+      "loss": 0.4606,
+      "step": 2944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.888374553583788e-05,
+      "loss": 0.4785,
+      "step": 2945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.863952067298042e-05,
+      "loss": 0.4397,
+      "step": 2946
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.839564218552054e-05,
+      "loss": 0.4685,
+      "step": 2947
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.815211027393615e-05,
+      "loss": 0.4958,
+      "step": 2948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.790892513842019e-05,
+      "loss": 0.4884,
+      "step": 2949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.766608697888095e-05,
+      "loss": 0.4713,
+      "step": 2950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.742359599494103e-05,
+      "loss": 0.4373,
+      "step": 2951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.718145238593793e-05,
+      "loss": 0.5111,
+      "step": 2952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.693965635092365e-05,
+      "loss": 0.469,
+      "step": 2953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.669820808866413e-05,
+      "loss": 0.4927,
+      "step": 2954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.645710779763981e-05,
+      "loss": 0.427,
+      "step": 2955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.621635567604468e-05,
+      "loss": 0.4713,
+      "step": 2956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.597595192178702e-05,
+      "loss": 0.4424,
+      "step": 2957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.573589673248832e-05,
+      "loss": 0.4656,
+      "step": 2958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.549619030548366e-05,
+      "loss": 0.4984,
+      "step": 2959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.525683283782137e-05,
+      "loss": 0.474,
+      "step": 2960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.501782452626305e-05,
+      "loss": 0.4882,
+      "step": 2961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.477916556728343e-05,
+      "loss": 0.4035,
+      "step": 2962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.454085615706952e-05,
+      "loss": 0.4955,
+      "step": 2963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.430289649152156e-05,
+      "loss": 0.4873,
+      "step": 2964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.40652867662519e-05,
+      "loss": 0.468,
+      "step": 2965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.382802717658548e-05,
+      "loss": 0.451,
+      "step": 2966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.359111791755917e-05,
+      "loss": 0.4978,
+      "step": 2967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.335455918392219e-05,
+      "loss": 0.4749,
+      "step": 2968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.311835117013526e-05,
+      "loss": 0.4465,
+      "step": 2969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.288249407037084e-05,
+      "loss": 0.474,
+      "step": 2970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.264698807851328e-05,
+      "loss": 0.5173,
+      "step": 2971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.241183338815782e-05,
+      "loss": 0.4723,
+      "step": 2972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.217703019261135e-05,
+      "loss": 0.4498,
+      "step": 2973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.194257868489145e-05,
+      "loss": 0.4587,
+      "step": 2974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.170847905772693e-05,
+      "loss": 0.4882,
+      "step": 2975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.147473150355693e-05,
+      "loss": 0.4954,
+      "step": 2976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.124133621453177e-05,
+      "loss": 0.4764,
+      "step": 2977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.100829338251146e-05,
+      "loss": 0.4728,
+      "step": 2978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.077560319906695e-05,
+      "loss": 0.5338,
+      "step": 2979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.054326585547893e-05,
+      "loss": 0.4504,
+      "step": 2980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.031128154273792e-05,
+      "loss": 0.5374,
+      "step": 2981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.007965045154474e-05,
+      "loss": 0.5002,
+      "step": 2982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.984837277230926e-05,
+      "loss": 0.4319,
+      "step": 2983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.961744869515135e-05,
+      "loss": 0.4688,
+      "step": 2984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.938687840989971e-05,
+      "loss": 0.4838,
+      "step": 2985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.915666210609278e-05,
+      "loss": 0.4385,
+      "step": 2986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.892679997297734e-05,
+      "loss": 0.5264,
+      "step": 2987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.86972921995096e-05,
+      "loss": 0.4342,
+      "step": 2988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.846813897435428e-05,
+      "loss": 0.425,
+      "step": 2989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.82393404858846e-05,
+      "loss": 0.4471,
+      "step": 2990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.801089692218215e-05,
+      "loss": 0.4829,
+      "step": 2991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.778280847103668e-05,
+      "loss": 0.5144,
+      "step": 2992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.755507531994636e-05,
+      "loss": 0.5063,
+      "step": 2993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.73276976561169e-05,
+      "loss": 0.5143,
+      "step": 2994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.71006756664621e-05,
+      "loss": 0.48,
+      "step": 2995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.687400953760309e-05,
+      "loss": 0.4333,
+      "step": 2996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.664769945586885e-05,
+      "loss": 0.4862,
+      "step": 2997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.642174560729514e-05,
+      "loss": 0.5066,
+      "step": 2998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.619614817762538e-05,
+      "loss": 0.4612,
+      "step": 2999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.597090735230987e-05,
+      "loss": 0.4799,
+      "step": 3000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.574602331650559e-05,
+      "loss": 0.4445,
+      "step": 3001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.552149625507636e-05,
+      "loss": 0.4692,
+      "step": 3002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.529732635259234e-05,
+      "loss": 0.4396,
+      "step": 3003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.50735137933306e-05,
+      "loss": 0.4464,
+      "step": 3004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.485005876127387e-05,
+      "loss": 0.5052,
+      "step": 3005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.462696144011149e-05,
+      "loss": 0.4666,
+      "step": 3006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.440422201323825e-05,
+      "loss": 0.4395,
+      "step": 3007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.41818406637551e-05,
+      "loss": 0.4507,
+      "step": 3008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.395981757446867e-05,
+      "loss": 0.5085,
+      "step": 3009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.37381529278907e-05,
+      "loss": 0.4524,
+      "step": 3010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.351684690623876e-05,
+      "loss": 0.4779,
+      "step": 3011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.329589969143517e-05,
+      "loss": 0.4186,
+      "step": 3012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.307531146510753e-05,
+      "loss": 0.4784,
+      "step": 3013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.28550824085884e-05,
+      "loss": 0.4867,
+      "step": 3014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2635212702915e-05,
+      "loss": 0.4656,
+      "step": 3015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.241570252882888e-05,
+      "loss": 0.4562,
+      "step": 3016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.21965520667766e-05,
+      "loss": 0.4956,
+      "step": 3017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.197776149690871e-05,
+      "loss": 0.4424,
+      "step": 3018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.175933099907971e-05,
+      "loss": 0.4899,
+      "step": 3019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.154126075284855e-05,
+      "loss": 0.468,
+      "step": 3020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.132355093747765e-05,
+      "loss": 0.484,
+      "step": 3021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.110620173193349e-05,
+      "loss": 0.4685,
+      "step": 3022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.088921331488567e-05,
+      "loss": 0.5029,
+      "step": 3023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0672585864707806e-05,
+      "loss": 0.4254,
+      "step": 3024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.045631955947617e-05,
+      "loss": 0.4635,
+      "step": 3025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0240414576970436e-05,
+      "loss": 0.4923,
+      "step": 3026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.002487109467347e-05,
+      "loss": 0.4603,
+      "step": 3027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.980968928977049e-05,
+      "loss": 0.4435,
+      "step": 3028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.959486933914993e-05,
+      "loss": 0.4343,
+      "step": 3029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9380411419402206e-05,
+      "loss": 0.4458,
+      "step": 3030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.916631570682068e-05,
+      "loss": 0.4639,
+      "step": 3031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.895258237740053e-05,
+      "loss": 0.4656,
+      "step": 3032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8739211606839427e-05,
+      "loss": 0.4816,
+      "step": 3033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.852620357053651e-05,
+      "loss": 0.4777,
+      "step": 3034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8313558443593264e-05,
+      "loss": 0.4926,
+      "step": 3035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.810127640081259e-05,
+      "loss": 0.4764,
+      "step": 3036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7889357616698665e-05,
+      "loss": 0.4559,
+      "step": 3037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7677802265457656e-05,
+      "loss": 0.4834,
+      "step": 3038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.746661052099639e-05,
+      "loss": 0.4331,
+      "step": 3039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.725578255692326e-05,
+      "loss": 0.4459,
+      "step": 3040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7045318546547206e-05,
+      "loss": 0.4441,
+      "step": 3041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.683521866287833e-05,
+      "loss": 0.4962,
+      "step": 3042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6625483078627136e-05,
+      "loss": 0.4686,
+      "step": 3043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.641611196620478e-05,
+      "loss": 0.4725,
+      "step": 3044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6207105497722956e-05,
+      "loss": 0.4537,
+      "step": 3045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5998463844993354e-05,
+      "loss": 0.4795,
+      "step": 3046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5790187179527885e-05,
+      "loss": 0.5042,
+      "step": 3047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5582275672538315e-05,
+      "loss": 0.4912,
+      "step": 3048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.537472949493649e-05,
+      "loss": 0.4209,
+      "step": 3049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.51675488173336e-05,
+      "loss": 0.4666,
+      "step": 3050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4960733810040797e-05,
+      "loss": 0.4425,
+      "step": 3051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.475428464306814e-05,
+      "loss": 0.4448,
+      "step": 3052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4548201486125324e-05,
+      "loss": 0.473,
+      "step": 3053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.434248450862123e-05,
+      "loss": 0.4391,
+      "step": 3054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.413713387966329e-05,
+      "loss": 0.4994,
+      "step": 3055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3932149768058325e-05,
+      "loss": 0.4548,
+      "step": 3056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.372753234231137e-05,
+      "loss": 0.4703,
+      "step": 3057
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.352328177062626e-05,
+      "loss": 0.4688,
+      "step": 3058
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3319398220905255e-05,
+      "loss": 0.4745,
+      "step": 3059
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3115881860749005e-05,
+      "loss": 0.4547,
+      "step": 3060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.291273285745601e-05,
+      "loss": 0.4412,
+      "step": 3061
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.270995137802315e-05,
+      "loss": 0.5387,
+      "step": 3062
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.250753758914506e-05,
+      "loss": 0.4827,
+      "step": 3063
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.230549165721393e-05,
+      "loss": 0.4802,
+      "step": 3064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.210381374831991e-05,
+      "loss": 0.4708,
+      "step": 3065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.190250402825014e-05,
+      "loss": 0.4661,
+      "step": 3066
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.17015626624896e-05,
+      "loss": 0.4757,
+      "step": 3067
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.15009898162202e-05,
+      "loss": 0.4471,
+      "step": 3068
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.130078565432089e-05,
+      "loss": 0.4662,
+      "step": 3069
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.110095034136758e-05,
+      "loss": 0.4684,
+      "step": 3070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.090148404163303e-05,
+      "loss": 0.4878,
+      "step": 3071
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0702386919086685e-05,
+      "loss": 0.473,
+      "step": 3072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0503659137394406e-05,
+      "loss": 0.4647,
+      "step": 3073
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.030530085991852e-05,
+      "loss": 0.4357,
+      "step": 3074
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.010731224971748e-05,
+      "loss": 0.4496,
+      "step": 3075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.99096934695461e-05,
+      "loss": 0.4749,
+      "step": 3076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.971244468185487e-05,
+      "loss": 0.4712,
+      "step": 3077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9515566048790485e-05,
+      "loss": 0.4951,
+      "step": 3078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9319057732194974e-05,
+      "loss": 0.4681,
+      "step": 3079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9122919893606175e-05,
+      "loss": 0.4797,
+      "step": 3080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.892715269425746e-05,
+      "loss": 0.4362,
+      "step": 3081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8731756295077256e-05,
+      "loss": 0.4482,
+      "step": 3082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.853673085668947e-05,
+      "loss": 0.4274,
+      "step": 3083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.834207653941269e-05,
+      "loss": 0.5154,
+      "step": 3084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.814779350326098e-05,
+      "loss": 0.4508,
+      "step": 3085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.795388190794259e-05,
+      "loss": 0.512,
+      "step": 3086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7760341912860936e-05,
+      "loss": 0.4827,
+      "step": 3087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.75671736771135e-05,
+      "loss": 0.4679,
+      "step": 3088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737437735949262e-05,
+      "loss": 0.4158,
+      "step": 3089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7181953118484556e-05,
+      "loss": 0.4525,
+      "step": 3090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.6989901112269973e-05,
+      "loss": 0.4465,
+      "step": 3091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.679822149872326e-05,
+      "loss": 0.4683,
+      "step": 3092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6606914435412815e-05,
+      "loss": 0.4681,
+      "step": 3093
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6415980079601015e-05,
+      "loss": 0.4601,
+      "step": 3094
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6225418588243364e-05,
+      "loss": 0.4751,
+      "step": 3095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.603523011798932e-05,
+      "loss": 0.4674,
+      "step": 3096
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.58454148251814e-05,
+      "loss": 0.4361,
+      "step": 3097
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5655972865855545e-05,
+      "loss": 0.4814,
+      "step": 3098
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.546690439574081e-05,
+      "loss": 0.434,
+      "step": 3099
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.527820957025891e-05,
+      "loss": 0.4727,
+      "step": 3100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5089888544524946e-05,
+      "loss": 0.4989,
+      "step": 3101
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.49019414733462e-05,
+      "loss": 0.45,
+      "step": 3102
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4714368511222906e-05,
+      "loss": 0.4966,
+      "step": 3103
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.452716981234745e-05,
+      "loss": 0.4747,
+      "step": 3104
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.434034553060501e-05,
+      "loss": 0.4502,
+      "step": 3105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.415389581957252e-05,
+      "loss": 0.4412,
+      "step": 3106
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3967820832519266e-05,
+      "loss": 0.5206,
+      "step": 3107
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.378212072240656e-05,
+      "loss": 0.4656,
+      "step": 3108
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3596795641887234e-05,
+      "loss": 0.501,
+      "step": 3109
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3411845743306146e-05,
+      "loss": 0.4818,
+      "step": 3110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.322727117869951e-05,
+      "loss": 0.4736,
+      "step": 3111
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3043072099795245e-05,
+      "loss": 0.4213,
+      "step": 3112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.285924865801233e-05,
+      "loss": 0.4796,
+      "step": 3113
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2675801004461165e-05,
+      "loss": 0.4424,
+      "step": 3114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2492729289943023e-05,
+      "loss": 0.4677,
+      "step": 3115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2310033664950344e-05,
+      "loss": 0.4878,
+      "step": 3116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.212771427966649e-05,
+      "loss": 0.4444,
+      "step": 3117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.194577128396521e-05,
+      "loss": 0.4167,
+      "step": 3118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.176420482741111e-05,
+      "loss": 0.4788,
+      "step": 3119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.158301505925904e-05,
+      "loss": 0.4678,
+      "step": 3120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.140220212845458e-05,
+      "loss": 0.4205,
+      "step": 3121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.122176618363305e-05,
+      "loss": 0.4775,
+      "step": 3122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1041707373120354e-05,
+      "loss": 0.4451,
+      "step": 3123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0862025844931924e-05,
+      "loss": 0.4794,
+      "step": 3124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0682721746773344e-05,
+      "loss": 0.4474,
+      "step": 3125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0503795226039805e-05,
+      "loss": 0.4382,
+      "step": 3126
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0325246429816167e-05,
+      "loss": 0.4594,
+      "step": 3127
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.014707550487684e-05,
+      "loss": 0.4784,
+      "step": 3128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.996928259768551e-05,
+      "loss": 0.4402,
+      "step": 3129
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9791867854395206e-05,
+      "loss": 0.4844,
+      "step": 3130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.961483142084793e-05,
+      "loss": 0.4156,
+      "step": 3131
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9438173442575e-05,
+      "loss": 0.478,
+      "step": 3132
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.926189406479613e-05,
+      "loss": 0.4537,
+      "step": 3133
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9085993432420476e-05,
+      "loss": 0.5057,
+      "step": 3134
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8910471690045243e-05,
+      "loss": 0.4741,
+      "step": 3135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.873532898195648e-05,
+      "loss": 0.4624,
+      "step": 3136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8560565452128605e-05,
+      "loss": 0.4974,
+      "step": 3137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8386181244224274e-05,
+      "loss": 0.4668,
+      "step": 3138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.821217650159453e-05,
+      "loss": 0.4634,
+      "step": 3139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.803855136727807e-05,
+      "loss": 0.4788,
+      "step": 3140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.786530598400206e-05,
+      "loss": 0.4877,
+      "step": 3141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.769244049418108e-05,
+      "loss": 0.4885,
+      "step": 3142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.751995503991762e-05,
+      "loss": 0.4325,
+      "step": 3143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.734784976300165e-05,
+      "loss": 0.5176,
+      "step": 3144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.717612480491078e-05,
+      "loss": 0.5177,
+      "step": 3145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.700478030680987e-05,
+      "loss": 0.4802,
+      "step": 3146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.683381640955097e-05,
+      "loss": 0.4976,
+      "step": 3147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6663233253673434e-05,
+      "loss": 0.4492,
+      "step": 3148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.649303097940337e-05,
+      "loss": 0.4456,
+      "step": 3149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.632320972665415e-05,
+      "loss": 0.476,
+      "step": 3150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.615376963502559e-05,
+      "loss": 0.4775,
+      "step": 3151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5984710843804446e-05,
+      "loss": 0.4556,
+      "step": 3152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5816033491963716e-05,
+      "loss": 0.4451,
+      "step": 3153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.56477377181631e-05,
+      "loss": 0.468,
+      "step": 3154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.54798236607487e-05,
+      "loss": 0.4465,
+      "step": 3155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53122914577525e-05,
+      "loss": 0.4937,
+      "step": 3156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.514514124689277e-05,
+      "loss": 0.4745,
+      "step": 3157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.497837316557384e-05,
+      "loss": 0.4476,
+      "step": 3158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4811987350885807e-05,
+      "loss": 0.464,
+      "step": 3159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.46459839396045e-05,
+      "loss": 0.4749,
+      "step": 3160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.448036306819152e-05,
+      "loss": 0.4823,
+      "step": 3161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.431512487279392e-05,
+      "loss": 0.4802,
+      "step": 3162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.41502694892441e-05,
+      "loss": 0.4315,
+      "step": 3163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3985797053060096e-05,
+      "loss": 0.4417,
+      "step": 3164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.38217076994447e-05,
+      "loss": 0.472,
+      "step": 3165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.365800156328619e-05,
+      "loss": 0.4801,
+      "step": 3166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.349467877915746e-05,
+      "loss": 0.5131,
+      "step": 3167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3331739481316624e-05,
+      "loss": 0.4946,
+      "step": 3168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.316918380370632e-05,
+      "loss": 0.4125,
+      "step": 3169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.300701187995392e-05,
+      "loss": 0.446,
+      "step": 3170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.284522384337124e-05,
+      "loss": 0.4814,
+      "step": 3171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2683819826954674e-05,
+      "loss": 0.5226,
+      "step": 3172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.252279996338492e-05,
+      "loss": 0.4896,
+      "step": 3173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.23621643850267e-05,
+      "loss": 0.4253,
+      "step": 3174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.220191322392918e-05,
+      "loss": 0.4557,
+      "step": 3175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.204204661182508e-05,
+      "loss": 0.474,
+      "step": 3176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.18825646801314e-05,
+      "loss": 0.4962,
+      "step": 3177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.172346755994865e-05,
+      "loss": 0.4675,
+      "step": 3178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1564755382061214e-05,
+      "loss": 0.4906,
+      "step": 3179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.14064282769369e-05,
+      "loss": 0.4586,
+      "step": 3180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.124848637472688e-05,
+      "loss": 0.4482,
+      "step": 3181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1090929805266014e-05,
+      "loss": 0.488,
+      "step": 3182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.093375869807202e-05,
+      "loss": 0.4902,
+      "step": 3183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.077697318234607e-05,
+      "loss": 0.4556,
+      "step": 3184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.062057338697199e-05,
+      "loss": 0.4673,
+      "step": 3185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0464559440517015e-05,
+      "loss": 0.4872,
+      "step": 3186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.030893147123076e-05,
+      "loss": 0.4662,
+      "step": 3187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0153689607045842e-05,
+      "loss": 0.4728,
+      "step": 3188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9998833975577233e-05,
+      "loss": 0.4677,
+      "step": 3189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9844364704122663e-05,
+      "loss": 0.4949,
+      "step": 3190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.969028191966211e-05,
+      "loss": 0.4674,
+      "step": 3191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.953658574885776e-05,
+      "loss": 0.4516,
+      "step": 3192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.938327631805421e-05,
+      "loss": 0.4604,
+      "step": 3193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9230353753277983e-05,
+      "loss": 0.4769,
+      "step": 3194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9077818180237692e-05,
+      "loss": 0.5333,
+      "step": 3195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8925669724323645e-05,
+      "loss": 0.4337,
+      "step": 3196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8773908510608182e-05,
+      "loss": 0.47,
+      "step": 3197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.862253466384507e-05,
+      "loss": 0.4806,
+      "step": 3198
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.847154830846971e-05,
+      "loss": 0.4518,
+      "step": 3199
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8320949568599207e-05,
+      "loss": 0.4586,
+      "step": 3200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8170738568031707e-05,
+      "loss": 0.4766,
+      "step": 3201
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.802091543024671e-05,
+      "loss": 0.4381,
+      "step": 3202
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.787148027840486e-05,
+      "loss": 0.4709,
+      "step": 3203
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7722433235348065e-05,
+      "loss": 0.496,
+      "step": 3204
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7573774423598875e-05,
+      "loss": 0.4954,
+      "step": 3205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7425503965361033e-05,
+      "loss": 0.4719,
+      "step": 3206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.727762198251871e-05,
+      "loss": 0.4673,
+      "step": 3207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.713012859663694e-05,
+      "loss": 0.4782,
+      "step": 3208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6983023928961405e-05,
+      "loss": 0.4834,
+      "step": 3209
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.683630810041787e-05,
+      "loss": 0.4568,
+      "step": 3210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6689981231612925e-05,
+      "loss": 0.455,
+      "step": 3211
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.654404344283301e-05,
+      "loss": 0.4495,
+      "step": 3212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6398494854045054e-05,
+      "loss": 0.4742,
+      "step": 3213
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6253335584895678e-05,
+      "loss": 0.4915,
+      "step": 3214
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6108565754711933e-05,
+      "loss": 0.4489,
+      "step": 3215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.596418548250029e-05,
+      "loss": 0.4233,
+      "step": 3216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5820194886947258e-05,
+      "loss": 0.4659,
+      "step": 3217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5676594086419036e-05,
+      "loss": 0.4795,
+      "step": 3218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.55333831989612e-05,
+      "loss": 0.4957,
+      "step": 3219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5390562342298973e-05,
+      "loss": 0.4874,
+      "step": 3220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5248131633836822e-05,
+      "loss": 0.481,
+      "step": 3221
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5106091190658597e-05,
+      "loss": 0.49,
+      "step": 3222
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4964441129527336e-05,
+      "loss": 0.4705,
+      "step": 3223
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4823181566885068e-05,
+      "loss": 0.4764,
+      "step": 3224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.46823126188529e-05,
+      "loss": 0.4885,
+      "step": 3225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4541834401230812e-05,
+      "loss": 0.4337,
+      "step": 3226
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4401747029497657e-05,
+      "loss": 0.4697,
+      "step": 3227
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4262050618810815e-05,
+      "loss": 0.4755,
+      "step": 3228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.412274528400654e-05,
+      "loss": 0.451,
+      "step": 3229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3983831139599287e-05,
+      "loss": 0.4578,
+      "step": 3230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.384530829978232e-05,
+      "loss": 0.4799,
+      "step": 3231
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3707176878426885e-05,
+      "loss": 0.4406,
+      "step": 3232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3569436989082704e-05,
+      "loss": 0.46,
+      "step": 3233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3432088744977487e-05,
+      "loss": 0.5096,
+      "step": 3234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3295132259017082e-05,
+      "loss": 0.446,
+      "step": 3235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3158567643785266e-05,
+      "loss": 0.4,
+      "step": 3236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3022395011543685e-05,
+      "loss": 0.4762,
+      "step": 3237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2886614474231792e-05,
+      "loss": 0.4575,
+      "step": 3238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2751226143466586e-05,
+      "loss": 0.4558,
+      "step": 3239
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.261623013054298e-05,
+      "loss": 0.4531,
+      "step": 3240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.248162654643293e-05,
+      "loss": 0.4714,
+      "step": 3241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2347415501786207e-05,
+      "loss": 0.4446,
+      "step": 3242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2213597106929607e-05,
+      "loss": 0.4657,
+      "step": 3243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.208017147186736e-05,
+      "loss": 0.4659,
+      "step": 3244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.194713870628079e-05,
+      "loss": 0.4865,
+      "step": 3245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.181449891952819e-05,
+      "loss": 0.459,
+      "step": 3246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1682252220644794e-05,
+      "loss": 0.4612,
+      "step": 3247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.155039871834269e-05,
+      "loss": 0.4569,
+      "step": 3248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1418938521010955e-05,
+      "loss": 0.4866,
+      "step": 3249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1287871736715092e-05,
+      "loss": 0.4927,
+      "step": 3250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1157198473197415e-05,
+      "loss": 0.4598,
+      "step": 3251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1026918837876497e-05,
+      "loss": 0.4865,
+      "step": 3252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0897032937847616e-05,
+      "loss": 0.4678,
+      "step": 3253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0767540879882142e-05,
+      "loss": 0.4739,
+      "step": 3254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0638442770427867e-05,
+      "loss": 0.4416,
+      "step": 3255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0509738715608682e-05,
+      "loss": 0.4619,
+      "step": 3256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.038142882122457e-05,
+      "loss": 0.4634,
+      "step": 3257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.025351319275137e-05,
+      "loss": 0.4996,
+      "step": 3258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.012599193534087e-05,
+      "loss": 0.4895,
+      "step": 3259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.999886515382088e-05,
+      "loss": 0.3826,
+      "step": 3260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9872132952694632e-05,
+      "loss": 0.4548,
+      "step": 3261
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9745795436141245e-05,
+      "loss": 0.4655,
+      "step": 3262
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.961985270801514e-05,
+      "loss": 0.4575,
+      "step": 3263
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9494304871846447e-05,
+      "loss": 0.4895,
+      "step": 3264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9369152030840554e-05,
+      "loss": 0.516,
+      "step": 3265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9244394287878166e-05,
+      "loss": 0.4262,
+      "step": 3266
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9120031745515298e-05,
+      "loss": 0.4938,
+      "step": 3267
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.89960645059829e-05,
+      "loss": 0.4453,
+      "step": 3268
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.887249267118707e-05,
+      "loss": 0.472,
+      "step": 3269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8749316342708823e-05,
+      "loss": 0.5052,
+      "step": 3270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8626535621804164e-05,
+      "loss": 0.4749,
+      "step": 3271
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.850415060940386e-05,
+      "loss": 0.4855,
+      "step": 3272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8382161406113208e-05,
+      "loss": 0.4834,
+      "step": 3273
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8260568112212494e-05,
+      "loss": 0.4933,
+      "step": 3274
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8139370827656153e-05,
+      "loss": 0.4445,
+      "step": 3275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.801856965207338e-05,
+      "loss": 0.5007,
+      "step": 3276
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7898164684767515e-05,
+      "loss": 0.4702,
+      "step": 3277
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7778156024716498e-05,
+      "loss": 0.4231,
+      "step": 3278
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.765854377057219e-05,
+      "loss": 0.4464,
+      "step": 3279
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7539328020660716e-05,
+      "loss": 0.5225,
+      "step": 3280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7420508872982244e-05,
+      "loss": 0.484,
+      "step": 3281
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7302086425210973e-05,
+      "loss": 0.4813,
+      "step": 3282
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.718406077469503e-05,
+      "loss": 0.4493,
+      "step": 3283
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.706643201845609e-05,
+      "loss": 0.4271,
+      "step": 3284
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6949200253189966e-05,
+      "loss": 0.499,
+      "step": 3285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.683236557526574e-05,
+      "loss": 0.4299,
+      "step": 3286
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6715928080726418e-05,
+      "loss": 0.4666,
+      "step": 3287
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.659988786528821e-05,
+      "loss": 0.4755,
+      "step": 3288
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6484245024340983e-05,
+      "loss": 0.4536,
+      "step": 3289
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.63689996529478e-05,
+      "loss": 0.4854,
+      "step": 3290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6254151845844943e-05,
+      "loss": 0.5039,
+      "step": 3291
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.613970169744211e-05,
+      "loss": 0.4769,
+      "step": 3292
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6025649301821876e-05,
+      "loss": 0.4917,
+      "step": 3293
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5911994752739965e-05,
+      "loss": 0.473,
+      "step": 3294
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.579873814362498e-05,
+      "loss": 0.4702,
+      "step": 3295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5685879567578554e-05,
+      "loss": 0.491,
+      "step": 3296
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.557341911737481e-05,
+      "loss": 0.4783,
+      "step": 3297
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5461356885461074e-05,
+      "loss": 0.4928,
+      "step": 3298
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5349692963956773e-05,
+      "loss": 0.4923,
+      "step": 3299
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5238427444654367e-05,
+      "loss": 0.4913,
+      "step": 3300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5127560419018527e-05,
+      "loss": 0.4453,
+      "step": 3301
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5017091978186515e-05,
+      "loss": 0.4834,
+      "step": 3302
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4907022212967803e-05,
+      "loss": 0.431,
+      "step": 3303
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4797351213844178e-05,
+      "loss": 0.4839,
+      "step": 3304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4688079070969752e-05,
+      "loss": 0.4478,
+      "step": 3305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4579205874170554e-05,
+      "loss": 0.4745,
+      "step": 3306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4470731712944884e-05,
+      "loss": 0.4567,
+      "step": 3307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4362656676462748e-05,
+      "loss": 0.4932,
+      "step": 3308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4254980853566246e-05,
+      "loss": 0.4771,
+      "step": 3309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4147704332769407e-05,
+      "loss": 0.4852,
+      "step": 3310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4040827202257744e-05,
+      "loss": 0.4957,
+      "step": 3311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3934349549888647e-05,
+      "loss": 0.4628,
+      "step": 3312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.38282714631911e-05,
+      "loss": 0.4706,
+      "step": 3313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.372259302936546e-05,
+      "loss": 0.4849,
+      "step": 3314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3617314335283792e-05,
+      "loss": 0.4718,
+      "step": 3315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3512435467489425e-05,
+      "loss": 0.4542,
+      "step": 3316
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3407956512197005e-05,
+      "loss": 0.5,
+      "step": 3317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3303877555292442e-05,
+      "loss": 0.4634,
+      "step": 3318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.320019868233302e-05,
+      "loss": 0.4778,
+      "step": 3319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.309691997854684e-05,
+      "loss": 0.4761,
+      "step": 3320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2994041528833267e-05,
+      "loss": 0.4493,
+      "step": 3321
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.289156341776243e-05,
+      "loss": 0.4543,
+      "step": 3322
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2789485729575613e-05,
+      "loss": 0.4626,
+      "step": 3323
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2687808548184798e-05,
+      "loss": 0.4884,
+      "step": 3324
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2586531957172687e-05,
+      "loss": 0.4944,
+      "step": 3325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2485656039792792e-05,
+      "loss": 0.4752,
+      "step": 3326
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2385180878969115e-05,
+      "loss": 0.4907,
+      "step": 3327
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2285106557296478e-05,
+      "loss": 0.4086,
+      "step": 3328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2185433157039904e-05,
+      "loss": 0.4564,
+      "step": 3329
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2086160760135078e-05,
+      "loss": 0.4656,
+      "step": 3330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1987289448187776e-05,
+      "loss": 0.4459,
+      "step": 3331
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1888819302474374e-05,
+      "loss": 0.4315,
+      "step": 3332
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1790750403941231e-05,
+      "loss": 0.4703,
+      "step": 3333
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1693082833205027e-05,
+      "loss": 0.4712,
+      "step": 3334
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1595816670552429e-05,
+      "loss": 0.475,
+      "step": 3335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1498951995940144e-05,
+      "loss": 0.4863,
+      "step": 3336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1402488888994922e-05,
+      "loss": 0.4756,
+      "step": 3337
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1306427429013222e-05,
+      "loss": 0.4783,
+      "step": 3338
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1210767694961654e-05,
+      "loss": 0.4822,
+      "step": 3339
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1115509765476206e-05,
+      "loss": 0.4689,
+      "step": 3340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1020653718862906e-05,
+      "loss": 0.4667,
+      "step": 3341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0926199633097156e-05,
+      "loss": 0.4491,
+      "step": 3342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0832147585824181e-05,
+      "loss": 0.4939,
+      "step": 3343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0738497654358414e-05,
+      "loss": 0.4735,
+      "step": 3344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0645249915683997e-05,
+      "loss": 0.4405,
+      "step": 3345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0552404446454333e-05,
+      "loss": 0.489,
+      "step": 3346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0459961322992152e-05,
+      "loss": 0.4694,
+      "step": 3347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0367920621289494e-05,
+      "loss": 0.4967,
+      "step": 3348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0276282417007399e-05,
+      "loss": 0.5215,
+      "step": 3349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0185046785476382e-05,
+      "loss": 0.4825,
+      "step": 3350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0094213801695729e-05,
+      "loss": 0.4338,
+      "step": 3351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0003783540333877e-05,
+      "loss": 0.474,
+      "step": 3352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.913756075728086e-06,
+      "loss": 0.4884,
+      "step": 3353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.824131481884658e-06,
+      "loss": 0.4669,
+      "step": 3354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.7349098324786e-06,
+      "loss": 0.4648,
+      "step": 3355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.646091200853802e-06,
+      "loss": 0.4557,
+      "step": 3356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.557675660022746e-06,
+      "loss": 0.4442,
+      "step": 3357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.469663282666518e-06,
+      "loss": 0.4888,
+      "step": 3358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.38205414113491e-06,
+      "loss": 0.4381,
+      "step": 3359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.294848307446201e-06,
+      "loss": 0.468,
+      "step": 3360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.208045853287105e-06,
+      "loss": 0.4558,
+      "step": 3361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.121646850012765e-06,
+      "loss": 0.4525,
+      "step": 3362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.035651368646646e-06,
+      "loss": 0.4604,
+      "step": 3363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.950059479880591e-06,
+      "loss": 0.4753,
+      "step": 3364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.864871254074481e-06,
+      "loss": 0.4388,
+      "step": 3365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.780086761256634e-06,
+      "loss": 0.4814,
+      "step": 3366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.695706071123244e-06,
+      "loss": 0.5011,
+      "step": 3367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.611729253038659e-06,
+      "loss": 0.4675,
+      "step": 3368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.528156376035324e-06,
+      "loss": 0.4562,
+      "step": 3369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.444987508813451e-06,
+      "loss": 0.4452,
+      "step": 3370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.36222271974113e-06,
+      "loss": 0.438,
+      "step": 3371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.27986207685455e-06,
+      "loss": 0.4423,
+      "step": 3372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.197905647857385e-06,
+      "loss": 0.4739,
+      "step": 3373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.116353500121187e-06,
+      "loss": 0.4843,
+      "step": 3374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.035205700685167e-06,
+      "loss": 0.4888,
+      "step": 3375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.954462316256016e-06,
+      "loss": 0.4492,
+      "step": 3376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.874123413208145e-06,
+      "loss": 0.4515,
+      "step": 3377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.794189057583333e-06,
+      "loss": 0.4431,
+      "step": 3378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.71465931509091e-06,
+      "loss": 0.454,
+      "step": 3379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.635534251107579e-06,
+      "loss": 0.4575,
+      "step": 3380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.5568139306771975e-06,
+      "loss": 0.4858,
+      "step": 3381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.4784984185112255e-06,
+      "loss": 0.4432,
+      "step": 3382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.400587778988055e-06,
+      "loss": 0.4451,
+      "step": 3383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.323082076153509e-06,
+      "loss": 0.4661,
+      "step": 3384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.2459813737203454e-06,
+      "loss": 0.4385,
+      "step": 3385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.1692857350685315e-06,
+      "loss": 0.4708,
+      "step": 3386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.092995223244858e-06,
+      "loss": 0.4631,
+      "step": 3387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.0171099009634365e-06,
+      "loss": 0.447,
+      "step": 3388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.941629830604868e-06,
+      "loss": 0.4736,
+      "step": 3389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.866555074216962e-06,
+      "loss": 0.4744,
+      "step": 3390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.791885693514133e-06,
+      "loss": 0.4795,
+      "step": 3391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.7176217498776675e-06,
+      "loss": 0.4364,
+      "step": 3392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.6437633043555655e-06,
+      "loss": 0.4545,
+      "step": 3393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.5703104176624285e-06,
+      "loss": 0.4441,
+      "step": 3394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.497263150179511e-06,
+      "loss": 0.4579,
+      "step": 3395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.424621561954613e-06,
+      "loss": 0.4996,
+      "step": 3396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.352385712702191e-06,
+      "loss": 0.4762,
+      "step": 3397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.2805556618028556e-06,
+      "loss": 0.4658,
+      "step": 3398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.2091314683039304e-06,
+      "loss": 0.4091,
+      "step": 3399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.138113190918948e-06,
+      "loss": 0.4464,
+      "step": 3400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.067500888027877e-06,
+      "loss": 0.4308,
+      "step": 3401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.997294617676841e-06,
+      "loss": 0.4803,
+      "step": 3402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.927494437578229e-06,
+      "loss": 0.4169,
+      "step": 3403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.858100405110645e-06,
+      "loss": 0.4822,
+      "step": 3404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.789112577318789e-06,
+      "loss": 0.4718,
+      "step": 3405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.720531010913521e-06,
+      "loss": 0.4579,
+      "step": 3406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.652355762271521e-06,
+      "loss": 0.4746,
+      "step": 3407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5845868874357386e-06,
+      "loss": 0.4801,
+      "step": 3408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.517224442114888e-06,
+      "loss": 0.4484,
+      "step": 3409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.450268481683618e-06,
+      "loss": 0.4923,
+      "step": 3410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.383719061182457e-06,
+      "loss": 0.3986,
+      "step": 3411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.317576235317756e-06,
+      "loss": 0.4623,
+      "step": 3412
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.251840058461577e-06,
+      "loss": 0.4729,
+      "step": 3413
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.186510584651693e-06,
+      "loss": 0.4746,
+      "step": 3414
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.121587867591649e-06,
+      "loss": 0.4811,
+      "step": 3415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0570719606504165e-06,
+      "loss": 0.4425,
+      "step": 3416
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.992962916862853e-06,
+      "loss": 0.4807,
+      "step": 3417
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.92926078892908e-06,
+      "loss": 0.4512,
+      "step": 3418
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.865965629214819e-06,
+      "loss": 0.4742,
+      "step": 3419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.803077489751395e-06,
+      "loss": 0.5055,
+      "step": 3420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.740596422235233e-06,
+      "loss": 0.498,
+      "step": 3421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6785224780284685e-06,
+      "loss": 0.4752,
+      "step": 3422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6168557081582855e-06,
+      "loss": 0.4624,
+      "step": 3423
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.555596163317299e-06,
+      "loss": 0.4586,
+      "step": 3424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.49474389386334e-06,
+      "loss": 0.4726,
+      "step": 3425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.434298949819449e-06,
+      "loss": 0.4799,
+      "step": 3426
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.374261380873768e-06,
+      "loss": 0.4746,
+      "step": 3427
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.314631236379707e-06,
+      "loss": 0.4556,
+      "step": 3428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.255408565355612e-06,
+      "loss": 0.4443,
+      "step": 3429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.196593416484873e-06,
+      "loss": 0.4811,
+      "step": 3430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.138185838116038e-06,
+      "loss": 0.4988,
+      "step": 3431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.080185878262421e-06,
+      "loss": 0.4287,
+      "step": 3432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.02259358460233e-06,
+      "loss": 0.483,
+      "step": 3433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.965409004479059e-06,
+      "loss": 0.4802,
+      "step": 3434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.908632184900562e-06,
+      "loss": 0.4746,
+      "step": 3435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.852263172539727e-06,
+      "loss": 0.4799,
+      "step": 3436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.796302013734099e-06,
+      "loss": 0.4559,
+      "step": 3437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7407487544861563e-06,
+      "loss": 0.4478,
+      "step": 3438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.685603440462759e-06,
+      "loss": 0.4561,
+      "step": 3439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.630866116995757e-06,
+      "loss": 0.4612,
+      "step": 3440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.576536829081323e-06,
+      "loss": 0.4635,
+      "step": 3441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.522615621380454e-06,
+      "loss": 0.454,
+      "step": 3442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.469102538218416e-06,
+      "loss": 0.4619,
+      "step": 3443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4159976235852984e-06,
+      "loss": 0.3906,
+      "step": 3444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.363300921135348e-06,
+      "loss": 0.4679,
+      "step": 3445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.311012474187469e-06,
+      "loss": 0.4231,
+      "step": 3446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2591323257248896e-06,
+      "loss": 0.4803,
+      "step": 3447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.207660518395161e-06,
+      "loss": 0.5065,
+      "step": 3448
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.156597094510272e-06,
+      "loss": 0.4422,
+      "step": 3449
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1059420960462548e-06,
+      "loss": 0.4736,
+      "step": 3450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.055695564643801e-06,
+      "loss": 0.4081,
+      "step": 3451
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.005857541607371e-06,
+      "loss": 0.4811,
+      "step": 3452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.956428067906025e-06,
+      "loss": 0.4613,
+      "step": 3453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9074071841727055e-06,
+      "loss": 0.4508,
+      "step": 3454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8587949307045113e-06,
+      "loss": 0.4801,
+      "step": 3455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8105913474628653e-06,
+      "loss": 0.4541,
+      "step": 3456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.762796474072904e-06,
+      "loss": 0.4561,
+      "step": 3457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7154103498239767e-06,
+      "loss": 0.479,
+      "step": 3458
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6684330136694245e-06,
+      "loss": 0.4789,
+      "step": 3459
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6218645042265786e-06,
+      "loss": 0.4783,
+      "step": 3460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5757048597765396e-06,
+      "loss": 0.4791,
+      "step": 3461
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.529954118264455e-06,
+      "loss": 0.4856,
+      "step": 3462
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4846123172992953e-06,
+      "loss": 0.5047,
+      "step": 3463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.439679494153857e-06,
+      "loss": 0.4962,
+      "step": 3464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.395155685764705e-06,
+      "loss": 0.4456,
+      "step": 3465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3510409287322844e-06,
+      "loss": 0.4392,
+      "step": 3466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3073352593206977e-06,
+      "loss": 0.4556,
+      "step": 3467
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2640387134577057e-06,
+      "loss": 0.4735,
+      "step": 3468
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2211513267348937e-06,
+      "loss": 0.4897,
+      "step": 3469
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.178673134407394e-06,
+      "loss": 0.4508,
+      "step": 3470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.136604171394052e-06,
+      "loss": 0.4773,
+      "step": 3471
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0949444722772047e-06,
+      "loss": 0.4617,
+      "step": 3472
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0536940713028474e-06,
+      "loss": 0.4783,
+      "step": 3473
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.012853002380466e-06,
+      "loss": 0.4792,
+      "step": 3474
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9724212990830937e-06,
+      "loss": 0.5193,
+      "step": 3475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9323989946472554e-06,
+      "loss": 0.5055,
+      "step": 3476
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8927861219728005e-06,
+      "loss": 0.4606,
+      "step": 3477
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8535827136232365e-06,
+      "loss": 0.4694,
+      "step": 3478
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.81478880182534e-06,
+      "loss": 0.5083,
+      "step": 3479
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.776404418469213e-06,
+      "loss": 0.4714,
+      "step": 3480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7384295951083927e-06,
+      "loss": 0.4581,
+      "step": 3481
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7008643629596864e-06,
+      "loss": 0.4586,
+      "step": 3482
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6637087529033922e-06,
+      "loss": 0.4924,
+      "step": 3483
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.626962795482745e-06,
+      "loss": 0.4838,
+      "step": 3484
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5906265209045257e-06,
+      "loss": 0.4882,
+      "step": 3485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.554699959038619e-06,
+      "loss": 0.4575,
+      "step": 3486
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5191831394180676e-06,
+      "loss": 0.4525,
+      "step": 3487
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4840760912391283e-06,
+      "loss": 0.5442,
+      "step": 3488
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4493788433612708e-06,
+      "loss": 0.4563,
+      "step": 3489
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4150914243070135e-06,
+      "loss": 0.5078,
+      "step": 3490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3812138622620319e-06,
+      "loss": 0.4113,
+      "step": 3491
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.347746185074994e-06,
+      "loss": 0.4591,
+      "step": 3492
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3146884202577259e-06,
+      "loss": 0.4716,
+      "step": 3493
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.282040594985101e-06,
+      "loss": 0.4469,
+      "step": 3494
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2498027360948739e-06,
+      "loss": 0.4519,
+      "step": 3495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2179748700879012e-06,
+      "loss": 0.4364,
+      "step": 3496
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.186557023127921e-06,
+      "loss": 0.4733,
+      "step": 3497
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1555492210418294e-06,
+      "loss": 0.4552,
+      "step": 3498
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.124951489319126e-06,
+      "loss": 0.4701,
+      "step": 3499
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0947638531125236e-06,
+      "loss": 0.4624,
+      "step": 3500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0649863372373946e-06,
+      "loss": 0.4618,
+      "step": 3501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0356189661721028e-06,
+      "loss": 0.4661,
+      "step": 3502
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.006661764057837e-06,
+      "loss": 0.5057,
+      "step": 3503
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.781147546985004e-07,
+      "loss": 0.4835,
+      "step": 3504
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.499779615609882e-07,
+      "loss": 0.4646,
+      "step": 3505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.222514077748545e-07,
+      "loss": 0.4786,
+      "step": 3506
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.949351161324226e-07,
+      "loss": 0.4285,
+      "step": 3507
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.680291090888415e-07,
+      "loss": 0.4274,
+      "step": 3508
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.415334087619186e-07,
+      "loss": 0.4542,
+      "step": 3509
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.15448036932176e-07,
+      "loss": 0.5106,
+      "step": 3510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.897730150428495e-07,
+      "loss": 0.4651,
+      "step": 3511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.645083641998896e-07,
+      "loss": 0.448,
+      "step": 3512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.396541051717942e-07,
+      "loss": 0.4691,
+      "step": 3513
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.152102583898312e-07,
+      "loss": 0.4442,
+      "step": 3514
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.911768439478161e-07,
+      "loss": 0.4457,
+      "step": 3515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.675538816022231e-07,
+      "loss": 0.4718,
+      "step": 3516
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.443413907720186e-07,
+      "loss": 0.5131,
+      "step": 3517
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.215393905388278e-07,
+      "loss": 0.4839,
+      "step": 3518
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.991478996468236e-07,
+      "loss": 0.4451,
+      "step": 3519
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.771669365027821e-07,
+      "loss": 0.4679,
+      "step": 3520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.555965191758606e-07,
+      "loss": 0.4295,
+      "step": 3521
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.344366653978194e-07,
+      "loss": 0.4596,
+      "step": 3522
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.136873925629671e-07,
+      "loss": 0.486,
+      "step": 3523
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.933487177280482e-07,
+      "loss": 0.4689,
+      "step": 3524
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7342065761224464e-07,
+      "loss": 0.4446,
+      "step": 3525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.539032285972855e-07,
+      "loss": 0.4419,
+      "step": 3526
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3479644672717035e-07,
+      "loss": 0.4998,
+      "step": 3527
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1610032770855735e-07,
+      "loss": 0.439,
+      "step": 3528
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9781488691037484e-07,
+      "loss": 0.4714,
+      "step": 3529
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7994013936398787e-07,
+      "loss": 0.4392,
+      "step": 3530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6247609976319816e-07,
+      "loss": 0.4648,
+      "step": 3531
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4542278246413314e-07,
+      "loss": 0.4679,
+      "step": 3532
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2878020148530144e-07,
+      "loss": 0.4485,
+      "step": 3533
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1254837050764817e-07,
+      "loss": 0.4435,
+      "step": 3534
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9672730287427786e-07,
+      "loss": 0.427,
+      "step": 3535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8131701159078705e-07,
+      "loss": 0.4729,
+      "step": 3536
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6631750932515354e-07,
+      "loss": 0.4891,
+      "step": 3537
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.517288084074587e-07,
+      "loss": 0.4893,
+      "step": 3538
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3755092083022068e-07,
+      "loss": 0.4844,
+      "step": 3539
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2378385824833868e-07,
+      "loss": 0.4736,
+      "step": 3540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1042763197876014e-07,
+      "loss": 0.442,
+      "step": 3541
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.974822530009801e-07,
+      "loss": 0.4917,
+      "step": 3542
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8494773195648628e-07,
+      "loss": 0.4724,
+      "step": 3543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7282407914931408e-07,
+      "loss": 0.5022,
+      "step": 3544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.61111304545436e-07,
+      "loss": 0.4758,
+      "step": 3545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.498094177733722e-07,
+      "loss": 0.4622,
+      "step": 3546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3891842812363554e-07,
+      "loss": 0.4928,
+      "step": 3547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2843834454911995e-07,
+      "loss": 0.4421,
+      "step": 3548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1836917566482308e-07,
+      "loss": 0.4386,
+      "step": 3549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0871092974806818e-07,
+      "loss": 0.4524,
+      "step": 3550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.946361473822663e-08,
+      "loss": 0.49,
+      "step": 3551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.06272382371065e-08,
+      "loss": 0.4918,
+      "step": 3552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.220180750850847e-08,
+      "loss": 0.4647,
+      "step": 3553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.418732947844786e-08,
+      "loss": 0.4731,
+      "step": 3554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.658381073515462e-08,
+      "loss": 0.4656,
+      "step": 3555
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.939125752901786e-08,
+      "loss": 0.4663,
+      "step": 3556
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.260967577269682e-08,
+      "loss": 0.4834,
+      "step": 3557
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.623907104084335e-08,
+      "loss": 0.4261,
+      "step": 3558
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.027944857032395e-08,
+      "loss": 0.4384,
+      "step": 3559
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4730813260275277e-08,
+      "loss": 0.464,
+      "step": 3560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.959316967188208e-08,
+      "loss": 0.4873,
+      "step": 3561
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.486652202848827e-08,
+      "loss": 0.5201,
+      "step": 3562
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.055087421554136e-08,
+      "loss": 0.4493,
+      "step": 3563
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6646229780759027e-08,
+      "loss": 0.4725,
+      "step": 3564
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3152591933851544e-08,
+      "loss": 0.4301,
+      "step": 3565
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0069963546743833e-08,
+      "loss": 0.4607,
+      "step": 3566
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.3983471535199464e-09,
+      "loss": 0.5264,
+      "step": 3567
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1377449503120514e-09,
+      "loss": 0.479,
+      "step": 3568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2881587954669555e-09,
+      "loss": 0.4252,
+      "step": 3569
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8495902093795814e-09,
+      "loss": 0.4821,
+      "step": 3570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.220403746039828e-10,
+      "loss": 0.4576,
+      "step": 3571
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0551013590885958e-10,
+      "loss": 0.4813,
+      "step": 3572
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.4642,
+      "step": 3573
+    },
+    {
+      "epoch": 1.0,
+      "step": 3573,
+      "total_flos": 1.039660262490112e+16,
+      "train_loss": 0.5047162341831313,
+      "train_runtime": 121049.3168,
+      "train_samples_per_second": 7.557,
+      "train_steps_per_second": 0.03
+    }
+  ],
+  "max_steps": 3573,
+  "num_train_epochs": 1,
+  "total_flos": 1.039660262490112e+16,
+  "trial_name": null,
+  "trial_params": null
+}