diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,50794 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 8463,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.839121190644934e-09,
+      "loss": 1.8715,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.839121190644934e-09,
+      "loss": 1.8444,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.839121190644934e-09,
+      "loss": 1.7487,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.839121190644934e-09,
+      "loss": 1.7431,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.839121190644934e-09,
+      "loss": 1.7051,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.839121190644934e-09,
+      "loss": 1.765,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3678242381289867e-08,
+      "loss": 1.6936,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.05173635719348e-08,
+      "loss": 1.7675,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7356484762579735e-08,
+      "loss": 1.8017,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.419560595322467e-08,
+      "loss": 1.6692,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.10347271438696e-08,
+      "loss": 1.5946,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.787384833451453e-08,
+      "loss": 1.6274,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.471296952515947e-08,
+      "loss": 1.7206,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.15520907158044e-08,
+      "loss": 1.5408,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.839121190644934e-08,
+      "loss": 1.6956,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.523033309709427e-08,
+      "loss": 1.5425,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.20694542877392e-08,
+      "loss": 1.331,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.890857547838413e-08,
+      "loss": 1.7456,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.574769666902907e-08,
+      "loss": 1.5695,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.02586817859674e-07,
+      "loss": 1.7221,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0942593905031894e-07,
+      "loss": 1.6627,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1626506024096388e-07,
+      "loss": 1.6837,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.231041814316088e-07,
+      "loss": 1.8599,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2994330262225372e-07,
+      "loss": 1.6502,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3678242381289867e-07,
+      "loss": 1.428,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.436215450035436e-07,
+      "loss": 1.5285,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5046066619418854e-07,
+      "loss": 1.557,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5729978738483344e-07,
+      "loss": 1.6779,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.641389085754784e-07,
+      "loss": 1.5442,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.709780297661233e-07,
+      "loss": 1.6438,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7781715095676826e-07,
+      "loss": 1.4316,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8465627214741318e-07,
+      "loss": 1.5006,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9149539333805813e-07,
+      "loss": 1.5123,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9833451452870306e-07,
+      "loss": 1.3744,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.05173635719348e-07,
+      "loss": 1.764,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1201275690999293e-07,
+      "loss": 1.4543,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1885187810063788e-07,
+      "loss": 1.5392,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.256909992912828e-07,
+      "loss": 1.6538,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3253012048192775e-07,
+      "loss": 1.5975,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.393692416725727e-07,
+      "loss": 1.4766,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.462083628632176e-07,
+      "loss": 1.8305,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.530474840538625e-07,
+      "loss": 1.519,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5988660524450744e-07,
+      "loss": 1.7496,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.667257264351524e-07,
+      "loss": 1.6243,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7356484762579734e-07,
+      "loss": 1.4024,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8040396881644224e-07,
+      "loss": 1.7447,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.872430900070872e-07,
+      "loss": 1.7954,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9408221119773214e-07,
+      "loss": 1.4861,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.009213323883771e-07,
+      "loss": 1.5682,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.07760453579022e-07,
+      "loss": 1.4324,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.145995747696669e-07,
+      "loss": 1.594,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2143869596031183e-07,
+      "loss": 1.3681,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.282778171509568e-07,
+      "loss": 1.4933,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.351169383416017e-07,
+      "loss": 1.6546,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.419560595322466e-07,
+      "loss": 1.5517,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4879518072289157e-07,
+      "loss": 1.5122,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.556343019135365e-07,
+      "loss": 1.6267,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6247342310418147e-07,
+      "loss": 1.7144,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6931254429482637e-07,
+      "loss": 1.4391,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.761516654854713e-07,
+      "loss": 1.6791,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8299078667611627e-07,
+      "loss": 1.4391,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.898299078667612e-07,
+      "loss": 1.4856,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.966690290574061e-07,
+      "loss": 1.5216,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0350815024805106e-07,
+      "loss": 1.2594,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.10347271438696e-07,
+      "loss": 1.668,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1718639262934096e-07,
+      "loss": 1.5657,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2402551381998586e-07,
+      "loss": 1.5136,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.308646350106308e-07,
+      "loss": 1.5883,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3770375620127576e-07,
+      "loss": 1.5008,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4454287739192065e-07,
+      "loss": 1.6395,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.513819985825656e-07,
+      "loss": 1.5169,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5822111977321055e-07,
+      "loss": 1.5823,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.650602409638555e-07,
+      "loss": 1.3637,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.718993621545004e-07,
+      "loss": 1.5409,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.787384833451454e-07,
+      "loss": 1.484,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.855776045357903e-07,
+      "loss": 1.5469,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.924167257264352e-07,
+      "loss": 1.4551,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992558469170801e-07,
+      "loss": 1.5268,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.06094968107725e-07,
+      "loss": 1.7707,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1293408929837e-07,
+      "loss": 1.5139,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.197732104890149e-07,
+      "loss": 1.5047,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.266123316796599e-07,
+      "loss": 1.5509,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.334514528703048e-07,
+      "loss": 1.5809,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.402905740609497e-07,
+      "loss": 1.522,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.471296952515947e-07,
+      "loss": 1.5008,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.539688164422396e-07,
+      "loss": 1.4134,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.608079376328845e-07,
+      "loss": 1.5975,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.676470588235295e-07,
+      "loss": 1.4516,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.744861800141744e-07,
+      "loss": 1.771,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.813253012048193e-07,
+      "loss": 1.2963,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.881644223954643e-07,
+      "loss": 1.6311,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.950035435861092e-07,
+      "loss": 1.4633,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.018426647767542e-07,
+      "loss": 1.4308,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.086817859673991e-07,
+      "loss": 1.4285,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.15520907158044e-07,
+      "loss": 1.3545,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.22360028348689e-07,
+      "loss": 1.6221,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.291991495393338e-07,
+      "loss": 1.782,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.360382707299788e-07,
+      "loss": 1.5305,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.428773919206237e-07,
+      "loss": 1.5877,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.497165131112687e-07,
+      "loss": 1.5455,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.565556343019136e-07,
+      "loss": 1.4851,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.633947554925586e-07,
+      "loss": 1.6264,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.702338766832035e-07,
+      "loss": 1.395,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.770729978738485e-07,
+      "loss": 1.4944,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.839121190644932e-07,
+      "loss": 1.6793,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.907512402551382e-07,
+      "loss": 1.5729,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.975903614457831e-07,
+      "loss": 1.3989,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.044294826364281e-07,
+      "loss": 1.4137,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.11268603827073e-07,
+      "loss": 1.3441,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.18107725017718e-07,
+      "loss": 1.2622,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.249468462083629e-07,
+      "loss": 1.4315,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.317859673990079e-07,
+      "loss": 1.4022,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.386250885896527e-07,
+      "loss": 1.6398,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.454642097802977e-07,
+      "loss": 1.802,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.523033309709426e-07,
+      "loss": 1.3356,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.591424521615876e-07,
+      "loss": 1.26,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.659815733522325e-07,
+      "loss": 1.5263,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.728206945428775e-07,
+      "loss": 1.5047,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.796598157335224e-07,
+      "loss": 1.471,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.864989369241674e-07,
+      "loss": 1.5811,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.933380581148122e-07,
+      "loss": 1.4855,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.001771793054571e-07,
+      "loss": 1.3252,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.070163004961021e-07,
+      "loss": 1.391,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.13855421686747e-07,
+      "loss": 1.5863,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.20694542877392e-07,
+      "loss": 1.6783,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.275336640680369e-07,
+      "loss": 1.5255,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.343727852586819e-07,
+      "loss": 1.46,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.412119064493267e-07,
+      "loss": 1.6057,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.480510276399717e-07,
+      "loss": 1.6283,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.548901488306166e-07,
+      "loss": 1.4631,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.617292700212616e-07,
+      "loss": 1.4629,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.685683912119065e-07,
+      "loss": 1.5159,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.754075124025515e-07,
+      "loss": 1.5731,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.822466335931963e-07,
+      "loss": 1.6601,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.890857547838413e-07,
+      "loss": 1.4904,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.959248759744862e-07,
+      "loss": 1.7212,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.027639971651312e-07,
+      "loss": 1.5519,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.096031183557761e-07,
+      "loss": 1.3859,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.164422395464211e-07,
+      "loss": 1.5933,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.23281360737066e-07,
+      "loss": 1.1922,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.30120481927711e-07,
+      "loss": 1.5382,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.369596031183558e-07,
+      "loss": 1.5093,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.437987243090008e-07,
+      "loss": 1.5495,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.506378454996457e-07,
+      "loss": 1.519,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.574769666902908e-07,
+      "loss": 1.5193,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.643160878809355e-07,
+      "loss": 1.5166,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.711552090715806e-07,
+      "loss": 1.487,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.779943302622255e-07,
+      "loss": 1.3993,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.848334514528704e-07,
+      "loss": 1.3559,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.916725726435153e-07,
+      "loss": 1.3952,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985116938341602e-07,
+      "loss": 1.5226,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0053508150248053e-06,
+      "loss": 1.4577,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.01218993621545e-06,
+      "loss": 1.6104,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.019029057406095e-06,
+      "loss": 1.6399,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.02586817859674e-06,
+      "loss": 1.724,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0327072997873849e-06,
+      "loss": 1.6623,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0395464209780298e-06,
+      "loss": 1.5186,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0463855421686749e-06,
+      "loss": 1.4716,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0532246633593198e-06,
+      "loss": 1.6722,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0600637845499647e-06,
+      "loss": 1.4189,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0669029057406096e-06,
+      "loss": 1.3879,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0737420269312545e-06,
+      "loss": 1.5922,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0805811481218994e-06,
+      "loss": 1.5659,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0874202693125445e-06,
+      "loss": 1.3949,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0942593905031894e-06,
+      "loss": 1.5814,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1010985116938343e-06,
+      "loss": 1.334,
+      "step": 166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1079376328844792e-06,
+      "loss": 1.5909,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1147767540751243e-06,
+      "loss": 1.594,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.121615875265769e-06,
+      "loss": 1.2914,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.128454996456414e-06,
+      "loss": 1.4019,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.135294117647059e-06,
+      "loss": 1.5838,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1421332388377038e-06,
+      "loss": 1.4015,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1489723600283487e-06,
+      "loss": 1.4188,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1558114812189939e-06,
+      "loss": 1.7561,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1626506024096385e-06,
+      "loss": 1.4874,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1694897236002836e-06,
+      "loss": 1.4665,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1763288447909285e-06,
+      "loss": 1.5998,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1831679659815734e-06,
+      "loss": 1.3722,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1900070871722183e-06,
+      "loss": 1.5706,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1968462083628632e-06,
+      "loss": 1.6116,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2036853295535083e-06,
+      "loss": 1.3618,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.210524450744153e-06,
+      "loss": 1.403,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2173635719347981e-06,
+      "loss": 1.397,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.224202693125443e-06,
+      "loss": 1.6803,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.231041814316088e-06,
+      "loss": 1.3016,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2378809355067328e-06,
+      "loss": 1.4439,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.244720056697378e-06,
+      "loss": 1.4795,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2515591778880228e-06,
+      "loss": 1.71,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2583982990786675e-06,
+      "loss": 1.4786,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2652374202693126e-06,
+      "loss": 1.4015,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2720765414599575e-06,
+      "loss": 1.5435,
+      "step": 191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2789156626506026e-06,
+      "loss": 1.6755,
+      "step": 192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2857547838412473e-06,
+      "loss": 1.4329,
+      "step": 193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2925939050318924e-06,
+      "loss": 1.6058,
+      "step": 194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2994330262225373e-06,
+      "loss": 1.3317,
+      "step": 195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3062721474131824e-06,
+      "loss": 1.4466,
+      "step": 196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3131112686038271e-06,
+      "loss": 1.5264,
+      "step": 197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.319950389794472e-06,
+      "loss": 1.4535,
+      "step": 198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3267895109851171e-06,
+      "loss": 1.3282,
+      "step": 199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.333628632175762e-06,
+      "loss": 1.6914,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.340467753366407e-06,
+      "loss": 1.5067,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3473068745570518e-06,
+      "loss": 1.2453,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.354145995747697e-06,
+      "loss": 1.4477,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3609851169383418e-06,
+      "loss": 1.5599,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3678242381289865e-06,
+      "loss": 1.5808,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3746633593196316e-06,
+      "loss": 1.5365,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3815024805102765e-06,
+      "loss": 1.6219,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3883416017009214e-06,
+      "loss": 1.4088,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3951807228915663e-06,
+      "loss": 1.4545,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4020198440822114e-06,
+      "loss": 1.439,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4088589652728563e-06,
+      "loss": 1.569,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.415698086463501e-06,
+      "loss": 1.2621,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.422537207654146e-06,
+      "loss": 1.4951,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.429376328844791e-06,
+      "loss": 1.5812,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.436215450035436e-06,
+      "loss": 1.5478,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4430545712260808e-06,
+      "loss": 1.5173,
+      "step": 216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4498936924167259e-06,
+      "loss": 1.4747,
+      "step": 217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4567328136073708e-06,
+      "loss": 1.2636,
+      "step": 218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4635719347980159e-06,
+      "loss": 1.5629,
+      "step": 219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4704110559886606e-06,
+      "loss": 1.6758,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4772501771793055e-06,
+      "loss": 1.7308,
+      "step": 221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4840892983699506e-06,
+      "loss": 1.4257,
+      "step": 222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4909284195605955e-06,
+      "loss": 1.6204,
+      "step": 223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4977675407512404e-06,
+      "loss": 1.4418,
+      "step": 224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5046066619418853e-06,
+      "loss": 1.5288,
+      "step": 225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5114457831325304e-06,
+      "loss": 1.3497,
+      "step": 226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5182849043231753e-06,
+      "loss": 1.4355,
+      "step": 227
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.52512402551382e-06,
+      "loss": 1.3728,
+      "step": 228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.531963146704465e-06,
+      "loss": 1.3832,
+      "step": 229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.53880226789511e-06,
+      "loss": 1.5348,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.545641389085755e-06,
+      "loss": 1.47,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5524805102763998e-06,
+      "loss": 1.3731,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5593196314670449e-06,
+      "loss": 1.6907,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5661587526576898e-06,
+      "loss": 1.5679,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5729978738483349e-06,
+      "loss": 1.3102,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5798369950389796e-06,
+      "loss": 1.354,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5866761162296244e-06,
+      "loss": 1.4913,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5935152374202696e-06,
+      "loss": 1.4311,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6003543586109142e-06,
+      "loss": 1.5376,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6071934798015593e-06,
+      "loss": 1.5139,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6140326009922042e-06,
+      "loss": 1.3753,
+      "step": 241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6208717221828494e-06,
+      "loss": 1.6419,
+      "step": 242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.627710843373494e-06,
+      "loss": 1.6436,
+      "step": 243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.634549964564139e-06,
+      "loss": 1.363,
+      "step": 244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.641389085754784e-06,
+      "loss": 1.3627,
+      "step": 245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.648228206945429e-06,
+      "loss": 1.5574,
+      "step": 246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6550673281360738e-06,
+      "loss": 1.5029,
+      "step": 247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6619064493267187e-06,
+      "loss": 1.5216,
+      "step": 248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6687455705173638e-06,
+      "loss": 1.4265,
+      "step": 249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6755846917080087e-06,
+      "loss": 1.4827,
+      "step": 250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6824238128986534e-06,
+      "loss": 1.4402,
+      "step": 251
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6892629340892985e-06,
+      "loss": 1.3257,
+      "step": 252
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6961020552799434e-06,
+      "loss": 1.5956,
+      "step": 253
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7029411764705885e-06,
+      "loss": 1.3966,
+      "step": 254
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7097802976612332e-06,
+      "loss": 1.4914,
+      "step": 255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7166194188518781e-06,
+      "loss": 1.4852,
+      "step": 256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7234585400425232e-06,
+      "loss": 1.5436,
+      "step": 257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7302976612331681e-06,
+      "loss": 1.4358,
+      "step": 258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.737136782423813e-06,
+      "loss": 1.4587,
+      "step": 259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.743975903614458e-06,
+      "loss": 1.483,
+      "step": 260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.750815024805103e-06,
+      "loss": 1.6225,
+      "step": 261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.757654145995748e-06,
+      "loss": 1.4948,
+      "step": 262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7644932671863926e-06,
+      "loss": 1.4562,
+      "step": 263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7713323883770377e-06,
+      "loss": 1.6169,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7781715095676826e-06,
+      "loss": 1.5434,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7850106307583275e-06,
+      "loss": 1.4315,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7918497519489724e-06,
+      "loss": 1.4823,
+      "step": 267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7986888731396175e-06,
+      "loss": 1.7113,
+      "step": 268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8055279943302624e-06,
+      "loss": 1.5952,
+      "step": 269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.812367115520907e-06,
+      "loss": 1.5285,
+      "step": 270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8192062367115522e-06,
+      "loss": 1.4147,
+      "step": 271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.826045357902197e-06,
+      "loss": 1.3552,
+      "step": 272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8328844790928422e-06,
+      "loss": 1.3659,
+      "step": 273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8397236002834869e-06,
+      "loss": 1.3553,
+      "step": 274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.846562721474132e-06,
+      "loss": 1.4068,
+      "step": 275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8534018426647769e-06,
+      "loss": 1.5675,
+      "step": 276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.860240963855422e-06,
+      "loss": 1.6656,
+      "step": 277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8670800850460667e-06,
+      "loss": 1.3317,
+      "step": 278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8739192062367116e-06,
+      "loss": 1.5659,
+      "step": 279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8807583274273567e-06,
+      "loss": 1.4274,
+      "step": 280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8875974486180016e-06,
+      "loss": 1.395,
+      "step": 281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8944365698086465e-06,
+      "loss": 1.3699,
+      "step": 282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9012756909992914e-06,
+      "loss": 1.421,
+      "step": 283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9081148121899365e-06,
+      "loss": 1.1859,
+      "step": 284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9149539333805816e-06,
+      "loss": 1.2602,
+      "step": 285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9217930545712263e-06,
+      "loss": 1.5275,
+      "step": 286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.928632175761871e-06,
+      "loss": 1.5932,
+      "step": 287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.935471296952516e-06,
+      "loss": 1.4615,
+      "step": 288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.942310418143161e-06,
+      "loss": 1.4693,
+      "step": 289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949149539333806e-06,
+      "loss": 1.5477,
+      "step": 290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955988660524451e-06,
+      "loss": 1.7134,
+      "step": 291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.962827781715096e-06,
+      "loss": 1.3821,
+      "step": 292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9696669029057408e-06,
+      "loss": 1.3775,
+      "step": 293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765060240963855e-06,
+      "loss": 1.439,
+      "step": 294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9833451452870306e-06,
+      "loss": 1.5947,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9901842664776757e-06,
+      "loss": 1.5893,
+      "step": 296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970233876683204e-06,
+      "loss": 1.4113,
+      "step": 297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.0038625088589655e-06,
+      "loss": 1.4532,
+      "step": 298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.0107016300496106e-06,
+      "loss": 1.3527,
+      "step": 299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.0175407512402553e-06,
+      "loss": 1.6066,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.0243798724309e-06,
+      "loss": 1.3007,
+      "step": 301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.031218993621545e-06,
+      "loss": 1.337,
+      "step": 302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.03805811481219e-06,
+      "loss": 1.5231,
+      "step": 303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.0448972360028353e-06,
+      "loss": 1.3204,
+      "step": 304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.05173635719348e-06,
+      "loss": 1.6934,
+      "step": 305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.058575478384125e-06,
+      "loss": 1.3971,
+      "step": 306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.0654145995747697e-06,
+      "loss": 1.4162,
+      "step": 307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.072253720765415e-06,
+      "loss": 1.6336,
+      "step": 308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.0790928419560595e-06,
+      "loss": 1.4454,
+      "step": 309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.0859319631467046e-06,
+      "loss": 1.389,
+      "step": 310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.0927710843373498e-06,
+      "loss": 1.3532,
+      "step": 311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.0996102055279944e-06,
+      "loss": 1.4574,
+      "step": 312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1064493267186395e-06,
+      "loss": 1.6152,
+      "step": 313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1132884479092842e-06,
+      "loss": 1.3837,
+      "step": 314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1201275690999293e-06,
+      "loss": 1.6821,
+      "step": 315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1269666902905744e-06,
+      "loss": 1.5569,
+      "step": 316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.133805811481219e-06,
+      "loss": 1.407,
+      "step": 317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1406449326718642e-06,
+      "loss": 1.2265,
+      "step": 318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.147484053862509e-06,
+      "loss": 1.4075,
+      "step": 319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.154323175053154e-06,
+      "loss": 1.5733,
+      "step": 320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1611622962437987e-06,
+      "loss": 1.6691,
+      "step": 321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168001417434444e-06,
+      "loss": 1.2532,
+      "step": 322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.174840538625089e-06,
+      "loss": 1.5834,
+      "step": 323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1816796598157336e-06,
+      "loss": 1.4876,
+      "step": 324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1885187810063787e-06,
+      "loss": 1.4078,
+      "step": 325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1953579021970234e-06,
+      "loss": 1.5793,
+      "step": 326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.2021970233876685e-06,
+      "loss": 1.3475,
+      "step": 327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.209036144578313e-06,
+      "loss": 1.4956,
+      "step": 328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.2158752657689583e-06,
+      "loss": 1.3979,
+      "step": 329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.2227143869596034e-06,
+      "loss": 1.3414,
+      "step": 330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.2295535081502485e-06,
+      "loss": 1.5282,
+      "step": 331
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.236392629340893e-06,
+      "loss": 1.6308,
+      "step": 332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.243231750531538e-06,
+      "loss": 1.3227,
+      "step": 333
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.250070871722183e-06,
+      "loss": 1.3984,
+      "step": 334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.256909992912828e-06,
+      "loss": 1.2708,
+      "step": 335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.263749114103473e-06,
+      "loss": 1.5071,
+      "step": 336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.270588235294118e-06,
+      "loss": 1.6033,
+      "step": 337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.2774273564847626e-06,
+      "loss": 1.578,
+      "step": 338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.2842664776754077e-06,
+      "loss": 1.495,
+      "step": 339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.2911055988660524e-06,
+      "loss": 1.4454,
+      "step": 340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.2979447200566975e-06,
+      "loss": 1.3735,
+      "step": 341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.3047838412473426e-06,
+      "loss": 1.5209,
+      "step": 342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.3116229624379877e-06,
+      "loss": 1.3207,
+      "step": 343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.3184620836286324e-06,
+      "loss": 1.4929,
+      "step": 344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.325301204819277e-06,
+      "loss": 1.239,
+      "step": 345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.332140326009922e-06,
+      "loss": 1.4765,
+      "step": 346
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.3389794472005673e-06,
+      "loss": 1.3025,
+      "step": 347
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.345818568391212e-06,
+      "loss": 1.3845,
+      "step": 348
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.352657689581857e-06,
+      "loss": 1.4877,
+      "step": 349
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.359496810772502e-06,
+      "loss": 1.3576,
+      "step": 350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.366335931963147e-06,
+      "loss": 1.5987,
+      "step": 351
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.3731750531537916e-06,
+      "loss": 1.3839,
+      "step": 352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3800141743444367e-06,
+      "loss": 1.3665,
+      "step": 353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3868532955350818e-06,
+      "loss": 1.5284,
+      "step": 354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3936924167257265e-06,
+      "loss": 1.4173,
+      "step": 355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4005315379163716e-06,
+      "loss": 1.5877,
+      "step": 356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4073706591070167e-06,
+      "loss": 1.3718,
+      "step": 357
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4142097802976614e-06,
+      "loss": 1.3945,
+      "step": 358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.421048901488306e-06,
+      "loss": 1.4284,
+      "step": 359
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4278880226789516e-06,
+      "loss": 1.4668,
+      "step": 360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4347271438695963e-06,
+      "loss": 1.5845,
+      "step": 361
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.441566265060241e-06,
+      "loss": 1.4568,
+      "step": 362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.448405386250886e-06,
+      "loss": 1.5375,
+      "step": 363
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.455244507441531e-06,
+      "loss": 1.604,
+      "step": 364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.462083628632176e-06,
+      "loss": 1.5883,
+      "step": 365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.468922749822821e-06,
+      "loss": 1.5841,
+      "step": 366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4757618710134656e-06,
+      "loss": 1.3456,
+      "step": 367
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.482600992204111e-06,
+      "loss": 1.4221,
+      "step": 368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.489440113394756e-06,
+      "loss": 1.3768,
+      "step": 369
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4962792345854005e-06,
+      "loss": 1.2467,
+      "step": 370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5031183557760457e-06,
+      "loss": 1.3872,
+      "step": 371
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5099574769666903e-06,
+      "loss": 1.4867,
+      "step": 372
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.516796598157335e-06,
+      "loss": 1.4102,
+      "step": 373
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5236357193479806e-06,
+      "loss": 1.6185,
+      "step": 374
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5304748405386252e-06,
+      "loss": 1.5516,
+      "step": 375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5373139617292703e-06,
+      "loss": 1.4949,
+      "step": 376
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.544153082919915e-06,
+      "loss": 1.4767,
+      "step": 377
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.55099220411056e-06,
+      "loss": 1.5722,
+      "step": 378
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5578313253012053e-06,
+      "loss": 1.5394,
+      "step": 379
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.56467044649185e-06,
+      "loss": 1.4005,
+      "step": 380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.5715095676824946e-06,
+      "loss": 1.4561,
+      "step": 381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.57834868887314e-06,
+      "loss": 1.3169,
+      "step": 382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.585187810063785e-06,
+      "loss": 1.5613,
+      "step": 383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.5920269312544295e-06,
+      "loss": 1.4469,
+      "step": 384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.5988660524450746e-06,
+      "loss": 1.3695,
+      "step": 385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6057051736357193e-06,
+      "loss": 1.3292,
+      "step": 386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.612544294826365e-06,
+      "loss": 1.6583,
+      "step": 387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6193834160170095e-06,
+      "loss": 1.4687,
+      "step": 388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6262225372076542e-06,
+      "loss": 1.4377,
+      "step": 389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6330616583982993e-06,
+      "loss": 1.5093,
+      "step": 390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.639900779588944e-06,
+      "loss": 1.3416,
+      "step": 391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.646739900779589e-06,
+      "loss": 1.4928,
+      "step": 392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6535790219702342e-06,
+      "loss": 1.4534,
+      "step": 393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.660418143160879e-06,
+      "loss": 1.4744,
+      "step": 394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.667257264351524e-06,
+      "loss": 1.1967,
+      "step": 395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.674096385542169e-06,
+      "loss": 1.4346,
+      "step": 396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.680935506732814e-06,
+      "loss": 1.6137,
+      "step": 397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.687774627923459e-06,
+      "loss": 1.3855,
+      "step": 398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6946137491141036e-06,
+      "loss": 1.5561,
+      "step": 399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7014528703047483e-06,
+      "loss": 1.5237,
+      "step": 400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.708291991495394e-06,
+      "loss": 1.5143,
+      "step": 401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7151311126860385e-06,
+      "loss": 1.5725,
+      "step": 402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7219702338766836e-06,
+      "loss": 1.2384,
+      "step": 403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7288093550673283e-06,
+      "loss": 1.331,
+      "step": 404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.735648476257973e-06,
+      "loss": 1.3861,
+      "step": 405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7424875974486185e-06,
+      "loss": 1.5006,
+      "step": 406
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.749326718639263e-06,
+      "loss": 1.505,
+      "step": 407
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.756165839829908e-06,
+      "loss": 1.6172,
+      "step": 408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.763004961020553e-06,
+      "loss": 1.6168,
+      "step": 409
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.769844082211198e-06,
+      "loss": 1.4731,
+      "step": 410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7766832034018428e-06,
+      "loss": 1.2798,
+      "step": 411
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.783522324592488e-06,
+      "loss": 1.4535,
+      "step": 412
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7903614457831326e-06,
+      "loss": 1.533,
+      "step": 413
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.797200566973778e-06,
+      "loss": 1.2132,
+      "step": 414
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.804039688164423e-06,
+      "loss": 1.6602,
+      "step": 415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8108788093550675e-06,
+      "loss": 1.2773,
+      "step": 416
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8177179305457126e-06,
+      "loss": 1.2671,
+      "step": 417
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8245570517363573e-06,
+      "loss": 1.5864,
+      "step": 418
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.831396172927002e-06,
+      "loss": 1.462,
+      "step": 419
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8382352941176475e-06,
+      "loss": 1.4615,
+      "step": 420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.845074415308292e-06,
+      "loss": 1.5308,
+      "step": 421
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8519135364989373e-06,
+      "loss": 1.4724,
+      "step": 422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.858752657689582e-06,
+      "loss": 1.405,
+      "step": 423
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8655917788802267e-06,
+      "loss": 1.4641,
+      "step": 424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.872430900070872e-06,
+      "loss": 1.503,
+      "step": 425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.879270021261517e-06,
+      "loss": 1.4809,
+      "step": 426
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8861091424521616e-06,
+      "loss": 1.5893,
+      "step": 427
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8929482636428067e-06,
+      "loss": 1.5087,
+      "step": 428
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8997873848334518e-06,
+      "loss": 1.3151,
+      "step": 429
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.906626506024097e-06,
+      "loss": 1.523,
+      "step": 430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9134656272147416e-06,
+      "loss": 1.3015,
+      "step": 431
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9203047484053862e-06,
+      "loss": 1.3316,
+      "step": 432
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9271438695960318e-06,
+      "loss": 1.4854,
+      "step": 433
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9339829907866765e-06,
+      "loss": 1.5881,
+      "step": 434
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.940822111977321e-06,
+      "loss": 1.486,
+      "step": 435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9476612331679663e-06,
+      "loss": 1.391,
+      "step": 436
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.954500354358611e-06,
+      "loss": 1.4863,
+      "step": 437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9613394755492556e-06,
+      "loss": 1.5521,
+      "step": 438
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.968178596739901e-06,
+      "loss": 1.3109,
+      "step": 439
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.975017717930546e-06,
+      "loss": 1.3784,
+      "step": 440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.981856839121191e-06,
+      "loss": 1.4054,
+      "step": 441
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9886959603118356e-06,
+      "loss": 1.5252,
+      "step": 442
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9955350815024807e-06,
+      "loss": 1.5296,
+      "step": 443
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.002374202693126e-06,
+      "loss": 1.3844,
+      "step": 444
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.0092133238837705e-06,
+      "loss": 1.3203,
+      "step": 445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.0160524450744152e-06,
+      "loss": 1.4948,
+      "step": 446
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.0228915662650608e-06,
+      "loss": 1.4986,
+      "step": 447
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.0297306874557054e-06,
+      "loss": 1.6544,
+      "step": 448
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.0365698086463505e-06,
+      "loss": 1.2748,
+      "step": 449
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.0434089298369952e-06,
+      "loss": 1.5376,
+      "step": 450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.05024805102764e-06,
+      "loss": 1.5373,
+      "step": 451
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.0570871722182854e-06,
+      "loss": 1.542,
+      "step": 452
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.06392629340893e-06,
+      "loss": 1.3466,
+      "step": 453
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.070765414599575e-06,
+      "loss": 1.4417,
+      "step": 454
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.07760453579022e-06,
+      "loss": 1.5483,
+      "step": 455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.0844436569808646e-06,
+      "loss": 1.6372,
+      "step": 456
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.09128277817151e-06,
+      "loss": 1.4326,
+      "step": 457
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.098121899362155e-06,
+      "loss": 1.5435,
+      "step": 458
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.1049610205527995e-06,
+      "loss": 1.7263,
+      "step": 459
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.1118001417434446e-06,
+      "loss": 1.476,
+      "step": 460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.1186392629340897e-06,
+      "loss": 1.5591,
+      "step": 461
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.1254783841247344e-06,
+      "loss": 1.4496,
+      "step": 462
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.1323175053153795e-06,
+      "loss": 1.457,
+      "step": 463
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.139156626506024e-06,
+      "loss": 1.435,
+      "step": 464
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.1459957476966697e-06,
+      "loss": 1.4846,
+      "step": 465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.1528348688873144e-06,
+      "loss": 1.4825,
+      "step": 466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.159673990077959e-06,
+      "loss": 1.2637,
+      "step": 467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.1665131112686042e-06,
+      "loss": 1.4008,
+      "step": 468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.173352232459249e-06,
+      "loss": 1.4888,
+      "step": 469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.1801913536498936e-06,
+      "loss": 1.4217,
+      "step": 470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.187030474840539e-06,
+      "loss": 1.3799,
+      "step": 471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.193869596031184e-06,
+      "loss": 1.4198,
+      "step": 472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.2007087172218285e-06,
+      "loss": 1.2073,
+      "step": 473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.2075478384124736e-06,
+      "loss": 1.3749,
+      "step": 474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.2143869596031187e-06,
+      "loss": 1.4519,
+      "step": 475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.221226080793764e-06,
+      "loss": 1.3673,
+      "step": 476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.2280652019844085e-06,
+      "loss": 1.4306,
+      "step": 477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.234904323175053e-06,
+      "loss": 1.3441,
+      "step": 478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.2417434443656987e-06,
+      "loss": 1.3653,
+      "step": 479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.2485825655563434e-06,
+      "loss": 1.5275,
+      "step": 480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.255421686746988e-06,
+      "loss": 1.4287,
+      "step": 481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.262260807937633e-06,
+      "loss": 1.2423,
+      "step": 482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.269099929128278e-06,
+      "loss": 1.2263,
+      "step": 483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.2759390503189234e-06,
+      "loss": 1.412,
+      "step": 484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.282778171509568e-06,
+      "loss": 1.3048,
+      "step": 485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.2896172927002128e-06,
+      "loss": 1.6132,
+      "step": 486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.296456413890858e-06,
+      "loss": 1.4367,
+      "step": 487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.3032955350815026e-06,
+      "loss": 1.4871,
+      "step": 488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.3101346562721477e-06,
+      "loss": 1.6591,
+      "step": 489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.3169737774627928e-06,
+      "loss": 1.5128,
+      "step": 490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.3238128986534375e-06,
+      "loss": 1.6612,
+      "step": 491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.3306520198440826e-06,
+      "loss": 1.3199,
+      "step": 492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.3374911410347277e-06,
+      "loss": 1.2728,
+      "step": 493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.3443302622253724e-06,
+      "loss": 1.5381,
+      "step": 494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.3511693834160175e-06,
+      "loss": 1.4725,
+      "step": 495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.358008504606662e-06,
+      "loss": 1.5458,
+      "step": 496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.364847625797307e-06,
+      "loss": 1.3567,
+      "step": 497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.3716867469879524e-06,
+      "loss": 1.563,
+      "step": 498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.378525868178597e-06,
+      "loss": 1.4825,
+      "step": 499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.3853649893692417e-06,
+      "loss": 1.6024,
+      "step": 500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.392204110559887e-06,
+      "loss": 1.5733,
+      "step": 501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.3990432317505315e-06,
+      "loss": 1.4582,
+      "step": 502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.405882352941177e-06,
+      "loss": 1.4569,
+      "step": 503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.4127214741318218e-06,
+      "loss": 1.4596,
+      "step": 504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.4195605953224664e-06,
+      "loss": 1.2713,
+      "step": 505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.4263997165131115e-06,
+      "loss": 1.4326,
+      "step": 506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.4332388377037562e-06,
+      "loss": 1.4892,
+      "step": 507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.4400779588944013e-06,
+      "loss": 1.5464,
+      "step": 508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.4469170800850464e-06,
+      "loss": 1.4276,
+      "step": 509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.453756201275691e-06,
+      "loss": 1.3592,
+      "step": 510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.4605953224663362e-06,
+      "loss": 1.4395,
+      "step": 511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.4674344436569813e-06,
+      "loss": 1.4188,
+      "step": 512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.474273564847626e-06,
+      "loss": 1.4539,
+      "step": 513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.481112686038271e-06,
+      "loss": 1.4052,
+      "step": 514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.487951807228916e-06,
+      "loss": 1.662,
+      "step": 515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.4947909284195605e-06,
+      "loss": 1.5029,
+      "step": 516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.501630049610206e-06,
+      "loss": 1.4132,
+      "step": 517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.5084691708008507e-06,
+      "loss": 1.4409,
+      "step": 518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.515308291991496e-06,
+      "loss": 1.557,
+      "step": 519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.5221474131821405e-06,
+      "loss": 1.3902,
+      "step": 520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.528986534372785e-06,
+      "loss": 1.2584,
+      "step": 521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.5358256555634307e-06,
+      "loss": 1.369,
+      "step": 522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.5426647767540754e-06,
+      "loss": 1.5062,
+      "step": 523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.54950389794472e-06,
+      "loss": 1.3024,
+      "step": 524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.5563430191353652e-06,
+      "loss": 1.439,
+      "step": 525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.5631821403260103e-06,
+      "loss": 1.3635,
+      "step": 526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.570021261516655e-06,
+      "loss": 1.6773,
+      "step": 527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.5768603827073e-06,
+      "loss": 1.4072,
+      "step": 528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.583699503897945e-06,
+      "loss": 1.3059,
+      "step": 529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.5905386250885903e-06,
+      "loss": 1.4315,
+      "step": 530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.597377746279235e-06,
+      "loss": 1.6508,
+      "step": 531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.6042168674698797e-06,
+      "loss": 1.5061,
+      "step": 532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.611055988660525e-06,
+      "loss": 1.4723,
+      "step": 533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.6178951098511695e-06,
+      "loss": 1.2792,
+      "step": 534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.624734231041814e-06,
+      "loss": 1.3722,
+      "step": 535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.6315733522324597e-06,
+      "loss": 1.5845,
+      "step": 536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.6384124734231044e-06,
+      "loss": 1.2646,
+      "step": 537
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.6452515946137495e-06,
+      "loss": 1.5166,
+      "step": 538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.652090715804394e-06,
+      "loss": 1.4042,
+      "step": 539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.6589298369950393e-06,
+      "loss": 1.6329,
+      "step": 540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.6657689581856844e-06,
+      "loss": 1.4617,
+      "step": 541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.672608079376329e-06,
+      "loss": 1.2684,
+      "step": 542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.6794472005669738e-06,
+      "loss": 1.3549,
+      "step": 543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.6862863217576193e-06,
+      "loss": 1.5399,
+      "step": 544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.693125442948264e-06,
+      "loss": 1.6374,
+      "step": 545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.699964564138909e-06,
+      "loss": 1.5276,
+      "step": 546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7068036853295538e-06,
+      "loss": 1.2705,
+      "step": 547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7136428065201985e-06,
+      "loss": 1.353,
+      "step": 548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.720481927710844e-06,
+      "loss": 1.4756,
+      "step": 549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7273210489014887e-06,
+      "loss": 1.5384,
+      "step": 550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7341601700921334e-06,
+      "loss": 1.5296,
+      "step": 551
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7409992912827785e-06,
+      "loss": 1.3179,
+      "step": 552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.747838412473423e-06,
+      "loss": 1.47,
+      "step": 553
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7546775336640683e-06,
+      "loss": 1.5823,
+      "step": 554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7615166548547134e-06,
+      "loss": 1.3614,
+      "step": 555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.768355776045358e-06,
+      "loss": 1.5106,
+      "step": 556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.775194897236003e-06,
+      "loss": 1.4042,
+      "step": 557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7820340184266483e-06,
+      "loss": 1.462,
+      "step": 558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.788873139617293e-06,
+      "loss": 1.5047,
+      "step": 559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.795712260807938e-06,
+      "loss": 1.4462,
+      "step": 560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.8025513819985828e-06,
+      "loss": 1.3342,
+      "step": 561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.8093905031892274e-06,
+      "loss": 1.4769,
+      "step": 562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.816229624379873e-06,
+      "loss": 1.3801,
+      "step": 563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.823068745570518e-06,
+      "loss": 1.4479,
+      "step": 564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.829907866761163e-06,
+      "loss": 1.4532,
+      "step": 565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.836746987951808e-06,
+      "loss": 1.4084,
+      "step": 566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.8435861091424526e-06,
+      "loss": 1.3414,
+      "step": 567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.850425230333097e-06,
+      "loss": 1.6028,
+      "step": 568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.857264351523742e-06,
+      "loss": 1.4299,
+      "step": 569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.8641034727143875e-06,
+      "loss": 1.4286,
+      "step": 570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.870942593905032e-06,
+      "loss": 1.5072,
+      "step": 571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.877781715095677e-06,
+      "loss": 1.4954,
+      "step": 572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.884620836286322e-06,
+      "loss": 1.1087,
+      "step": 573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.891459957476967e-06,
+      "loss": 1.3225,
+      "step": 574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.898299078667612e-06,
+      "loss": 1.6836,
+      "step": 575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.905138199858257e-06,
+      "loss": 1.5701,
+      "step": 576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.911977321048902e-06,
+      "loss": 1.4734,
+      "step": 577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.918816442239547e-06,
+      "loss": 1.4421,
+      "step": 578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.925655563430192e-06,
+      "loss": 1.4623,
+      "step": 579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.932494684620837e-06,
+      "loss": 1.5029,
+      "step": 580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9393338058114815e-06,
+      "loss": 1.427,
+      "step": 581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.946172927002126e-06,
+      "loss": 1.5172,
+      "step": 582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.953012048192771e-06,
+      "loss": 1.3919,
+      "step": 583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9598511693834164e-06,
+      "loss": 1.3256,
+      "step": 584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.966690290574061e-06,
+      "loss": 1.3074,
+      "step": 585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.973529411764706e-06,
+      "loss": 1.4226,
+      "step": 586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.980368532955351e-06,
+      "loss": 1.3951,
+      "step": 587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.987207654145996e-06,
+      "loss": 1.4965,
+      "step": 588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.994046775336641e-06,
+      "loss": 1.5741,
+      "step": 589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.000885896527286e-06,
+      "loss": 1.5386,
+      "step": 590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.007725017717931e-06,
+      "loss": 1.3827,
+      "step": 591
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.014564138908576e-06,
+      "loss": 1.3192,
+      "step": 592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.021403260099221e-06,
+      "loss": 1.672,
+      "step": 593
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.028242381289866e-06,
+      "loss": 1.2499,
+      "step": 594
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.0350815024805105e-06,
+      "loss": 1.442,
+      "step": 595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.041920623671155e-06,
+      "loss": 1.4431,
+      "step": 596
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.0487597448618e-06,
+      "loss": 1.6406,
+      "step": 597
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.055598866052445e-06,
+      "loss": 1.5904,
+      "step": 598
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.06243798724309e-06,
+      "loss": 1.262,
+      "step": 599
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.069277108433736e-06,
+      "loss": 1.3862,
+      "step": 600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.07611622962438e-06,
+      "loss": 1.4074,
+      "step": 601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.082955350815025e-06,
+      "loss": 1.385,
+      "step": 602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.0897944720056705e-06,
+      "loss": 1.2434,
+      "step": 603
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.096633593196315e-06,
+      "loss": 1.4026,
+      "step": 604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.10347271438696e-06,
+      "loss": 1.2408,
+      "step": 605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.110311835577605e-06,
+      "loss": 1.3635,
+      "step": 606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.11715095676825e-06,
+      "loss": 1.2396,
+      "step": 607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.123990077958895e-06,
+      "loss": 1.5136,
+      "step": 608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.1308291991495395e-06,
+      "loss": 1.5621,
+      "step": 609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.137668320340184e-06,
+      "loss": 1.2891,
+      "step": 610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.14450744153083e-06,
+      "loss": 1.3997,
+      "step": 611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.151346562721474e-06,
+      "loss": 1.4552,
+      "step": 612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.158185683912119e-06,
+      "loss": 1.3528,
+      "step": 613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.165024805102765e-06,
+      "loss": 1.2161,
+      "step": 614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.171863926293409e-06,
+      "loss": 1.291,
+      "step": 615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.178703047484054e-06,
+      "loss": 1.4603,
+      "step": 616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.1855421686746995e-06,
+      "loss": 1.4261,
+      "step": 617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.192381289865344e-06,
+      "loss": 1.4793,
+      "step": 618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.199220411055989e-06,
+      "loss": 1.5209,
+      "step": 619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2060595322466336e-06,
+      "loss": 1.3715,
+      "step": 620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.212898653437279e-06,
+      "loss": 1.4378,
+      "step": 621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.219737774627924e-06,
+      "loss": 1.2491,
+      "step": 622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2265768958185685e-06,
+      "loss": 1.4236,
+      "step": 623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.233416017009213e-06,
+      "loss": 1.4648,
+      "step": 624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.240255138199859e-06,
+      "loss": 1.1777,
+      "step": 625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.247094259390503e-06,
+      "loss": 1.4574,
+      "step": 626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.253933380581149e-06,
+      "loss": 1.3109,
+      "step": 627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2607725017717936e-06,
+      "loss": 1.5761,
+      "step": 628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.267611622962438e-06,
+      "loss": 1.6044,
+      "step": 629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.274450744153084e-06,
+      "loss": 1.5157,
+      "step": 630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2812898653437285e-06,
+      "loss": 1.3396,
+      "step": 631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.288128986534373e-06,
+      "loss": 1.3652,
+      "step": 632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.294968107725018e-06,
+      "loss": 1.2655,
+      "step": 633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.3018072289156625e-06,
+      "loss": 1.6335,
+      "step": 634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.308646350106308e-06,
+      "loss": 1.6558,
+      "step": 635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.315485471296953e-06,
+      "loss": 1.5505,
+      "step": 636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3223245924875974e-06,
+      "loss": 1.7017,
+      "step": 637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.329163713678243e-06,
+      "loss": 1.5214,
+      "step": 638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.336002834868888e-06,
+      "loss": 1.4149,
+      "step": 639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.342841956059532e-06,
+      "loss": 1.4456,
+      "step": 640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.349681077250178e-06,
+      "loss": 1.4904,
+      "step": 641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3565201984408225e-06,
+      "loss": 1.3723,
+      "step": 642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.363359319631467e-06,
+      "loss": 1.207,
+      "step": 643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.370198440822113e-06,
+      "loss": 1.563,
+      "step": 644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3770375620127574e-06,
+      "loss": 1.5267,
+      "step": 645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.383876683203402e-06,
+      "loss": 1.2942,
+      "step": 646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.390715804394047e-06,
+      "loss": 1.4081,
+      "step": 647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3975549255846915e-06,
+      "loss": 1.415,
+      "step": 648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.404394046775337e-06,
+      "loss": 1.4245,
+      "step": 649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.411233167965982e-06,
+      "loss": 1.3412,
+      "step": 650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.418072289156626e-06,
+      "loss": 1.6336,
+      "step": 651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.424911410347272e-06,
+      "loss": 1.277,
+      "step": 652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.431750531537917e-06,
+      "loss": 1.4614,
+      "step": 653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.438589652728562e-06,
+      "loss": 1.3168,
+      "step": 654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.445428773919207e-06,
+      "loss": 1.5636,
+      "step": 655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4522678951098515e-06,
+      "loss": 1.3785,
+      "step": 656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.459107016300497e-06,
+      "loss": 1.4564,
+      "step": 657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.465946137491142e-06,
+      "loss": 1.5395,
+      "step": 658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.472785258681786e-06,
+      "loss": 1.4932,
+      "step": 659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.479624379872431e-06,
+      "loss": 1.5384,
+      "step": 660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.486463501063076e-06,
+      "loss": 1.2064,
+      "step": 661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.493302622253721e-06,
+      "loss": 1.3875,
+      "step": 662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.500141743444366e-06,
+      "loss": 1.4349,
+      "step": 663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.506980864635011e-06,
+      "loss": 1.5439,
+      "step": 664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.513819985825656e-06,
+      "loss": 1.3215,
+      "step": 665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.520659107016301e-06,
+      "loss": 1.4153,
+      "step": 666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.527498228206946e-06,
+      "loss": 1.3765,
+      "step": 667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.534337349397591e-06,
+      "loss": 1.475,
+      "step": 668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.541176470588236e-06,
+      "loss": 1.2878,
+      "step": 669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5480155917788805e-06,
+      "loss": 1.4492,
+      "step": 670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.554854712969525e-06,
+      "loss": 1.5405,
+      "step": 671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.561693834160171e-06,
+      "loss": 1.5548,
+      "step": 672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.568532955350815e-06,
+      "loss": 1.6272,
+      "step": 673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.57537207654146e-06,
+      "loss": 1.1585,
+      "step": 674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.582211197732105e-06,
+      "loss": 1.3594,
+      "step": 675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.58905031892275e-06,
+      "loss": 1.6026,
+      "step": 676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.595889440113395e-06,
+      "loss": 1.3947,
+      "step": 677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.60272856130404e-06,
+      "loss": 1.3053,
+      "step": 678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609567682494685e-06,
+      "loss": 1.452,
+      "step": 679
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.61640680368533e-06,
+      "loss": 1.3663,
+      "step": 680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.623245924875975e-06,
+      "loss": 1.4655,
+      "step": 681
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.63008504606662e-06,
+      "loss": 1.3123,
+      "step": 682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.636924167257265e-06,
+      "loss": 1.6231,
+      "step": 683
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6437632884479095e-06,
+      "loss": 1.2821,
+      "step": 684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.650602409638554e-06,
+      "loss": 1.5182,
+      "step": 685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6574415308292e-06,
+      "loss": 1.5316,
+      "step": 686
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.664280652019844e-06,
+      "loss": 1.3208,
+      "step": 687
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.671119773210489e-06,
+      "loss": 1.3648,
+      "step": 688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.677958894401135e-06,
+      "loss": 1.5536,
+      "step": 689
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.684798015591779e-06,
+      "loss": 1.4597,
+      "step": 690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.691637136782424e-06,
+      "loss": 1.5339,
+      "step": 691
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6984762579730695e-06,
+      "loss": 1.4841,
+      "step": 692
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.705315379163714e-06,
+      "loss": 1.604,
+      "step": 693
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.712154500354359e-06,
+      "loss": 1.5355,
+      "step": 694
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.718993621545004e-06,
+      "loss": 1.4097,
+      "step": 695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.725832742735649e-06,
+      "loss": 1.4694,
+      "step": 696
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.732671863926294e-06,
+      "loss": 1.5263,
+      "step": 697
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7395109851169384e-06,
+      "loss": 1.557,
+      "step": 698
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.746350106307583e-06,
+      "loss": 1.2086,
+      "step": 699
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.753189227498229e-06,
+      "loss": 1.4142,
+      "step": 700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.760028348688873e-06,
+      "loss": 1.2563,
+      "step": 701
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.766867469879518e-06,
+      "loss": 1.4078,
+      "step": 702
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7737065910701636e-06,
+      "loss": 1.422,
+      "step": 703
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.780545712260808e-06,
+      "loss": 1.6681,
+      "step": 704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.787384833451453e-06,
+      "loss": 1.5196,
+      "step": 705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7942239546420985e-06,
+      "loss": 1.4838,
+      "step": 706
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.801063075832743e-06,
+      "loss": 1.2835,
+      "step": 707
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.807902197023389e-06,
+      "loss": 1.4242,
+      "step": 708
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.814741318214033e-06,
+      "loss": 1.4439,
+      "step": 709
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.821580439404678e-06,
+      "loss": 1.5584,
+      "step": 710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.828419560595323e-06,
+      "loss": 1.516,
+      "step": 711
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.835258681785968e-06,
+      "loss": 1.6753,
+      "step": 712
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.842097802976612e-06,
+      "loss": 1.2701,
+      "step": 713
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.848936924167258e-06,
+      "loss": 1.4611,
+      "step": 714
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.855776045357903e-06,
+      "loss": 1.5257,
+      "step": 715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.862615166548547e-06,
+      "loss": 1.2856,
+      "step": 716
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8694542877391925e-06,
+      "loss": 1.4945,
+      "step": 717
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.876293408929838e-06,
+      "loss": 1.4991,
+      "step": 718
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.883132530120482e-06,
+      "loss": 1.2995,
+      "step": 719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8899716513111274e-06,
+      "loss": 1.363,
+      "step": 720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.896810772501772e-06,
+      "loss": 1.4531,
+      "step": 721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.903649893692417e-06,
+      "loss": 1.5304,
+      "step": 722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.910489014883062e-06,
+      "loss": 1.5327,
+      "step": 723
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.917328136073707e-06,
+      "loss": 1.321,
+      "step": 724
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.924167257264352e-06,
+      "loss": 1.6692,
+      "step": 725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.931006378454996e-06,
+      "loss": 1.4016,
+      "step": 726
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.937845499645642e-06,
+      "loss": 1.2809,
+      "step": 727
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.944684620836287e-06,
+      "loss": 1.3369,
+      "step": 728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.951523742026931e-06,
+      "loss": 1.4269,
+      "step": 729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.958362863217577e-06,
+      "loss": 1.2484,
+      "step": 730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.965201984408222e-06,
+      "loss": 1.4684,
+      "step": 731
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.972041105598866e-06,
+      "loss": 1.5707,
+      "step": 732
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.978880226789512e-06,
+      "loss": 1.6972,
+      "step": 733
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.985719347980156e-06,
+      "loss": 1.4214,
+      "step": 734
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.992558469170801e-06,
+      "loss": 1.4125,
+      "step": 735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.999397590361447e-06,
+      "loss": 1.3745,
+      "step": 736
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.006236711552091e-06,
+      "loss": 1.3144,
+      "step": 737
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.013075832742736e-06,
+      "loss": 1.5752,
+      "step": 738
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.019914953933381e-06,
+      "loss": 1.4107,
+      "step": 739
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.026754075124026e-06,
+      "loss": 1.2986,
+      "step": 740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.03359319631467e-06,
+      "loss": 1.5031,
+      "step": 741
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.040432317505316e-06,
+      "loss": 1.5232,
+      "step": 742
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.047271438695961e-06,
+      "loss": 1.534,
+      "step": 743
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.054110559886605e-06,
+      "loss": 1.4171,
+      "step": 744
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.0609496810772505e-06,
+      "loss": 1.3993,
+      "step": 745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.067788802267896e-06,
+      "loss": 1.4774,
+      "step": 746
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.074627923458541e-06,
+      "loss": 1.3258,
+      "step": 747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.081467044649185e-06,
+      "loss": 1.454,
+      "step": 748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.08830616583983e-06,
+      "loss": 1.2553,
+      "step": 749
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.095145287030476e-06,
+      "loss": 1.3637,
+      "step": 750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.10198440822112e-06,
+      "loss": 1.6148,
+      "step": 751
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.108823529411765e-06,
+      "loss": 1.3962,
+      "step": 752
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.1156626506024105e-06,
+      "loss": 1.4365,
+      "step": 753
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.122501771793054e-06,
+      "loss": 1.6721,
+      "step": 754
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.1293408929837e-06,
+      "loss": 1.3598,
+      "step": 755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.136180014174345e-06,
+      "loss": 1.2297,
+      "step": 756
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.143019135364989e-06,
+      "loss": 1.6021,
+      "step": 757
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.149858256555635e-06,
+      "loss": 1.3298,
+      "step": 758
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.15669737774628e-06,
+      "loss": 1.393,
+      "step": 759
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.163536498936924e-06,
+      "loss": 1.5476,
+      "step": 760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.17037562012757e-06,
+      "loss": 1.337,
+      "step": 761
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.177214741318214e-06,
+      "loss": 1.3058,
+      "step": 762
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.184053862508859e-06,
+      "loss": 1.4285,
+      "step": 763
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.190892983699504e-06,
+      "loss": 1.2713,
+      "step": 764
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.197732104890149e-06,
+      "loss": 1.1921,
+      "step": 765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.204571226080795e-06,
+      "loss": 1.4354,
+      "step": 766
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.211410347271439e-06,
+      "loss": 1.3961,
+      "step": 767
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.218249468462084e-06,
+      "loss": 1.2099,
+      "step": 768
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.22508858965273e-06,
+      "loss": 1.3196,
+      "step": 769
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.2319277108433735e-06,
+      "loss": 1.5765,
+      "step": 770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.238766832034019e-06,
+      "loss": 1.3433,
+      "step": 771
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.245605953224664e-06,
+      "loss": 1.3933,
+      "step": 772
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.2524450744153084e-06,
+      "loss": 1.3362,
+      "step": 773
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.259284195605954e-06,
+      "loss": 1.4447,
+      "step": 774
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.266123316796599e-06,
+      "loss": 1.203,
+      "step": 775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.272962437987243e-06,
+      "loss": 1.3982,
+      "step": 776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.279801559177888e-06,
+      "loss": 1.3501,
+      "step": 777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.2866406803685335e-06,
+      "loss": 1.4542,
+      "step": 778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.293479801559178e-06,
+      "loss": 1.3856,
+      "step": 779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.300318922749823e-06,
+      "loss": 1.3478,
+      "step": 780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.3071580439404684e-06,
+      "loss": 1.5092,
+      "step": 781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.313997165131114e-06,
+      "loss": 1.382,
+      "step": 782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.320836286321758e-06,
+      "loss": 1.4151,
+      "step": 783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.327675407512403e-06,
+      "loss": 1.3178,
+      "step": 784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.334514528703048e-06,
+      "loss": 1.3228,
+      "step": 785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.341353649893693e-06,
+      "loss": 1.4805,
+      "step": 786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.348192771084338e-06,
+      "loss": 1.4018,
+      "step": 787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.355031892274983e-06,
+      "loss": 1.4349,
+      "step": 788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.361871013465628e-06,
+      "loss": 1.5248,
+      "step": 789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.368710134656272e-06,
+      "loss": 1.3459,
+      "step": 790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.375549255846918e-06,
+      "loss": 1.748,
+      "step": 791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.382388377037562e-06,
+      "loss": 1.3315,
+      "step": 792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.389227498228207e-06,
+      "loss": 1.3503,
+      "step": 793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.396066619418853e-06,
+      "loss": 1.5265,
+      "step": 794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.4029057406094966e-06,
+      "loss": 1.5608,
+      "step": 795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.409744861800142e-06,
+      "loss": 1.4146,
+      "step": 796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.416583982990788e-06,
+      "loss": 1.1639,
+      "step": 797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.4234231041814315e-06,
+      "loss": 1.3962,
+      "step": 798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.430262225372077e-06,
+      "loss": 1.3513,
+      "step": 799
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.437101346562722e-06,
+      "loss": 1.5561,
+      "step": 800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.443940467753367e-06,
+      "loss": 1.2406,
+      "step": 801
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.450779588944012e-06,
+      "loss": 1.6421,
+      "step": 802
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.457618710134657e-06,
+      "loss": 1.2742,
+      "step": 803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.464457831325302e-06,
+      "loss": 1.2864,
+      "step": 804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.471296952515946e-06,
+      "loss": 1.2796,
+      "step": 805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.4781360737065915e-06,
+      "loss": 1.405,
+      "step": 806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.484975194897237e-06,
+      "loss": 1.4565,
+      "step": 807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.491814316087881e-06,
+      "loss": 1.321,
+      "step": 808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.498653437278526e-06,
+      "loss": 1.6137,
+      "step": 809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.505492558469172e-06,
+      "loss": 1.5407,
+      "step": 810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.512331679659816e-06,
+      "loss": 1.3316,
+      "step": 811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.519170800850461e-06,
+      "loss": 1.3551,
+      "step": 812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.526009922041106e-06,
+      "loss": 1.2876,
+      "step": 813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.532849043231751e-06,
+      "loss": 1.4159,
+      "step": 814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.539688164422396e-06,
+      "loss": 1.5564,
+      "step": 815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.546527285613041e-06,
+      "loss": 1.3134,
+      "step": 816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.5533664068036856e-06,
+      "loss": 1.4915,
+      "step": 817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.56020552799433e-06,
+      "loss": 1.4424,
+      "step": 818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.567044649184976e-06,
+      "loss": 1.34,
+      "step": 819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.573883770375621e-06,
+      "loss": 1.3336,
+      "step": 820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.580722891566265e-06,
+      "loss": 1.3078,
+      "step": 821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.587562012756911e-06,
+      "loss": 1.375,
+      "step": 822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.594401133947556e-06,
+      "loss": 1.3208,
+      "step": 823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.6012402551382e-06,
+      "loss": 1.2888,
+      "step": 824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.608079376328846e-06,
+      "loss": 1.4237,
+      "step": 825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.61491849751949e-06,
+      "loss": 1.5424,
+      "step": 826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.621757618710135e-06,
+      "loss": 1.3466,
+      "step": 827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.62859673990078e-06,
+      "loss": 1.4849,
+      "step": 828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.635435861091425e-06,
+      "loss": 1.2663,
+      "step": 829
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.64227498228207e-06,
+      "loss": 1.2037,
+      "step": 830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.6491141034727145e-06,
+      "loss": 1.4038,
+      "step": 831
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.65595322466336e-06,
+      "loss": 1.489,
+      "step": 832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.662792345854004e-06,
+      "loss": 1.355,
+      "step": 833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6696314670446494e-06,
+      "loss": 1.462,
+      "step": 834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.676470588235295e-06,
+      "loss": 1.3426,
+      "step": 835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.68330970942594e-06,
+      "loss": 1.3372,
+      "step": 836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.690148830616584e-06,
+      "loss": 1.5733,
+      "step": 837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.69698795180723e-06,
+      "loss": 1.3974,
+      "step": 838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.7038270729978746e-06,
+      "loss": 1.5864,
+      "step": 839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.710666194188519e-06,
+      "loss": 1.543,
+      "step": 840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.717505315379164e-06,
+      "loss": 1.304,
+      "step": 841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.7243444365698095e-06,
+      "loss": 1.2431,
+      "step": 842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.731183557760453e-06,
+      "loss": 1.3556,
+      "step": 843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.738022678951099e-06,
+      "loss": 1.5154,
+      "step": 844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.744861800141744e-06,
+      "loss": 1.4695,
+      "step": 845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.751700921332388e-06,
+      "loss": 1.5256,
+      "step": 846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.758540042523034e-06,
+      "loss": 1.3392,
+      "step": 847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.765379163713679e-06,
+      "loss": 1.4356,
+      "step": 848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.772218284904323e-06,
+      "loss": 1.5091,
+      "step": 849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.779057406094969e-06,
+      "loss": 1.5996,
+      "step": 850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.785896527285613e-06,
+      "loss": 1.2103,
+      "step": 851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.792735648476258e-06,
+      "loss": 1.1692,
+      "step": 852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.7995747696669035e-06,
+      "loss": 1.3536,
+      "step": 853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.806413890857548e-06,
+      "loss": 1.5173,
+      "step": 854
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.813253012048194e-06,
+      "loss": 1.5292,
+      "step": 855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.820092133238838e-06,
+      "loss": 1.3394,
+      "step": 856
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.826931254429483e-06,
+      "loss": 1.3732,
+      "step": 857
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.833770375620129e-06,
+      "loss": 1.6056,
+      "step": 858
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.8406094968107725e-06,
+      "loss": 1.5194,
+      "step": 859
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.847448618001418e-06,
+      "loss": 1.4137,
+      "step": 860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.8542877391920636e-06,
+      "loss": 1.4529,
+      "step": 861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.861126860382707e-06,
+      "loss": 1.5093,
+      "step": 862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.867965981573353e-06,
+      "loss": 1.414,
+      "step": 863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.874805102763998e-06,
+      "loss": 1.3544,
+      "step": 864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.881644223954642e-06,
+      "loss": 1.3974,
+      "step": 865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.888483345145288e-06,
+      "loss": 1.3674,
+      "step": 866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.8953224663359325e-06,
+      "loss": 1.3244,
+      "step": 867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.902161587526577e-06,
+      "loss": 1.3709,
+      "step": 868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.909000708717222e-06,
+      "loss": 1.4984,
+      "step": 869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.915839829907867e-06,
+      "loss": 1.3566,
+      "step": 870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.922678951098511e-06,
+      "loss": 1.2979,
+      "step": 871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.929518072289157e-06,
+      "loss": 1.2436,
+      "step": 872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.936357193479802e-06,
+      "loss": 1.3635,
+      "step": 873
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.943196314670448e-06,
+      "loss": 1.5546,
+      "step": 874
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.950035435861092e-06,
+      "loss": 1.2754,
+      "step": 875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.956874557051737e-06,
+      "loss": 1.397,
+      "step": 876
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.963713678242382e-06,
+      "loss": 1.3786,
+      "step": 877
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.970552799433027e-06,
+      "loss": 1.5498,
+      "step": 878
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.977391920623671e-06,
+      "loss": 1.4366,
+      "step": 879
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.984231041814317e-06,
+      "loss": 1.2782,
+      "step": 880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9910701630049615e-06,
+      "loss": 1.3585,
+      "step": 881
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.997909284195606e-06,
+      "loss": 1.1601,
+      "step": 882
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.004748405386252e-06,
+      "loss": 1.4551,
+      "step": 883
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.0115875265768955e-06,
+      "loss": 1.6559,
+      "step": 884
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.018426647767541e-06,
+      "loss": 1.4938,
+      "step": 885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.025265768958187e-06,
+      "loss": 1.3821,
+      "step": 886
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.0321048901488304e-06,
+      "loss": 1.5307,
+      "step": 887
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.038944011339476e-06,
+      "loss": 1.3103,
+      "step": 888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.0457831325301215e-06,
+      "loss": 1.5182,
+      "step": 889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.052622253720766e-06,
+      "loss": 1.264,
+      "step": 890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.059461374911411e-06,
+      "loss": 1.4314,
+      "step": 891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.0663004961020556e-06,
+      "loss": 1.3327,
+      "step": 892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.073139617292701e-06,
+      "loss": 1.4212,
+      "step": 893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.079978738483346e-06,
+      "loss": 1.3056,
+      "step": 894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.0868178596739905e-06,
+      "loss": 1.4693,
+      "step": 895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.093656980864636e-06,
+      "loss": 1.4791,
+      "step": 896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.10049610205528e-06,
+      "loss": 1.4171,
+      "step": 897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.107335223245925e-06,
+      "loss": 1.4682,
+      "step": 898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.114174344436571e-06,
+      "loss": 1.3624,
+      "step": 899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.121013465627215e-06,
+      "loss": 1.2344,
+      "step": 900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.12785258681786e-06,
+      "loss": 1.6477,
+      "step": 901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.134691708008506e-06,
+      "loss": 1.3435,
+      "step": 902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.14153082919915e-06,
+      "loss": 1.4023,
+      "step": 903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.148369950389795e-06,
+      "loss": 1.3617,
+      "step": 904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.15520907158044e-06,
+      "loss": 1.3227,
+      "step": 905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.1620481927710845e-06,
+      "loss": 1.3602,
+      "step": 906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.168887313961729e-06,
+      "loss": 1.4213,
+      "step": 907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.175726435152375e-06,
+      "loss": 1.5043,
+      "step": 908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.18256555634302e-06,
+      "loss": 1.2789,
+      "step": 909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.189404677533664e-06,
+      "loss": 1.4996,
+      "step": 910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.19624379872431e-06,
+      "loss": 1.3946,
+      "step": 911
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.203082919914955e-06,
+      "loss": 1.3771,
+      "step": 912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.209922041105599e-06,
+      "loss": 1.4711,
+      "step": 913
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.2167611622962445e-06,
+      "loss": 1.4072,
+      "step": 914
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.223600283486889e-06,
+      "loss": 1.8436,
+      "step": 915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.230439404677534e-06,
+      "loss": 1.6291,
+      "step": 916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.2372785258681794e-06,
+      "loss": 1.6147,
+      "step": 917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.244117647058824e-06,
+      "loss": 1.3144,
+      "step": 918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.250956768249469e-06,
+      "loss": 1.3756,
+      "step": 919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.2577958894401135e-06,
+      "loss": 1.471,
+      "step": 920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.264635010630759e-06,
+      "loss": 1.3764,
+      "step": 921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.271474131821403e-06,
+      "loss": 1.3836,
+      "step": 922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.278313253012048e-06,
+      "loss": 1.2349,
+      "step": 923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.285152374202694e-06,
+      "loss": 1.4337,
+      "step": 924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.2919914953933395e-06,
+      "loss": 1.4094,
+      "step": 925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.298830616583983e-06,
+      "loss": 1.2986,
+      "step": 926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.305669737774629e-06,
+      "loss": 1.4725,
+      "step": 927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.3125088589652735e-06,
+      "loss": 1.3004,
+      "step": 928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.319347980155918e-06,
+      "loss": 1.4524,
+      "step": 929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.326187101346563e-06,
+      "loss": 1.4148,
+      "step": 930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.3330262225372084e-06,
+      "loss": 1.4444,
+      "step": 931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.339865343727853e-06,
+      "loss": 1.3111,
+      "step": 932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.346704464918498e-06,
+      "loss": 1.3837,
+      "step": 933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.353543586109143e-06,
+      "loss": 1.5378,
+      "step": 934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.360382707299787e-06,
+      "loss": 1.5412,
+      "step": 935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.367221828490433e-06,
+      "loss": 1.4228,
+      "step": 936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.374060949681078e-06,
+      "loss": 1.3675,
+      "step": 937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.380900070871722e-06,
+      "loss": 1.3888,
+      "step": 938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.387739192062368e-06,
+      "loss": 1.3815,
+      "step": 939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.394578313253013e-06,
+      "loss": 1.5788,
+      "step": 940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.401417434443657e-06,
+      "loss": 1.4399,
+      "step": 941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.4082565556343025e-06,
+      "loss": 1.4442,
+      "step": 942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.415095676824947e-06,
+      "loss": 1.3586,
+      "step": 943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.421934798015593e-06,
+      "loss": 1.4293,
+      "step": 944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.428773919206237e-06,
+      "loss": 1.5049,
+      "step": 945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.435613040396882e-06,
+      "loss": 1.3498,
+      "step": 946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.442452161587528e-06,
+      "loss": 1.2293,
+      "step": 947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.4492912827781715e-06,
+      "loss": 1.0788,
+      "step": 948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.456130403968817e-06,
+      "loss": 1.3656,
+      "step": 949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.4629695251594625e-06,
+      "loss": 1.3569,
+      "step": 950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.469808646350106e-06,
+      "loss": 1.4778,
+      "step": 951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.476647767540752e-06,
+      "loss": 1.2583,
+      "step": 952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.483486888731397e-06,
+      "loss": 1.5143,
+      "step": 953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.490326009922041e-06,
+      "loss": 1.4245,
+      "step": 954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.497165131112687e-06,
+      "loss": 1.5249,
+      "step": 955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.5040042523033315e-06,
+      "loss": 1.3733,
+      "step": 956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.510843373493976e-06,
+      "loss": 1.5298,
+      "step": 957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.517682494684621e-06,
+      "loss": 1.6669,
+      "step": 958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.524521615875266e-06,
+      "loss": 1.5165,
+      "step": 959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.531360737065911e-06,
+      "loss": 1.3748,
+      "step": 960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.538199858256556e-06,
+      "loss": 1.4714,
+      "step": 961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.545038979447201e-06,
+      "loss": 1.3802,
+      "step": 962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.551878100637847e-06,
+      "loss": 1.6009,
+      "step": 963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.558717221828491e-06,
+      "loss": 1.4523,
+      "step": 964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.565556343019136e-06,
+      "loss": 1.4421,
+      "step": 965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.572395464209781e-06,
+      "loss": 1.2568,
+      "step": 966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.5792345854004255e-06,
+      "loss": 1.3212,
+      "step": 967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.586073706591071e-06,
+      "loss": 1.3814,
+      "step": 968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.592912827781716e-06,
+      "loss": 1.2959,
+      "step": 969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.5997519489723604e-06,
+      "loss": 1.5446,
+      "step": 970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.606591070163005e-06,
+      "loss": 1.5721,
+      "step": 971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.613430191353651e-06,
+      "loss": 1.5038,
+      "step": 972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.620269312544295e-06,
+      "loss": 1.2834,
+      "step": 973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.62710843373494e-06,
+      "loss": 1.3057,
+      "step": 974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.6339475549255856e-06,
+      "loss": 1.4996,
+      "step": 975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.640786676116229e-06,
+      "loss": 1.4962,
+      "step": 976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.647625797306875e-06,
+      "loss": 1.2428,
+      "step": 977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.6544649184975205e-06,
+      "loss": 1.662,
+      "step": 978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.661304039688165e-06,
+      "loss": 1.5698,
+      "step": 979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.66814316087881e-06,
+      "loss": 1.3461,
+      "step": 980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.674982282069455e-06,
+      "loss": 1.6628,
+      "step": 981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.6818214032601e-06,
+      "loss": 1.2681,
+      "step": 982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.688660524450745e-06,
+      "loss": 1.4383,
+      "step": 983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.695499645641389e-06,
+      "loss": 1.4515,
+      "step": 984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.702338766832035e-06,
+      "loss": 1.4903,
+      "step": 985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.709177888022679e-06,
+      "loss": 1.3754,
+      "step": 986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.716017009213324e-06,
+      "loss": 1.2739,
+      "step": 987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.72285613040397e-06,
+      "loss": 1.396,
+      "step": 988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.729695251594614e-06,
+      "loss": 1.2713,
+      "step": 989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.736534372785259e-06,
+      "loss": 1.5057,
+      "step": 990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.743373493975905e-06,
+      "loss": 1.4431,
+      "step": 991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.750212615166549e-06,
+      "loss": 1.6582,
+      "step": 992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.757051736357194e-06,
+      "loss": 1.4198,
+      "step": 993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.763890857547839e-06,
+      "loss": 1.5554,
+      "step": 994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.7707299787384835e-06,
+      "loss": 1.3,
+      "step": 995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.777569099929129e-06,
+      "loss": 1.5008,
+      "step": 996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.784408221119774e-06,
+      "loss": 1.3194,
+      "step": 997
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.791247342310419e-06,
+      "loss": 1.268,
+      "step": 998
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.798086463501063e-06,
+      "loss": 1.4928,
+      "step": 999
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.804925584691709e-06,
+      "loss": 1.4934,
+      "step": 1000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.811764705882354e-06,
+      "loss": 1.4903,
+      "step": 1001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.818603827072998e-06,
+      "loss": 1.5324,
+      "step": 1002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.8254429482636435e-06,
+      "loss": 1.4662,
+      "step": 1003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.832282069454289e-06,
+      "loss": 1.3841,
+      "step": 1004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.839121190644933e-06,
+      "loss": 1.4389,
+      "step": 1005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.845960311835578e-06,
+      "loss": 1.5431,
+      "step": 1006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.852799433026223e-06,
+      "loss": 1.4936,
+      "step": 1007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.859638554216868e-06,
+      "loss": 1.5375,
+      "step": 1008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.8664776754075125e-06,
+      "loss": 1.3961,
+      "step": 1009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.873316796598158e-06,
+      "loss": 1.552,
+      "step": 1010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.880155917788803e-06,
+      "loss": 1.4659,
+      "step": 1011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.886995038979447e-06,
+      "loss": 1.2046,
+      "step": 1012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.893834160170093e-06,
+      "loss": 1.5444,
+      "step": 1013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.900673281360737e-06,
+      "loss": 1.206,
+      "step": 1014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.907512402551382e-06,
+      "loss": 1.466,
+      "step": 1015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.914351523742028e-06,
+      "loss": 1.5057,
+      "step": 1016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.9211906449326725e-06,
+      "loss": 1.4925,
+      "step": 1017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.928029766123317e-06,
+      "loss": 1.1947,
+      "step": 1018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.934868887313963e-06,
+      "loss": 1.3747,
+      "step": 1019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.941708008504607e-06,
+      "loss": 1.4471,
+      "step": 1020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.948547129695252e-06,
+      "loss": 1.2134,
+      "step": 1021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.955386250885897e-06,
+      "loss": 1.2402,
+      "step": 1022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.962225372076542e-06,
+      "loss": 1.2341,
+      "step": 1023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.969064493267187e-06,
+      "loss": 1.3572,
+      "step": 1024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.975903614457832e-06,
+      "loss": 1.3698,
+      "step": 1025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.982742735648477e-06,
+      "loss": 1.3429,
+      "step": 1026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.989581856839121e-06,
+      "loss": 1.563,
+      "step": 1027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.9964209780297666e-06,
+      "loss": 1.397,
+      "step": 1028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.003260099220412e-06,
+      "loss": 1.4778,
+      "step": 1029
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.010099220411056e-06,
+      "loss": 1.455,
+      "step": 1030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.0169383416017015e-06,
+      "loss": 1.5583,
+      "step": 1031
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.023777462792347e-06,
+      "loss": 1.2699,
+      "step": 1032
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.030616583982992e-06,
+      "loss": 1.4381,
+      "step": 1033
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.037455705173636e-06,
+      "loss": 1.5558,
+      "step": 1034
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.044294826364281e-06,
+      "loss": 1.2357,
+      "step": 1035
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.0511339475549266e-06,
+      "loss": 1.2535,
+      "step": 1036
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.05797306874557e-06,
+      "loss": 1.4076,
+      "step": 1037
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.064812189936216e-06,
+      "loss": 1.3878,
+      "step": 1038
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.0716513111268615e-06,
+      "loss": 1.5936,
+      "step": 1039
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.078490432317505e-06,
+      "loss": 1.3765,
+      "step": 1040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.085329553508151e-06,
+      "loss": 1.4701,
+      "step": 1041
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.092168674698796e-06,
+      "loss": 1.4722,
+      "step": 1042
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.09900779588944e-06,
+      "loss": 1.4084,
+      "step": 1043
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.105846917080086e-06,
+      "loss": 1.6563,
+      "step": 1044
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.1126860382707304e-06,
+      "loss": 1.5007,
+      "step": 1045
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.119525159461375e-06,
+      "loss": 1.2659,
+      "step": 1046
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.126364280652021e-06,
+      "loss": 1.3742,
+      "step": 1047
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.133203401842665e-06,
+      "loss": 1.4699,
+      "step": 1048
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.14004252303331e-06,
+      "loss": 1.3159,
+      "step": 1049
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.146881644223955e-06,
+      "loss": 1.4396,
+      "step": 1050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.1537207654146e-06,
+      "loss": 1.3686,
+      "step": 1051
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.160559886605246e-06,
+      "loss": 1.1869,
+      "step": 1052
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.16739900779589e-06,
+      "loss": 1.1517,
+      "step": 1053
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.174238128986535e-06,
+      "loss": 1.2439,
+      "step": 1054
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.181077250177181e-06,
+      "loss": 1.478,
+      "step": 1055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.1879163713678245e-06,
+      "loss": 1.5165,
+      "step": 1056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.19475549255847e-06,
+      "loss": 1.3978,
+      "step": 1057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.201594613749115e-06,
+      "loss": 1.2541,
+      "step": 1058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.208433734939759e-06,
+      "loss": 1.4703,
+      "step": 1059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.215272856130405e-06,
+      "loss": 1.4346,
+      "step": 1060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.22211197732105e-06,
+      "loss": 1.6357,
+      "step": 1061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.228951098511694e-06,
+      "loss": 1.2759,
+      "step": 1062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.235790219702339e-06,
+      "loss": 1.4618,
+      "step": 1063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.2426293408929845e-06,
+      "loss": 1.3134,
+      "step": 1064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.249468462083628e-06,
+      "loss": 1.5325,
+      "step": 1065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.256307583274274e-06,
+      "loss": 1.4045,
+      "step": 1066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.2631467044649194e-06,
+      "loss": 1.5951,
+      "step": 1067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.269985825655565e-06,
+      "loss": 1.4283,
+      "step": 1068
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.276824946846209e-06,
+      "loss": 1.4858,
+      "step": 1069
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.283664068036854e-06,
+      "loss": 1.4861,
+      "step": 1070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.290503189227499e-06,
+      "loss": 1.2685,
+      "step": 1071
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.297342310418144e-06,
+      "loss": 1.0321,
+      "step": 1072
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.304181431608788e-06,
+      "loss": 1.1699,
+      "step": 1073
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.311020552799434e-06,
+      "loss": 1.2928,
+      "step": 1074
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.317859673990079e-06,
+      "loss": 1.4338,
+      "step": 1075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.324698795180723e-06,
+      "loss": 1.3391,
+      "step": 1076
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.331537916371369e-06,
+      "loss": 1.3727,
+      "step": 1077
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.338377037562013e-06,
+      "loss": 1.4142,
+      "step": 1078
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.345216158752658e-06,
+      "loss": 1.3999,
+      "step": 1079
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.352055279943304e-06,
+      "loss": 1.335,
+      "step": 1080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3588944011339476e-06,
+      "loss": 1.3696,
+      "step": 1081
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.365733522324593e-06,
+      "loss": 1.2417,
+      "step": 1082
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.372572643515239e-06,
+      "loss": 1.4803,
+      "step": 1083
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3794117647058825e-06,
+      "loss": 1.5263,
+      "step": 1084
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.386250885896528e-06,
+      "loss": 1.388,
+      "step": 1085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.393090007087173e-06,
+      "loss": 1.5678,
+      "step": 1086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.399929128277818e-06,
+      "loss": 1.388,
+      "step": 1087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.406768249468462e-06,
+      "loss": 1.2865,
+      "step": 1088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.4136073706591076e-06,
+      "loss": 1.496,
+      "step": 1089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.420446491849753e-06,
+      "loss": 1.5241,
+      "step": 1090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.427285613040397e-06,
+      "loss": 1.5177,
+      "step": 1091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.4341247342310425e-06,
+      "loss": 1.4494,
+      "step": 1092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.440963855421688e-06,
+      "loss": 1.3572,
+      "step": 1093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.447802976612332e-06,
+      "loss": 1.5573,
+      "step": 1094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.454642097802977e-06,
+      "loss": 1.3783,
+      "step": 1095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.461481218993622e-06,
+      "loss": 1.3816,
+      "step": 1096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.468320340184267e-06,
+      "loss": 1.4491,
+      "step": 1097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.475159461374912e-06,
+      "loss": 1.4559,
+      "step": 1098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.481998582565557e-06,
+      "loss": 1.5222,
+      "step": 1099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.488837703756202e-06,
+      "loss": 1.4003,
+      "step": 1100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.495676824946846e-06,
+      "loss": 1.1939,
+      "step": 1101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.502515946137492e-06,
+      "loss": 1.4063,
+      "step": 1102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.5093550673281365e-06,
+      "loss": 1.1073,
+      "step": 1103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.516194188518781e-06,
+      "loss": 1.1469,
+      "step": 1104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.523033309709427e-06,
+      "loss": 1.2487,
+      "step": 1105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.529872430900072e-06,
+      "loss": 1.4487,
+      "step": 1106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.536711552090716e-06,
+      "loss": 1.2332,
+      "step": 1107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.543550673281362e-06,
+      "loss": 1.3484,
+      "step": 1108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.550389794472006e-06,
+      "loss": 1.5135,
+      "step": 1109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.557228915662651e-06,
+      "loss": 1.268,
+      "step": 1110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.5640680368532966e-06,
+      "loss": 1.2971,
+      "step": 1111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.570907158043941e-06,
+      "loss": 1.4493,
+      "step": 1112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.577746279234586e-06,
+      "loss": 1.4672,
+      "step": 1113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.584585400425231e-06,
+      "loss": 1.5237,
+      "step": 1114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.591424521615876e-06,
+      "loss": 1.6687,
+      "step": 1115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.59826364280652e-06,
+      "loss": 1.3356,
+      "step": 1116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.6051027639971655e-06,
+      "loss": 1.1509,
+      "step": 1117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.611941885187811e-06,
+      "loss": 1.6902,
+      "step": 1118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.618781006378455e-06,
+      "loss": 1.3932,
+      "step": 1119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.6256201275691e-06,
+      "loss": 1.4753,
+      "step": 1120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.632459248759746e-06,
+      "loss": 1.3868,
+      "step": 1121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.63929836995039e-06,
+      "loss": 1.3127,
+      "step": 1122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.646137491141035e-06,
+      "loss": 1.5383,
+      "step": 1123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.65297661233168e-06,
+      "loss": 1.5338,
+      "step": 1124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.659815733522326e-06,
+      "loss": 1.5072,
+      "step": 1125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.66665485471297e-06,
+      "loss": 1.3266,
+      "step": 1126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.673493975903616e-06,
+      "loss": 1.4938,
+      "step": 1127
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.68033309709426e-06,
+      "loss": 1.224,
+      "step": 1128
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.687172218284905e-06,
+      "loss": 1.7013,
+      "step": 1129
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.69401133947555e-06,
+      "loss": 1.4497,
+      "step": 1130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.700850460666194e-06,
+      "loss": 1.4636,
+      "step": 1131
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.70768958185684e-06,
+      "loss": 1.5673,
+      "step": 1132
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.714528703047484e-06,
+      "loss": 1.4554,
+      "step": 1133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.72136782423813e-06,
+      "loss": 1.3966,
+      "step": 1134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.728206945428775e-06,
+      "loss": 1.3031,
+      "step": 1135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.73504606661942e-06,
+      "loss": 1.6264,
+      "step": 1136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.741885187810064e-06,
+      "loss": 1.1104,
+      "step": 1137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.748724309000709e-06,
+      "loss": 1.4016,
+      "step": 1138
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.755563430191354e-06,
+      "loss": 1.3646,
+      "step": 1139
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.762402551382e-06,
+      "loss": 1.4676,
+      "step": 1140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.769241672572645e-06,
+      "loss": 1.5394,
+      "step": 1141
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.77608079376329e-06,
+      "loss": 1.3554,
+      "step": 1142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.782919914953934e-06,
+      "loss": 1.5256,
+      "step": 1143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.789759036144579e-06,
+      "loss": 1.4154,
+      "step": 1144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.796598157335223e-06,
+      "loss": 1.1885,
+      "step": 1145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.803437278525868e-06,
+      "loss": 1.5355,
+      "step": 1146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.810276399716515e-06,
+      "loss": 1.5502,
+      "step": 1147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.817115520907158e-06,
+      "loss": 1.2375,
+      "step": 1148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.823954642097804e-06,
+      "loss": 1.61,
+      "step": 1149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.830793763288449e-06,
+      "loss": 1.3781,
+      "step": 1150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.837632884479093e-06,
+      "loss": 1.436,
+      "step": 1151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.844472005669738e-06,
+      "loss": 1.3594,
+      "step": 1152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.851311126860384e-06,
+      "loss": 1.465,
+      "step": 1153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.858150248051027e-06,
+      "loss": 1.3648,
+      "step": 1154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.864989369241674e-06,
+      "loss": 1.4319,
+      "step": 1155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.871828490432318e-06,
+      "loss": 1.2989,
+      "step": 1156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.878667611622963e-06,
+      "loss": 1.4862,
+      "step": 1157
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.885506732813608e-06,
+      "loss": 1.3556,
+      "step": 1158
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.892345854004252e-06,
+      "loss": 1.5391,
+      "step": 1159
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.899184975194899e-06,
+      "loss": 1.4403,
+      "step": 1160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.906024096385542e-06,
+      "loss": 1.5372,
+      "step": 1161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.912863217576188e-06,
+      "loss": 1.3268,
+      "step": 1162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.919702338766833e-06,
+      "loss": 1.48,
+      "step": 1163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.926541459957478e-06,
+      "loss": 1.595,
+      "step": 1164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.933380581148122e-06,
+      "loss": 1.3374,
+      "step": 1165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.940219702338769e-06,
+      "loss": 1.3499,
+      "step": 1166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.947058823529412e-06,
+      "loss": 1.6332,
+      "step": 1167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.953897944720058e-06,
+      "loss": 1.2295,
+      "step": 1168
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.960737065910703e-06,
+      "loss": 1.4402,
+      "step": 1169
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.967576187101347e-06,
+      "loss": 1.4585,
+      "step": 1170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.974415308291992e-06,
+      "loss": 1.5351,
+      "step": 1171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.981254429482637e-06,
+      "loss": 1.293,
+      "step": 1172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.988093550673281e-06,
+      "loss": 1.4271,
+      "step": 1173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.994932671863926e-06,
+      "loss": 1.4713,
+      "step": 1174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.001771793054572e-06,
+      "loss": 1.3483,
+      "step": 1175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.008610914245217e-06,
+      "loss": 1.3812,
+      "step": 1176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.015450035435862e-06,
+      "loss": 1.5094,
+      "step": 1177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.022289156626507e-06,
+      "loss": 1.2921,
+      "step": 1178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.029128277817151e-06,
+      "loss": 1.4632,
+      "step": 1179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.035967399007796e-06,
+      "loss": 1.2707,
+      "step": 1180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.042806520198442e-06,
+      "loss": 1.4512,
+      "step": 1181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.049645641389087e-06,
+      "loss": 1.2353,
+      "step": 1182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.056484762579732e-06,
+      "loss": 1.1912,
+      "step": 1183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.063323883770376e-06,
+      "loss": 1.5899,
+      "step": 1184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.070163004961021e-06,
+      "loss": 1.4538,
+      "step": 1185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.077002126151666e-06,
+      "loss": 1.4461,
+      "step": 1186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.08384124734231e-06,
+      "loss": 1.6826,
+      "step": 1187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.090680368532957e-06,
+      "loss": 1.4766,
+      "step": 1188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.0975194897236e-06,
+      "loss": 1.3884,
+      "step": 1189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.104358610914246e-06,
+      "loss": 1.2889,
+      "step": 1190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.11119773210489e-06,
+      "loss": 1.5564,
+      "step": 1191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.118036853295536e-06,
+      "loss": 1.5121,
+      "step": 1192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.12487597448618e-06,
+      "loss": 1.4409,
+      "step": 1193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.131715095676827e-06,
+      "loss": 1.2823,
+      "step": 1194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.138554216867471e-06,
+      "loss": 1.4492,
+      "step": 1195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.145393338058116e-06,
+      "loss": 1.6502,
+      "step": 1196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.15223245924876e-06,
+      "loss": 1.6354,
+      "step": 1197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.159071580439405e-06,
+      "loss": 1.3516,
+      "step": 1198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.16591070163005e-06,
+      "loss": 1.2582,
+      "step": 1199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.172749822820695e-06,
+      "loss": 1.3733,
+      "step": 1200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.179588944011341e-06,
+      "loss": 1.4276,
+      "step": 1201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.186428065201984e-06,
+      "loss": 1.2028,
+      "step": 1202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.19326718639263e-06,
+      "loss": 1.228,
+      "step": 1203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.200106307583275e-06,
+      "loss": 1.3071,
+      "step": 1204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.20694542877392e-06,
+      "loss": 1.4397,
+      "step": 1205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.213784549964564e-06,
+      "loss": 1.4623,
+      "step": 1206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.22062367115521e-06,
+      "loss": 1.4585,
+      "step": 1207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.227462792345854e-06,
+      "loss": 1.4288,
+      "step": 1208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.2343019135365e-06,
+      "loss": 1.2757,
+      "step": 1209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.241141034727145e-06,
+      "loss": 1.4409,
+      "step": 1210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.24798015591779e-06,
+      "loss": 1.549,
+      "step": 1211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.254819277108434e-06,
+      "loss": 1.3236,
+      "step": 1212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.261658398299079e-06,
+      "loss": 1.4549,
+      "step": 1213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.268497519489725e-06,
+      "loss": 1.2961,
+      "step": 1214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.275336640680368e-06,
+      "loss": 1.2405,
+      "step": 1215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.282175761871015e-06,
+      "loss": 1.4646,
+      "step": 1216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.28901488306166e-06,
+      "loss": 1.3793,
+      "step": 1217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.295854004252304e-06,
+      "loss": 1.3433,
+      "step": 1218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.302693125442949e-06,
+      "loss": 1.396,
+      "step": 1219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.309532246633593e-06,
+      "loss": 1.5204,
+      "step": 1220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.316371367824238e-06,
+      "loss": 1.3953,
+      "step": 1221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.323210489014885e-06,
+      "loss": 1.5722,
+      "step": 1222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.33004961020553e-06,
+      "loss": 1.1954,
+      "step": 1223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.336888731396174e-06,
+      "loss": 1.2732,
+      "step": 1224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.343727852586819e-06,
+      "loss": 1.458,
+      "step": 1225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.350566973777463e-06,
+      "loss": 1.4931,
+      "step": 1226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.357406094968108e-06,
+      "loss": 1.3502,
+      "step": 1227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.364245216158753e-06,
+      "loss": 1.5591,
+      "step": 1228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.371084337349399e-06,
+      "loss": 1.6048,
+      "step": 1229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.377923458540044e-06,
+      "loss": 1.4332,
+      "step": 1230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.384762579730688e-06,
+      "loss": 1.3309,
+      "step": 1231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.391601700921333e-06,
+      "loss": 1.4314,
+      "step": 1232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.398440822111978e-06,
+      "loss": 1.3339,
+      "step": 1233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.405279943302622e-06,
+      "loss": 1.3,
+      "step": 1234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.412119064493267e-06,
+      "loss": 1.29,
+      "step": 1235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.418958185683913e-06,
+      "loss": 1.3646,
+      "step": 1236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.425797306874558e-06,
+      "loss": 1.4977,
+      "step": 1237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.432636428065203e-06,
+      "loss": 1.3172,
+      "step": 1238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.439475549255848e-06,
+      "loss": 1.5652,
+      "step": 1239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.446314670446492e-06,
+      "loss": 1.3143,
+      "step": 1240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.453153791637137e-06,
+      "loss": 1.2315,
+      "step": 1241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.459992912827783e-06,
+      "loss": 1.6137,
+      "step": 1242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.466832034018426e-06,
+      "loss": 1.3443,
+      "step": 1243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.473671155209073e-06,
+      "loss": 1.2444,
+      "step": 1244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.480510276399717e-06,
+      "loss": 1.3619,
+      "step": 1245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.487349397590362e-06,
+      "loss": 1.4227,
+      "step": 1246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.494188518781007e-06,
+      "loss": 1.3198,
+      "step": 1247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.501027639971651e-06,
+      "loss": 1.4404,
+      "step": 1248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.507866761162298e-06,
+      "loss": 1.3438,
+      "step": 1249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.51470588235294e-06,
+      "loss": 1.3333,
+      "step": 1250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.521545003543587e-06,
+      "loss": 1.5367,
+      "step": 1251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.528384124734232e-06,
+      "loss": 1.5021,
+      "step": 1252
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.535223245924877e-06,
+      "loss": 1.3428,
+      "step": 1253
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.542062367115521e-06,
+      "loss": 1.3096,
+      "step": 1254
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.548901488306168e-06,
+      "loss": 1.3505,
+      "step": 1255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.55574060949681e-06,
+      "loss": 1.3771,
+      "step": 1256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.562579730687457e-06,
+      "loss": 1.4177,
+      "step": 1257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.569418851878102e-06,
+      "loss": 1.5589,
+      "step": 1258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.576257973068746e-06,
+      "loss": 1.4217,
+      "step": 1259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.583097094259391e-06,
+      "loss": 1.3281,
+      "step": 1260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.589936215450036e-06,
+      "loss": 1.1818,
+      "step": 1261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.59677533664068e-06,
+      "loss": 1.3597,
+      "step": 1262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.603614457831325e-06,
+      "loss": 1.4628,
+      "step": 1263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.610453579021971e-06,
+      "loss": 1.3882,
+      "step": 1264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.617292700212616e-06,
+      "loss": 1.4083,
+      "step": 1265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.62413182140326e-06,
+      "loss": 1.4814,
+      "step": 1266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.630970942593905e-06,
+      "loss": 1.4496,
+      "step": 1267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.637810063784552e-06,
+      "loss": 1.3413,
+      "step": 1268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.644649184975195e-06,
+      "loss": 1.3545,
+      "step": 1269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.651488306165841e-06,
+      "loss": 1.3598,
+      "step": 1270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.658327427356486e-06,
+      "loss": 1.2456,
+      "step": 1271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.66516654854713e-06,
+      "loss": 1.3662,
+      "step": 1272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.672005669737775e-06,
+      "loss": 1.3409,
+      "step": 1273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.67884479092842e-06,
+      "loss": 1.4225,
+      "step": 1274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.685683912119065e-06,
+      "loss": 1.4778,
+      "step": 1275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.69252303330971e-06,
+      "loss": 1.4259,
+      "step": 1276
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.699362154500356e-06,
+      "loss": 1.4151,
+      "step": 1277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.706201275690999e-06,
+      "loss": 1.2993,
+      "step": 1278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.713040396881645e-06,
+      "loss": 1.3833,
+      "step": 1279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.71987951807229e-06,
+      "loss": 1.3049,
+      "step": 1280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.726718639262934e-06,
+      "loss": 1.188,
+      "step": 1281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.733557760453579e-06,
+      "loss": 1.379,
+      "step": 1282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.740396881644226e-06,
+      "loss": 1.3109,
+      "step": 1283
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.74723600283487e-06,
+      "loss": 1.3724,
+      "step": 1284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.754075124025515e-06,
+      "loss": 1.2501,
+      "step": 1285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.76091424521616e-06,
+      "loss": 1.3047,
+      "step": 1286
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.767753366406804e-06,
+      "loss": 1.5306,
+      "step": 1287
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.774592487597449e-06,
+      "loss": 1.1759,
+      "step": 1288
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.781431608788094e-06,
+      "loss": 1.4016,
+      "step": 1289
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.78827072997874e-06,
+      "loss": 1.3934,
+      "step": 1290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.795109851169383e-06,
+      "loss": 1.2835,
+      "step": 1291
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.80194897236003e-06,
+      "loss": 1.5794,
+      "step": 1292
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.808788093550674e-06,
+      "loss": 1.4908,
+      "step": 1293
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.815627214741319e-06,
+      "loss": 1.4501,
+      "step": 1294
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.822466335931963e-06,
+      "loss": 1.4617,
+      "step": 1295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.82930545712261e-06,
+      "loss": 1.3126,
+      "step": 1296
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.836144578313253e-06,
+      "loss": 1.4814,
+      "step": 1297
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.8429836995039e-06,
+      "loss": 1.3619,
+      "step": 1298
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.849822820694544e-06,
+      "loss": 1.3431,
+      "step": 1299
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.856661941885189e-06,
+      "loss": 1.4085,
+      "step": 1300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.863501063075833e-06,
+      "loss": 1.4013,
+      "step": 1301
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.870340184266478e-06,
+      "loss": 1.4121,
+      "step": 1302
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.877179305457124e-06,
+      "loss": 1.4684,
+      "step": 1303
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.884018426647767e-06,
+      "loss": 1.3522,
+      "step": 1304
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.890857547838414e-06,
+      "loss": 1.5432,
+      "step": 1305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.897696669029058e-06,
+      "loss": 1.4305,
+      "step": 1306
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.904535790219703e-06,
+      "loss": 1.3962,
+      "step": 1307
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.911374911410348e-06,
+      "loss": 1.2972,
+      "step": 1308
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.918214032600994e-06,
+      "loss": 1.2031,
+      "step": 1309
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.925053153791637e-06,
+      "loss": 1.397,
+      "step": 1310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.931892274982283e-06,
+      "loss": 1.4166,
+      "step": 1311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.938731396172928e-06,
+      "loss": 1.4304,
+      "step": 1312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.945570517363573e-06,
+      "loss": 1.4234,
+      "step": 1313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.952409638554218e-06,
+      "loss": 1.1726,
+      "step": 1314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.959248759744862e-06,
+      "loss": 1.5927,
+      "step": 1315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.966087880935507e-06,
+      "loss": 1.1889,
+      "step": 1316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.972927002126152e-06,
+      "loss": 1.3355,
+      "step": 1317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.979766123316798e-06,
+      "loss": 1.3981,
+      "step": 1318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.986605244507443e-06,
+      "loss": 1.3046,
+      "step": 1319
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.993444365698087e-06,
+      "loss": 1.3799,
+      "step": 1320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.000283486888732e-06,
+      "loss": 1.607,
+      "step": 1321
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.007122608079377e-06,
+      "loss": 1.4027,
+      "step": 1322
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.013961729270021e-06,
+      "loss": 1.5786,
+      "step": 1323
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.020800850460668e-06,
+      "loss": 1.3394,
+      "step": 1324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.027639971651312e-06,
+      "loss": 1.3674,
+      "step": 1325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.034479092841957e-06,
+      "loss": 1.3337,
+      "step": 1326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.041318214032602e-06,
+      "loss": 1.5311,
+      "step": 1327
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.048157335223247e-06,
+      "loss": 1.2767,
+      "step": 1328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.054996456413891e-06,
+      "loss": 1.3614,
+      "step": 1329
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.061835577604536e-06,
+      "loss": 1.3969,
+      "step": 1330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.068674698795182e-06,
+      "loss": 1.3295,
+      "step": 1331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.075513819985825e-06,
+      "loss": 1.3579,
+      "step": 1332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.082352941176472e-06,
+      "loss": 1.316,
+      "step": 1333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.089192062367116e-06,
+      "loss": 1.4583,
+      "step": 1334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.096031183557761e-06,
+      "loss": 1.16,
+      "step": 1335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.102870304748406e-06,
+      "loss": 1.3143,
+      "step": 1336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.10970942593905e-06,
+      "loss": 1.6445,
+      "step": 1337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.116548547129697e-06,
+      "loss": 1.1482,
+      "step": 1338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.123387668320341e-06,
+      "loss": 1.4833,
+      "step": 1339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.130226789510986e-06,
+      "loss": 1.3816,
+      "step": 1340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.13706591070163e-06,
+      "loss": 1.2887,
+      "step": 1341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.143905031892275e-06,
+      "loss": 1.2734,
+      "step": 1342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.15074415308292e-06,
+      "loss": 1.3582,
+      "step": 1343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.157583274273567e-06,
+      "loss": 1.346,
+      "step": 1344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.16442239546421e-06,
+      "loss": 1.6533,
+      "step": 1345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.171261516654856e-06,
+      "loss": 1.5231,
+      "step": 1346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.1781006378455e-06,
+      "loss": 1.5618,
+      "step": 1347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.184939759036145e-06,
+      "loss": 1.1859,
+      "step": 1348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.19177888022679e-06,
+      "loss": 1.2894,
+      "step": 1349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.198618001417435e-06,
+      "loss": 1.6841,
+      "step": 1350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.20545712260808e-06,
+      "loss": 1.1777,
+      "step": 1351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.212296243798726e-06,
+      "loss": 1.3333,
+      "step": 1352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.21913536498937e-06,
+      "loss": 1.4408,
+      "step": 1353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.225974486180015e-06,
+      "loss": 1.5927,
+      "step": 1354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.23281360737066e-06,
+      "loss": 1.3881,
+      "step": 1355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.239652728561304e-06,
+      "loss": 1.3953,
+      "step": 1356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.24649184975195e-06,
+      "loss": 1.2752,
+      "step": 1357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.253330970942594e-06,
+      "loss": 1.3727,
+      "step": 1358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.26017009213324e-06,
+      "loss": 1.3249,
+      "step": 1359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.267009213323885e-06,
+      "loss": 1.3883,
+      "step": 1360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.27384833451453e-06,
+      "loss": 1.3074,
+      "step": 1361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.280687455705174e-06,
+      "loss": 1.2646,
+      "step": 1362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.287526576895819e-06,
+      "loss": 1.3814,
+      "step": 1363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.294365698086464e-06,
+      "loss": 1.3101,
+      "step": 1364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.301204819277108e-06,
+      "loss": 1.683,
+      "step": 1365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.308043940467755e-06,
+      "loss": 1.1766,
+      "step": 1366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.3148830616584e-06,
+      "loss": 1.2266,
+      "step": 1367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.321722182849044e-06,
+      "loss": 1.4606,
+      "step": 1368
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.328561304039689e-06,
+      "loss": 1.5355,
+      "step": 1369
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.335400425230333e-06,
+      "loss": 1.4505,
+      "step": 1370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.342239546420978e-06,
+      "loss": 1.4566,
+      "step": 1371
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.349078667611624e-06,
+      "loss": 1.4431,
+      "step": 1372
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.35591778880227e-06,
+      "loss": 1.5272,
+      "step": 1373
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.362756909992914e-06,
+      "loss": 1.4545,
+      "step": 1374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.369596031183559e-06,
+      "loss": 1.3254,
+      "step": 1375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.376435152374203e-06,
+      "loss": 1.4969,
+      "step": 1376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.383274273564848e-06,
+      "loss": 1.2233,
+      "step": 1377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.390113394755493e-06,
+      "loss": 1.5018,
+      "step": 1378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.396952515946139e-06,
+      "loss": 1.4999,
+      "step": 1379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.403791637136784e-06,
+      "loss": 1.5745,
+      "step": 1380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.410630758327428e-06,
+      "loss": 1.3771,
+      "step": 1381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.417469879518073e-06,
+      "loss": 1.5133,
+      "step": 1382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.424309000708718e-06,
+      "loss": 1.355,
+      "step": 1383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.431148121899362e-06,
+      "loss": 1.5467,
+      "step": 1384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.437987243090009e-06,
+      "loss": 1.5154,
+      "step": 1385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.444826364280652e-06,
+      "loss": 1.5566,
+      "step": 1386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.451665485471298e-06,
+      "loss": 1.3082,
+      "step": 1387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.458504606661943e-06,
+      "loss": 1.3023,
+      "step": 1388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.465343727852588e-06,
+      "loss": 1.3594,
+      "step": 1389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.472182849043232e-06,
+      "loss": 1.669,
+      "step": 1390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.479021970233877e-06,
+      "loss": 1.4847,
+      "step": 1391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.485861091424523e-06,
+      "loss": 1.489,
+      "step": 1392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.492700212615166e-06,
+      "loss": 1.3994,
+      "step": 1393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.499539333805813e-06,
+      "loss": 1.408,
+      "step": 1394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.506378454996457e-06,
+      "loss": 1.6402,
+      "step": 1395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.513217576187102e-06,
+      "loss": 1.4489,
+      "step": 1396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.520056697377747e-06,
+      "loss": 1.5158,
+      "step": 1397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.526895818568393e-06,
+      "loss": 1.3697,
+      "step": 1398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.533734939759036e-06,
+      "loss": 1.2377,
+      "step": 1399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.540574060949682e-06,
+      "loss": 1.4568,
+      "step": 1400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.547413182140327e-06,
+      "loss": 1.3931,
+      "step": 1401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.554252303330972e-06,
+      "loss": 1.432,
+      "step": 1402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.561091424521616e-06,
+      "loss": 1.2266,
+      "step": 1403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.567930545712261e-06,
+      "loss": 1.6791,
+      "step": 1404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.574769666902906e-06,
+      "loss": 1.4856,
+      "step": 1405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.58160878809355e-06,
+      "loss": 1.2692,
+      "step": 1406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.588447909284197e-06,
+      "loss": 1.3335,
+      "step": 1407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.595287030474842e-06,
+      "loss": 1.3938,
+      "step": 1408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.602126151665486e-06,
+      "loss": 1.3975,
+      "step": 1409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.608965272856131e-06,
+      "loss": 1.3538,
+      "step": 1410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.615804394046777e-06,
+      "loss": 1.3637,
+      "step": 1411
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.62264351523742e-06,
+      "loss": 1.5683,
+      "step": 1412
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.629482636428067e-06,
+      "loss": 1.4706,
+      "step": 1413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.636321757618711e-06,
+      "loss": 1.2678,
+      "step": 1414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.643160878809356e-06,
+      "loss": 1.2967,
+      "step": 1415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.65e-06,
+      "loss": 1.2649,
+      "step": 1416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.64923979832992e-06,
+      "loss": 1.3737,
+      "step": 1417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.64847959665984e-06,
+      "loss": 1.5335,
+      "step": 1418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.647719394989759e-06,
+      "loss": 1.4782,
+      "step": 1419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.646959193319679e-06,
+      "loss": 1.5224,
+      "step": 1420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.646198991649598e-06,
+      "loss": 1.4176,
+      "step": 1421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.645438789979518e-06,
+      "loss": 1.3995,
+      "step": 1422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.644678588309438e-06,
+      "loss": 1.4308,
+      "step": 1423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.643918386639358e-06,
+      "loss": 1.5546,
+      "step": 1424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.643158184969278e-06,
+      "loss": 1.496,
+      "step": 1425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.642397983299198e-06,
+      "loss": 1.3678,
+      "step": 1426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.641637781629118e-06,
+      "loss": 1.4639,
+      "step": 1427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.640877579959037e-06,
+      "loss": 1.4144,
+      "step": 1428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.640117378288957e-06,
+      "loss": 1.5198,
+      "step": 1429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.639357176618875e-06,
+      "loss": 1.3861,
+      "step": 1430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.638596974948795e-06,
+      "loss": 1.5638,
+      "step": 1431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.637836773278715e-06,
+      "loss": 1.1789,
+      "step": 1432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.637076571608635e-06,
+      "loss": 1.5702,
+      "step": 1433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.636316369938555e-06,
+      "loss": 1.4639,
+      "step": 1434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.635556168268473e-06,
+      "loss": 1.6509,
+      "step": 1435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.634795966598393e-06,
+      "loss": 1.4595,
+      "step": 1436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.634035764928313e-06,
+      "loss": 1.4494,
+      "step": 1437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.633275563258233e-06,
+      "loss": 1.4794,
+      "step": 1438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.632515361588152e-06,
+      "loss": 1.4242,
+      "step": 1439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.631755159918072e-06,
+      "loss": 1.2877,
+      "step": 1440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.630994958247992e-06,
+      "loss": 1.4095,
+      "step": 1441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.630234756577912e-06,
+      "loss": 1.4501,
+      "step": 1442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.629474554907832e-06,
+      "loss": 1.4788,
+      "step": 1443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.628714353237752e-06,
+      "loss": 1.3864,
+      "step": 1444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.627954151567672e-06,
+      "loss": 1.2877,
+      "step": 1445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.62719394989759e-06,
+      "loss": 1.3149,
+      "step": 1446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.62643374822751e-06,
+      "loss": 1.4075,
+      "step": 1447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.62567354655743e-06,
+      "loss": 1.3126,
+      "step": 1448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.62491334488735e-06,
+      "loss": 1.5238,
+      "step": 1449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.624153143217269e-06,
+      "loss": 1.4871,
+      "step": 1450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.623392941547187e-06,
+      "loss": 1.2435,
+      "step": 1451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.622632739877107e-06,
+      "loss": 1.2739,
+      "step": 1452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.621872538207027e-06,
+      "loss": 1.3998,
+      "step": 1453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.621112336536947e-06,
+      "loss": 1.4419,
+      "step": 1454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.620352134866867e-06,
+      "loss": 1.3974,
+      "step": 1455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.619591933196787e-06,
+      "loss": 1.3609,
+      "step": 1456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.618831731526706e-06,
+      "loss": 1.4891,
+      "step": 1457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.618071529856626e-06,
+      "loss": 1.3381,
+      "step": 1458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.617311328186546e-06,
+      "loss": 1.3948,
+      "step": 1459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.616551126516466e-06,
+      "loss": 1.2775,
+      "step": 1460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.615790924846386e-06,
+      "loss": 1.2258,
+      "step": 1461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.615030723176304e-06,
+      "loss": 1.3358,
+      "step": 1462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.614270521506224e-06,
+      "loss": 1.2964,
+      "step": 1463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.613510319836144e-06,
+      "loss": 1.4727,
+      "step": 1464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.612750118166064e-06,
+      "loss": 1.4351,
+      "step": 1465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.611989916495983e-06,
+      "loss": 1.5628,
+      "step": 1466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.611229714825902e-06,
+      "loss": 1.3529,
+      "step": 1467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.610469513155821e-06,
+      "loss": 1.4719,
+      "step": 1468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.609709311485741e-06,
+      "loss": 1.3918,
+      "step": 1469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.608949109815661e-06,
+      "loss": 1.4903,
+      "step": 1470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.608188908145581e-06,
+      "loss": 1.383,
+      "step": 1471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.607428706475501e-06,
+      "loss": 1.2196,
+      "step": 1472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.60666850480542e-06,
+      "loss": 1.4782,
+      "step": 1473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.60590830313534e-06,
+      "loss": 1.4255,
+      "step": 1474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.60514810146526e-06,
+      "loss": 1.5154,
+      "step": 1475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.60438789979518e-06,
+      "loss": 1.2837,
+      "step": 1476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.6036276981251e-06,
+      "loss": 1.3525,
+      "step": 1477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.602867496455018e-06,
+      "loss": 1.3548,
+      "step": 1478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.602107294784938e-06,
+      "loss": 1.4628,
+      "step": 1479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.601347093114858e-06,
+      "loss": 1.6122,
+      "step": 1480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.600586891444778e-06,
+      "loss": 1.5763,
+      "step": 1481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599826689774698e-06,
+      "loss": 1.3046,
+      "step": 1482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599066488104616e-06,
+      "loss": 1.2141,
+      "step": 1483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.598306286434536e-06,
+      "loss": 1.4136,
+      "step": 1484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597546084764456e-06,
+      "loss": 1.3757,
+      "step": 1485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596785883094375e-06,
+      "loss": 1.7127,
+      "step": 1486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596025681424295e-06,
+      "loss": 1.3335,
+      "step": 1487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.595265479754215e-06,
+      "loss": 1.4815,
+      "step": 1488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594505278084135e-06,
+      "loss": 1.4014,
+      "step": 1489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.593745076414055e-06,
+      "loss": 1.2981,
+      "step": 1490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592984874743975e-06,
+      "loss": 1.4177,
+      "step": 1491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592224673073895e-06,
+      "loss": 1.3184,
+      "step": 1492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591464471403814e-06,
+      "loss": 1.4789,
+      "step": 1493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.590704269733733e-06,
+      "loss": 1.5083,
+      "step": 1494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.589944068063652e-06,
+      "loss": 1.3599,
+      "step": 1495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.589183866393572e-06,
+      "loss": 1.2377,
+      "step": 1496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588423664723492e-06,
+      "loss": 1.4046,
+      "step": 1497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.587663463053412e-06,
+      "loss": 1.4095,
+      "step": 1498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586903261383332e-06,
+      "loss": 1.6165,
+      "step": 1499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58614305971325e-06,
+      "loss": 1.357,
+      "step": 1500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58538285804317e-06,
+      "loss": 1.6148,
+      "step": 1501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58462265637309e-06,
+      "loss": 1.6572,
+      "step": 1502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58386245470301e-06,
+      "loss": 1.6085,
+      "step": 1503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58310225303293e-06,
+      "loss": 1.3749,
+      "step": 1504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58234205136285e-06,
+      "loss": 1.3336,
+      "step": 1505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58158184969277e-06,
+      "loss": 1.4688,
+      "step": 1506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.580821648022689e-06,
+      "loss": 1.4512,
+      "step": 1507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.580061446352609e-06,
+      "loss": 1.2293,
+      "step": 1508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.579301244682529e-06,
+      "loss": 1.4701,
+      "step": 1509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.578541043012447e-06,
+      "loss": 1.3222,
+      "step": 1510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.577780841342367e-06,
+      "loss": 1.4138,
+      "step": 1511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.577020639672287e-06,
+      "loss": 1.2745,
+      "step": 1512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.576260438002206e-06,
+      "loss": 1.2798,
+      "step": 1513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.575500236332126e-06,
+      "loss": 1.3101,
+      "step": 1514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.574740034662046e-06,
+      "loss": 1.451,
+      "step": 1515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.573979832991966e-06,
+      "loss": 1.4659,
+      "step": 1516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.573219631321886e-06,
+      "loss": 1.402,
+      "step": 1517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.572459429651806e-06,
+      "loss": 1.3164,
+      "step": 1518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.571699227981726e-06,
+      "loss": 1.2531,
+      "step": 1519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.570939026311644e-06,
+      "loss": 1.3287,
+      "step": 1520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.570178824641564e-06,
+      "loss": 1.2967,
+      "step": 1521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.569418622971483e-06,
+      "loss": 1.2964,
+      "step": 1522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.568658421301403e-06,
+      "loss": 1.412,
+      "step": 1523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.567898219631323e-06,
+      "loss": 1.5136,
+      "step": 1524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.567138017961243e-06,
+      "loss": 1.3654,
+      "step": 1525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.566377816291161e-06,
+      "loss": 1.4518,
+      "step": 1526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.565617614621081e-06,
+      "loss": 1.4053,
+      "step": 1527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.564857412951001e-06,
+      "loss": 1.5531,
+      "step": 1528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.56409721128092e-06,
+      "loss": 1.254,
+      "step": 1529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.56333700961084e-06,
+      "loss": 1.4144,
+      "step": 1530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.56257680794076e-06,
+      "loss": 1.4686,
+      "step": 1531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.56181660627068e-06,
+      "loss": 1.3685,
+      "step": 1532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.5610564046006e-06,
+      "loss": 1.1566,
+      "step": 1533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.56029620293052e-06,
+      "loss": 1.3067,
+      "step": 1534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.55953600126044e-06,
+      "loss": 1.6265,
+      "step": 1535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.558775799590358e-06,
+      "loss": 1.5893,
+      "step": 1536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.558015597920278e-06,
+      "loss": 1.3285,
+      "step": 1537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.557255396250198e-06,
+      "loss": 1.4286,
+      "step": 1538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.556495194580118e-06,
+      "loss": 1.1953,
+      "step": 1539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.555734992910038e-06,
+      "loss": 1.2664,
+      "step": 1540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.554974791239957e-06,
+      "loss": 1.4058,
+      "step": 1541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.554214589569876e-06,
+      "loss": 1.3149,
+      "step": 1542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.553454387899795e-06,
+      "loss": 1.21,
+      "step": 1543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.552694186229715e-06,
+      "loss": 1.349,
+      "step": 1544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.551933984559635e-06,
+      "loss": 1.2475,
+      "step": 1545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.551173782889555e-06,
+      "loss": 1.5278,
+      "step": 1546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.550413581219475e-06,
+      "loss": 1.173,
+      "step": 1547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.549653379549395e-06,
+      "loss": 1.3403,
+      "step": 1548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.548893177879315e-06,
+      "loss": 1.41,
+      "step": 1549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.548132976209234e-06,
+      "loss": 1.2517,
+      "step": 1550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.547372774539154e-06,
+      "loss": 1.3474,
+      "step": 1551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.546612572869072e-06,
+      "loss": 1.4051,
+      "step": 1552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.545852371198992e-06,
+      "loss": 1.444,
+      "step": 1553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.545092169528912e-06,
+      "loss": 1.2011,
+      "step": 1554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.544331967858832e-06,
+      "loss": 1.2855,
+      "step": 1555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.543571766188752e-06,
+      "loss": 1.4307,
+      "step": 1556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.542811564518672e-06,
+      "loss": 1.4184,
+      "step": 1557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.54205136284859e-06,
+      "loss": 1.3908,
+      "step": 1558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.54129116117851e-06,
+      "loss": 1.415,
+      "step": 1559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.54053095950843e-06,
+      "loss": 1.1986,
+      "step": 1560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.53977075783835e-06,
+      "loss": 1.5207,
+      "step": 1561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.53901055616827e-06,
+      "loss": 1.5282,
+      "step": 1562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.538250354498189e-06,
+      "loss": 1.0538,
+      "step": 1563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.537490152828109e-06,
+      "loss": 1.5077,
+      "step": 1564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.536729951158029e-06,
+      "loss": 1.4205,
+      "step": 1565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.535969749487949e-06,
+      "loss": 1.3032,
+      "step": 1566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.535209547817869e-06,
+      "loss": 1.4756,
+      "step": 1567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.534449346147787e-06,
+      "loss": 1.0812,
+      "step": 1568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.533689144477707e-06,
+      "loss": 1.3467,
+      "step": 1569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.532928942807626e-06,
+      "loss": 1.3923,
+      "step": 1570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.532168741137546e-06,
+      "loss": 1.4255,
+      "step": 1571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.531408539467466e-06,
+      "loss": 1.387,
+      "step": 1572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.530648337797386e-06,
+      "loss": 1.2701,
+      "step": 1573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.529888136127304e-06,
+      "loss": 1.2131,
+      "step": 1574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.529127934457224e-06,
+      "loss": 1.384,
+      "step": 1575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.528367732787144e-06,
+      "loss": 1.3066,
+      "step": 1576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.527607531117064e-06,
+      "loss": 1.3073,
+      "step": 1577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.526847329446984e-06,
+      "loss": 1.2772,
+      "step": 1578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.526087127776903e-06,
+      "loss": 1.3299,
+      "step": 1579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.525326926106823e-06,
+      "loss": 1.1874,
+      "step": 1580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.524566724436743e-06,
+      "loss": 1.214,
+      "step": 1581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.523806522766663e-06,
+      "loss": 1.274,
+      "step": 1582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.523046321096583e-06,
+      "loss": 1.1815,
+      "step": 1583
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.522286119426501e-06,
+      "loss": 1.3168,
+      "step": 1584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.521525917756421e-06,
+      "loss": 1.4158,
+      "step": 1585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.52076571608634e-06,
+      "loss": 1.3429,
+      "step": 1586
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.52000551441626e-06,
+      "loss": 1.348,
+      "step": 1587
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.51924531274618e-06,
+      "loss": 1.4661,
+      "step": 1588
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.5184851110761e-06,
+      "loss": 1.3636,
+      "step": 1589
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.517724909406018e-06,
+      "loss": 1.2939,
+      "step": 1590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.516964707735938e-06,
+      "loss": 1.3934,
+      "step": 1591
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.516204506065858e-06,
+      "loss": 1.2125,
+      "step": 1592
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.515444304395778e-06,
+      "loss": 1.3041,
+      "step": 1593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.514684102725698e-06,
+      "loss": 1.2102,
+      "step": 1594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.513923901055618e-06,
+      "loss": 1.2738,
+      "step": 1595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.513163699385538e-06,
+      "loss": 1.3138,
+      "step": 1596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.512403497715457e-06,
+      "loss": 1.2087,
+      "step": 1597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.511643296045377e-06,
+      "loss": 1.2621,
+      "step": 1598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.510883094375297e-06,
+      "loss": 1.3845,
+      "step": 1599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.510122892705215e-06,
+      "loss": 1.3296,
+      "step": 1600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.509362691035135e-06,
+      "loss": 1.3385,
+      "step": 1601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.508602489365055e-06,
+      "loss": 1.3095,
+      "step": 1602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.507842287694975e-06,
+      "loss": 1.4463,
+      "step": 1603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.507082086024895e-06,
+      "loss": 1.3127,
+      "step": 1604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.506321884354815e-06,
+      "loss": 1.4048,
+      "step": 1605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.505561682684733e-06,
+      "loss": 1.3221,
+      "step": 1606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.504801481014653e-06,
+      "loss": 1.3941,
+      "step": 1607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.504041279344572e-06,
+      "loss": 1.5774,
+      "step": 1608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.503281077674492e-06,
+      "loss": 1.273,
+      "step": 1609
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.502520876004412e-06,
+      "loss": 1.378,
+      "step": 1610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.501760674334332e-06,
+      "loss": 1.1985,
+      "step": 1611
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.501000472664252e-06,
+      "loss": 1.4,
+      "step": 1612
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.500240270994172e-06,
+      "loss": 1.4136,
+      "step": 1613
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.499480069324092e-06,
+      "loss": 1.4006,
+      "step": 1614
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.498719867654011e-06,
+      "loss": 1.2703,
+      "step": 1615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.49795966598393e-06,
+      "loss": 1.2826,
+      "step": 1616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.49719946431385e-06,
+      "loss": 1.2001,
+      "step": 1617
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.49643926264377e-06,
+      "loss": 1.1194,
+      "step": 1618
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.49567906097369e-06,
+      "loss": 1.2897,
+      "step": 1619
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.494918859303609e-06,
+      "loss": 1.1796,
+      "step": 1620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.494158657633527e-06,
+      "loss": 1.1554,
+      "step": 1621
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.493398455963447e-06,
+      "loss": 1.1706,
+      "step": 1622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.492638254293367e-06,
+      "loss": 1.3624,
+      "step": 1623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.491878052623287e-06,
+      "loss": 1.2864,
+      "step": 1624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.491117850953207e-06,
+      "loss": 1.4634,
+      "step": 1625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.490357649283126e-06,
+      "loss": 1.3776,
+      "step": 1626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.489597447613046e-06,
+      "loss": 1.3392,
+      "step": 1627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.488837245942966e-06,
+      "loss": 1.7144,
+      "step": 1628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.488077044272886e-06,
+      "loss": 1.3656,
+      "step": 1629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.487316842602806e-06,
+      "loss": 1.4388,
+      "step": 1630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.486556640932726e-06,
+      "loss": 1.3942,
+      "step": 1631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.485796439262644e-06,
+      "loss": 1.423,
+      "step": 1632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.485036237592564e-06,
+      "loss": 1.4026,
+      "step": 1633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.484276035922484e-06,
+      "loss": 1.3667,
+      "step": 1634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.483515834252404e-06,
+      "loss": 1.5018,
+      "step": 1635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.482755632582323e-06,
+      "loss": 1.2863,
+      "step": 1636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.481995430912242e-06,
+      "loss": 1.6069,
+      "step": 1637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.481235229242161e-06,
+      "loss": 1.3593,
+      "step": 1638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.480475027572081e-06,
+      "loss": 1.346,
+      "step": 1639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.479714825902001e-06,
+      "loss": 1.4572,
+      "step": 1640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.478954624231921e-06,
+      "loss": 1.4255,
+      "step": 1641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.47819442256184e-06,
+      "loss": 1.418,
+      "step": 1642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.47743422089176e-06,
+      "loss": 1.009,
+      "step": 1643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.47667401922168e-06,
+      "loss": 1.5054,
+      "step": 1644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.4759138175516e-06,
+      "loss": 1.3198,
+      "step": 1645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.47515361588152e-06,
+      "loss": 1.4973,
+      "step": 1646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.47439341421144e-06,
+      "loss": 1.3759,
+      "step": 1647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.473633212541358e-06,
+      "loss": 1.4636,
+      "step": 1648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.472873010871278e-06,
+      "loss": 1.431,
+      "step": 1649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.472112809201198e-06,
+      "loss": 1.425,
+      "step": 1650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.471352607531118e-06,
+      "loss": 1.1695,
+      "step": 1651
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.470592405861038e-06,
+      "loss": 1.6119,
+      "step": 1652
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.469832204190956e-06,
+      "loss": 1.2753,
+      "step": 1653
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.469072002520876e-06,
+      "loss": 1.4214,
+      "step": 1654
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.468311800850796e-06,
+      "loss": 1.4032,
+      "step": 1655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.467551599180715e-06,
+      "loss": 1.1992,
+      "step": 1656
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.466791397510635e-06,
+      "loss": 1.3483,
+      "step": 1657
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.466031195840555e-06,
+      "loss": 1.2886,
+      "step": 1658
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.465270994170475e-06,
+      "loss": 1.4551,
+      "step": 1659
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.464510792500395e-06,
+      "loss": 1.3931,
+      "step": 1660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.463750590830315e-06,
+      "loss": 1.5145,
+      "step": 1661
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.462990389160235e-06,
+      "loss": 1.587,
+      "step": 1662
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.462230187490154e-06,
+      "loss": 1.5279,
+      "step": 1663
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.461469985820073e-06,
+      "loss": 1.2726,
+      "step": 1664
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.460709784149992e-06,
+      "loss": 1.3044,
+      "step": 1665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.459949582479912e-06,
+      "loss": 1.5186,
+      "step": 1666
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.459189380809832e-06,
+      "loss": 1.3741,
+      "step": 1667
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.458429179139752e-06,
+      "loss": 1.545,
+      "step": 1668
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45766897746967e-06,
+      "loss": 1.4269,
+      "step": 1669
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45690877579959e-06,
+      "loss": 1.3499,
+      "step": 1670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45614857412951e-06,
+      "loss": 1.3459,
+      "step": 1671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45538837245943e-06,
+      "loss": 1.4309,
+      "step": 1672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45462817078935e-06,
+      "loss": 1.3796,
+      "step": 1673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45386796911927e-06,
+      "loss": 1.2724,
+      "step": 1674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45310776744919e-06,
+      "loss": 1.4519,
+      "step": 1675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.452347565779109e-06,
+      "loss": 1.3971,
+      "step": 1676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.451587364109029e-06,
+      "loss": 1.4562,
+      "step": 1677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.450827162438949e-06,
+      "loss": 1.3134,
+      "step": 1678
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.450066960768869e-06,
+      "loss": 1.4939,
+      "step": 1679
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.449306759098787e-06,
+      "loss": 1.4384,
+      "step": 1680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.448546557428707e-06,
+      "loss": 1.3902,
+      "step": 1681
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.447786355758627e-06,
+      "loss": 1.3964,
+      "step": 1682
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.447026154088546e-06,
+      "loss": 1.3045,
+      "step": 1683
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.446265952418466e-06,
+      "loss": 1.4732,
+      "step": 1684
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.445505750748386e-06,
+      "loss": 1.4056,
+      "step": 1685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.444745549078306e-06,
+      "loss": 1.3019,
+      "step": 1686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.443985347408226e-06,
+      "loss": 1.3607,
+      "step": 1687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.443225145738144e-06,
+      "loss": 1.3341,
+      "step": 1688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.442464944068064e-06,
+      "loss": 1.5496,
+      "step": 1689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.441704742397984e-06,
+      "loss": 1.3845,
+      "step": 1690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.440944540727904e-06,
+      "loss": 1.2773,
+      "step": 1691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.440184339057823e-06,
+      "loss": 1.362,
+      "step": 1692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.439424137387743e-06,
+      "loss": 1.3822,
+      "step": 1693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.438663935717663e-06,
+      "loss": 1.3242,
+      "step": 1694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.437903734047583e-06,
+      "loss": 1.5248,
+      "step": 1695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.437143532377501e-06,
+      "loss": 1.3101,
+      "step": 1696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.436383330707421e-06,
+      "loss": 1.2467,
+      "step": 1697
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.435623129037341e-06,
+      "loss": 1.6031,
+      "step": 1698
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.43486292736726e-06,
+      "loss": 1.4828,
+      "step": 1699
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.43410272569718e-06,
+      "loss": 1.3782,
+      "step": 1700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.4333425240271e-06,
+      "loss": 1.1876,
+      "step": 1701
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.43258232235702e-06,
+      "loss": 1.0637,
+      "step": 1702
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.43182212068694e-06,
+      "loss": 1.0631,
+      "step": 1703
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.43106191901686e-06,
+      "loss": 1.349,
+      "step": 1704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.43030171734678e-06,
+      "loss": 1.3973,
+      "step": 1705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.4295415156767e-06,
+      "loss": 1.3436,
+      "step": 1706
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.428781314006618e-06,
+      "loss": 1.4079,
+      "step": 1707
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.428021112336538e-06,
+      "loss": 1.5668,
+      "step": 1708
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.427260910666458e-06,
+      "loss": 1.384,
+      "step": 1709
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.426500708996377e-06,
+      "loss": 1.2124,
+      "step": 1710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.425740507326297e-06,
+      "loss": 1.4961,
+      "step": 1711
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.424980305656215e-06,
+      "loss": 1.625,
+      "step": 1712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.424220103986135e-06,
+      "loss": 1.4717,
+      "step": 1713
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.423459902316055e-06,
+      "loss": 1.3217,
+      "step": 1714
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.422699700645975e-06,
+      "loss": 1.258,
+      "step": 1715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.421939498975895e-06,
+      "loss": 1.21,
+      "step": 1716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.421179297305815e-06,
+      "loss": 1.3132,
+      "step": 1717
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.420419095635735e-06,
+      "loss": 1.6623,
+      "step": 1718
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.419658893965654e-06,
+      "loss": 1.5256,
+      "step": 1719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.418898692295574e-06,
+      "loss": 1.2645,
+      "step": 1720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.418138490625494e-06,
+      "loss": 1.5833,
+      "step": 1721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.417378288955414e-06,
+      "loss": 1.3642,
+      "step": 1722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.416618087285332e-06,
+      "loss": 1.5207,
+      "step": 1723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.415857885615252e-06,
+      "loss": 1.3546,
+      "step": 1724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.415097683945172e-06,
+      "loss": 1.3814,
+      "step": 1725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.414337482275092e-06,
+      "loss": 1.3789,
+      "step": 1726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.413577280605012e-06,
+      "loss": 1.2993,
+      "step": 1727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.41281707893493e-06,
+      "loss": 1.3542,
+      "step": 1728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.41205687726485e-06,
+      "loss": 1.4631,
+      "step": 1729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.41129667559477e-06,
+      "loss": 1.228,
+      "step": 1730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.41053647392469e-06,
+      "loss": 1.4374,
+      "step": 1731
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.40977627225461e-06,
+      "loss": 1.264,
+      "step": 1732
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.409016070584529e-06,
+      "loss": 1.4896,
+      "step": 1733
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.408255868914449e-06,
+      "loss": 1.417,
+      "step": 1734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.407495667244369e-06,
+      "loss": 1.6178,
+      "step": 1735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.406735465574289e-06,
+      "loss": 1.4296,
+      "step": 1736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.405975263904208e-06,
+      "loss": 1.5811,
+      "step": 1737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.405215062234128e-06,
+      "loss": 1.5814,
+      "step": 1738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.404454860564047e-06,
+      "loss": 1.222,
+      "step": 1739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.403694658893966e-06,
+      "loss": 1.195,
+      "step": 1740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.402934457223886e-06,
+      "loss": 1.3003,
+      "step": 1741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.402174255553806e-06,
+      "loss": 1.3153,
+      "step": 1742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.401414053883726e-06,
+      "loss": 1.4128,
+      "step": 1743
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.400653852213644e-06,
+      "loss": 1.1678,
+      "step": 1744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.399893650543564e-06,
+      "loss": 1.0683,
+      "step": 1745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.399133448873484e-06,
+      "loss": 1.3394,
+      "step": 1746
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.398373247203404e-06,
+      "loss": 1.1642,
+      "step": 1747
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.397613045533324e-06,
+      "loss": 1.4929,
+      "step": 1748
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.396852843863243e-06,
+      "loss": 1.41,
+      "step": 1749
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.396092642193163e-06,
+      "loss": 1.4419,
+      "step": 1750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.395332440523083e-06,
+      "loss": 1.332,
+      "step": 1751
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.394572238853003e-06,
+      "loss": 1.2731,
+      "step": 1752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.393812037182923e-06,
+      "loss": 1.456,
+      "step": 1753
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.393051835512843e-06,
+      "loss": 1.4359,
+      "step": 1754
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.39229163384276e-06,
+      "loss": 1.4203,
+      "step": 1755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.39153143217268e-06,
+      "loss": 1.4521,
+      "step": 1756
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.3907712305026e-06,
+      "loss": 1.3757,
+      "step": 1757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.39001102883252e-06,
+      "loss": 1.4959,
+      "step": 1758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.38925082716244e-06,
+      "loss": 1.219,
+      "step": 1759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.388490625492358e-06,
+      "loss": 1.3832,
+      "step": 1760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.387730423822278e-06,
+      "loss": 1.4481,
+      "step": 1761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.386970222152198e-06,
+      "loss": 1.4765,
+      "step": 1762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.386210020482118e-06,
+      "loss": 1.3403,
+      "step": 1763
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.385449818812038e-06,
+      "loss": 1.5001,
+      "step": 1764
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.384689617141958e-06,
+      "loss": 1.2376,
+      "step": 1765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.383929415471878e-06,
+      "loss": 1.2154,
+      "step": 1766
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.383169213801797e-06,
+      "loss": 1.3498,
+      "step": 1767
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.382409012131717e-06,
+      "loss": 1.1033,
+      "step": 1768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.381648810461637e-06,
+      "loss": 1.2278,
+      "step": 1769
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.380888608791557e-06,
+      "loss": 1.5479,
+      "step": 1770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.380128407121475e-06,
+      "loss": 1.4408,
+      "step": 1771
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.379368205451395e-06,
+      "loss": 1.24,
+      "step": 1772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.378608003781315e-06,
+      "loss": 1.5124,
+      "step": 1773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.377847802111235e-06,
+      "loss": 1.475,
+      "step": 1774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.377087600441155e-06,
+      "loss": 1.5395,
+      "step": 1775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.376327398771073e-06,
+      "loss": 1.5259,
+      "step": 1776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.375567197100993e-06,
+      "loss": 1.3997,
+      "step": 1777
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.374806995430912e-06,
+      "loss": 1.1309,
+      "step": 1778
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.374046793760832e-06,
+      "loss": 1.2717,
+      "step": 1779
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.373286592090752e-06,
+      "loss": 1.526,
+      "step": 1780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.372526390420672e-06,
+      "loss": 1.2044,
+      "step": 1781
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.371766188750592e-06,
+      "loss": 1.2842,
+      "step": 1782
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.371005987080512e-06,
+      "loss": 1.3924,
+      "step": 1783
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.370245785410432e-06,
+      "loss": 1.429,
+      "step": 1784
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.369485583740351e-06,
+      "loss": 1.1351,
+      "step": 1785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.368725382070271e-06,
+      "loss": 1.2203,
+      "step": 1786
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.36796518040019e-06,
+      "loss": 1.4443,
+      "step": 1787
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.36720497873011e-06,
+      "loss": 1.3354,
+      "step": 1788
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.366444777060029e-06,
+      "loss": 1.535,
+      "step": 1789
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.365684575389949e-06,
+      "loss": 1.4893,
+      "step": 1790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.364924373719869e-06,
+      "loss": 1.3788,
+      "step": 1791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.364164172049787e-06,
+      "loss": 1.1529,
+      "step": 1792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.363403970379707e-06,
+      "loss": 1.4477,
+      "step": 1793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.362643768709627e-06,
+      "loss": 1.1785,
+      "step": 1794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.361883567039547e-06,
+      "loss": 1.282,
+      "step": 1795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.361123365369466e-06,
+      "loss": 1.5173,
+      "step": 1796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.360363163699386e-06,
+      "loss": 1.4664,
+      "step": 1797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.359602962029306e-06,
+      "loss": 1.3739,
+      "step": 1798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.358842760359226e-06,
+      "loss": 1.3202,
+      "step": 1799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.358082558689146e-06,
+      "loss": 1.2388,
+      "step": 1800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.357322357019066e-06,
+      "loss": 1.4432,
+      "step": 1801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.356562155348986e-06,
+      "loss": 1.5763,
+      "step": 1802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.355801953678904e-06,
+      "loss": 1.3051,
+      "step": 1803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.355041752008824e-06,
+      "loss": 1.5668,
+      "step": 1804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.354281550338743e-06,
+      "loss": 1.2939,
+      "step": 1805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.353521348668663e-06,
+      "loss": 1.3379,
+      "step": 1806
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.352761146998583e-06,
+      "loss": 1.2325,
+      "step": 1807
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.352000945328501e-06,
+      "loss": 1.4347,
+      "step": 1808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.351240743658421e-06,
+      "loss": 1.3954,
+      "step": 1809
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.350480541988341e-06,
+      "loss": 1.2655,
+      "step": 1810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.349720340318261e-06,
+      "loss": 1.4621,
+      "step": 1811
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.34896013864818e-06,
+      "loss": 1.4938,
+      "step": 1812
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.3481999369781e-06,
+      "loss": 1.385,
+      "step": 1813
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.34743973530802e-06,
+      "loss": 1.3137,
+      "step": 1814
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.34667953363794e-06,
+      "loss": 1.3065,
+      "step": 1815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.34591933196786e-06,
+      "loss": 1.2598,
+      "step": 1816
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.34515913029778e-06,
+      "loss": 1.3357,
+      "step": 1817
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.3443989286277e-06,
+      "loss": 1.2449,
+      "step": 1818
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.343638726957618e-06,
+      "loss": 1.1168,
+      "step": 1819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.342878525287538e-06,
+      "loss": 1.5801,
+      "step": 1820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.342118323617458e-06,
+      "loss": 1.397,
+      "step": 1821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.341358121947378e-06,
+      "loss": 1.3536,
+      "step": 1822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.340597920277297e-06,
+      "loss": 1.5536,
+      "step": 1823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.339837718607216e-06,
+      "loss": 1.3908,
+      "step": 1824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.339077516937135e-06,
+      "loss": 1.337,
+      "step": 1825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.338317315267055e-06,
+      "loss": 1.5065,
+      "step": 1826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.337557113596975e-06,
+      "loss": 1.4857,
+      "step": 1827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.336796911926895e-06,
+      "loss": 1.3389,
+      "step": 1828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.336036710256815e-06,
+      "loss": 1.3547,
+      "step": 1829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.335276508586735e-06,
+      "loss": 1.3677,
+      "step": 1830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.334516306916655e-06,
+      "loss": 1.4225,
+      "step": 1831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.333756105246574e-06,
+      "loss": 1.6334,
+      "step": 1832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.332995903576494e-06,
+      "loss": 1.3153,
+      "step": 1833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.332235701906412e-06,
+      "loss": 1.328,
+      "step": 1834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.331475500236332e-06,
+      "loss": 1.2443,
+      "step": 1835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.330715298566252e-06,
+      "loss": 1.2722,
+      "step": 1836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.329955096896172e-06,
+      "loss": 1.5149,
+      "step": 1837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.329194895226092e-06,
+      "loss": 1.2842,
+      "step": 1838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.328434693556012e-06,
+      "loss": 1.2565,
+      "step": 1839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.32767449188593e-06,
+      "loss": 1.2037,
+      "step": 1840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.32691429021585e-06,
+      "loss": 1.4241,
+      "step": 1841
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.32615408854577e-06,
+      "loss": 1.3617,
+      "step": 1842
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.32539388687569e-06,
+      "loss": 1.284,
+      "step": 1843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.32463368520561e-06,
+      "loss": 1.3021,
+      "step": 1844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.32387348353553e-06,
+      "loss": 1.4538,
+      "step": 1845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.323113281865449e-06,
+      "loss": 1.2602,
+      "step": 1846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.322353080195369e-06,
+      "loss": 1.4409,
+      "step": 1847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.321592878525289e-06,
+      "loss": 1.4311,
+      "step": 1848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.320832676855209e-06,
+      "loss": 1.3723,
+      "step": 1849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.320072475185127e-06,
+      "loss": 1.3483,
+      "step": 1850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.319312273515047e-06,
+      "loss": 1.3553,
+      "step": 1851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.318552071844967e-06,
+      "loss": 1.2314,
+      "step": 1852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.317791870174886e-06,
+      "loss": 1.4645,
+      "step": 1853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.317031668504806e-06,
+      "loss": 1.5661,
+      "step": 1854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.316271466834726e-06,
+      "loss": 1.6596,
+      "step": 1855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.315511265164644e-06,
+      "loss": 1.3587,
+      "step": 1856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.314751063494564e-06,
+      "loss": 1.2405,
+      "step": 1857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.313990861824484e-06,
+      "loss": 1.414,
+      "step": 1858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.313230660154404e-06,
+      "loss": 1.4137,
+      "step": 1859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.312470458484324e-06,
+      "loss": 1.3521,
+      "step": 1860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.311710256814244e-06,
+      "loss": 1.333,
+      "step": 1861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.310950055144163e-06,
+      "loss": 1.3086,
+      "step": 1862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.310189853474083e-06,
+      "loss": 1.2078,
+      "step": 1863
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.309429651804003e-06,
+      "loss": 1.4562,
+      "step": 1864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.308669450133923e-06,
+      "loss": 1.4703,
+      "step": 1865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.307909248463841e-06,
+      "loss": 1.4048,
+      "step": 1866
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.307149046793761e-06,
+      "loss": 1.3151,
+      "step": 1867
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.30638884512368e-06,
+      "loss": 1.3553,
+      "step": 1868
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.3056286434536e-06,
+      "loss": 1.3348,
+      "step": 1869
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.30486844178352e-06,
+      "loss": 1.2401,
+      "step": 1870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.30410824011344e-06,
+      "loss": 1.3524,
+      "step": 1871
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.30334803844336e-06,
+      "loss": 1.5631,
+      "step": 1872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.30258783677328e-06,
+      "loss": 1.6162,
+      "step": 1873
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.3018276351032e-06,
+      "loss": 1.3639,
+      "step": 1874
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.30106743343312e-06,
+      "loss": 1.4761,
+      "step": 1875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.300307231763038e-06,
+      "loss": 1.4964,
+      "step": 1876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.299547030092958e-06,
+      "loss": 1.4495,
+      "step": 1877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.298786828422878e-06,
+      "loss": 1.3722,
+      "step": 1878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.298026626752798e-06,
+      "loss": 1.2379,
+      "step": 1879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.297266425082717e-06,
+      "loss": 1.2151,
+      "step": 1880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.296506223412637e-06,
+      "loss": 1.3885,
+      "step": 1881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.295746021742555e-06,
+      "loss": 1.2472,
+      "step": 1882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.294985820072475e-06,
+      "loss": 1.3227,
+      "step": 1883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.294225618402395e-06,
+      "loss": 1.5528,
+      "step": 1884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.293465416732315e-06,
+      "loss": 1.5558,
+      "step": 1885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.292705215062235e-06,
+      "loss": 1.2983,
+      "step": 1886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.291945013392155e-06,
+      "loss": 1.4675,
+      "step": 1887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.291184811722075e-06,
+      "loss": 1.5095,
+      "step": 1888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.290424610051994e-06,
+      "loss": 1.4874,
+      "step": 1889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.289664408381914e-06,
+      "loss": 1.4294,
+      "step": 1890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.288904206711834e-06,
+      "loss": 1.366,
+      "step": 1891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.288144005041754e-06,
+      "loss": 1.3928,
+      "step": 1892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.287383803371672e-06,
+      "loss": 1.2127,
+      "step": 1893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.286623601701592e-06,
+      "loss": 1.3474,
+      "step": 1894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.285863400031512e-06,
+      "loss": 1.3461,
+      "step": 1895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.285103198361432e-06,
+      "loss": 1.3826,
+      "step": 1896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.284342996691352e-06,
+      "loss": 1.3619,
+      "step": 1897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.28358279502127e-06,
+      "loss": 1.5323,
+      "step": 1898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.28282259335119e-06,
+      "loss": 1.2059,
+      "step": 1899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.28206239168111e-06,
+      "loss": 1.2923,
+      "step": 1900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.28130219001103e-06,
+      "loss": 1.4715,
+      "step": 1901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.28054198834095e-06,
+      "loss": 1.2106,
+      "step": 1902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.279781786670869e-06,
+      "loss": 1.3863,
+      "step": 1903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.279021585000789e-06,
+      "loss": 1.3362,
+      "step": 1904
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.278261383330709e-06,
+      "loss": 1.126,
+      "step": 1905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.277501181660629e-06,
+      "loss": 1.4016,
+      "step": 1906
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.276740979990548e-06,
+      "loss": 1.3847,
+      "step": 1907
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.275980778320468e-06,
+      "loss": 1.6439,
+      "step": 1908
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.275220576650386e-06,
+      "loss": 1.2735,
+      "step": 1909
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.274460374980306e-06,
+      "loss": 1.2774,
+      "step": 1910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.273700173310226e-06,
+      "loss": 1.5266,
+      "step": 1911
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.272939971640146e-06,
+      "loss": 1.5195,
+      "step": 1912
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.272179769970066e-06,
+      "loss": 1.3756,
+      "step": 1913
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.271419568299984e-06,
+      "loss": 1.2264,
+      "step": 1914
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.270659366629904e-06,
+      "loss": 1.4167,
+      "step": 1915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.269899164959824e-06,
+      "loss": 1.5746,
+      "step": 1916
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.269138963289744e-06,
+      "loss": 1.479,
+      "step": 1917
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.268378761619663e-06,
+      "loss": 1.3869,
+      "step": 1918
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.267618559949583e-06,
+      "loss": 1.4349,
+      "step": 1919
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.266858358279503e-06,
+      "loss": 1.5388,
+      "step": 1920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.266098156609423e-06,
+      "loss": 1.1452,
+      "step": 1921
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.265337954939343e-06,
+      "loss": 1.3469,
+      "step": 1922
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.264577753269263e-06,
+      "loss": 1.3358,
+      "step": 1923
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.263817551599183e-06,
+      "loss": 1.2935,
+      "step": 1924
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.2630573499291e-06,
+      "loss": 1.4482,
+      "step": 1925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.26229714825902e-06,
+      "loss": 1.3748,
+      "step": 1926
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.26153694658894e-06,
+      "loss": 1.115,
+      "step": 1927
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.26077674491886e-06,
+      "loss": 1.2499,
+      "step": 1928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.26001654324878e-06,
+      "loss": 1.3502,
+      "step": 1929
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.259256341578698e-06,
+      "loss": 1.1865,
+      "step": 1930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.258496139908618e-06,
+      "loss": 1.2873,
+      "step": 1931
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.257735938238538e-06,
+      "loss": 1.4542,
+      "step": 1932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.256975736568458e-06,
+      "loss": 1.4403,
+      "step": 1933
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.256215534898378e-06,
+      "loss": 1.4661,
+      "step": 1934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.255455333228298e-06,
+      "loss": 1.2857,
+      "step": 1935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.254695131558217e-06,
+      "loss": 1.6667,
+      "step": 1936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.253934929888137e-06,
+      "loss": 1.418,
+      "step": 1937
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.253174728218057e-06,
+      "loss": 1.2644,
+      "step": 1938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.252414526547977e-06,
+      "loss": 1.1874,
+      "step": 1939
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.251654324877897e-06,
+      "loss": 1.7629,
+      "step": 1940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.250894123207815e-06,
+      "loss": 1.3369,
+      "step": 1941
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.250133921537735e-06,
+      "loss": 1.5978,
+      "step": 1942
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.249373719867655e-06,
+      "loss": 1.4653,
+      "step": 1943
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.248613518197575e-06,
+      "loss": 1.1347,
+      "step": 1944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.247853316527494e-06,
+      "loss": 1.5568,
+      "step": 1945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.247093114857413e-06,
+      "loss": 1.3412,
+      "step": 1946
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.246332913187333e-06,
+      "loss": 1.3845,
+      "step": 1947
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.245572711517252e-06,
+      "loss": 1.3072,
+      "step": 1948
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.244812509847172e-06,
+      "loss": 1.4252,
+      "step": 1949
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.244052308177092e-06,
+      "loss": 1.2077,
+      "step": 1950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.243292106507012e-06,
+      "loss": 1.4087,
+      "step": 1951
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.242531904836932e-06,
+      "loss": 1.3946,
+      "step": 1952
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.241771703166852e-06,
+      "loss": 1.5528,
+      "step": 1953
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.241011501496772e-06,
+      "loss": 1.4097,
+      "step": 1954
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.240251299826691e-06,
+      "loss": 1.3278,
+      "step": 1955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.239491098156611e-06,
+      "loss": 1.2643,
+      "step": 1956
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.23873089648653e-06,
+      "loss": 1.4371,
+      "step": 1957
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.23797069481645e-06,
+      "loss": 1.4819,
+      "step": 1958
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.237210493146369e-06,
+      "loss": 1.2409,
+      "step": 1959
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.236450291476289e-06,
+      "loss": 1.5432,
+      "step": 1960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.235690089806209e-06,
+      "loss": 1.3632,
+      "step": 1961
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.234929888136127e-06,
+      "loss": 1.4532,
+      "step": 1962
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.234169686466047e-06,
+      "loss": 1.4816,
+      "step": 1963
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.233409484795967e-06,
+      "loss": 1.5985,
+      "step": 1964
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.232649283125887e-06,
+      "loss": 1.5842,
+      "step": 1965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.231889081455806e-06,
+      "loss": 1.4346,
+      "step": 1966
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.231128879785726e-06,
+      "loss": 1.2618,
+      "step": 1967
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.230368678115646e-06,
+      "loss": 1.4822,
+      "step": 1968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.229608476445566e-06,
+      "loss": 1.4432,
+      "step": 1969
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.228848274775486e-06,
+      "loss": 1.4821,
+      "step": 1970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.228088073105406e-06,
+      "loss": 1.3665,
+      "step": 1971
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.227327871435326e-06,
+      "loss": 1.4281,
+      "step": 1972
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.226567669765244e-06,
+      "loss": 1.3008,
+      "step": 1973
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.225807468095164e-06,
+      "loss": 1.3864,
+      "step": 1974
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.225047266425083e-06,
+      "loss": 1.2664,
+      "step": 1975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.224287064755003e-06,
+      "loss": 1.4096,
+      "step": 1976
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.223526863084923e-06,
+      "loss": 1.4578,
+      "step": 1977
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.222766661414841e-06,
+      "loss": 1.2244,
+      "step": 1978
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.222006459744761e-06,
+      "loss": 1.4833,
+      "step": 1979
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.221246258074681e-06,
+      "loss": 1.3807,
+      "step": 1980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.2204860564046e-06,
+      "loss": 1.5249,
+      "step": 1981
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.21972585473452e-06,
+      "loss": 1.1829,
+      "step": 1982
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.21896565306444e-06,
+      "loss": 1.3978,
+      "step": 1983
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.21820545139436e-06,
+      "loss": 1.2765,
+      "step": 1984
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.21744524972428e-06,
+      "loss": 1.2426,
+      "step": 1985
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.2166850480542e-06,
+      "loss": 1.4261,
+      "step": 1986
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.21592484638412e-06,
+      "loss": 1.2995,
+      "step": 1987
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.21516464471404e-06,
+      "loss": 1.387,
+      "step": 1988
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.214404443043958e-06,
+      "loss": 1.5486,
+      "step": 1989
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.213644241373878e-06,
+      "loss": 1.3222,
+      "step": 1990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.212884039703798e-06,
+      "loss": 1.2084,
+      "step": 1991
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.212123838033718e-06,
+      "loss": 1.48,
+      "step": 1992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.211363636363637e-06,
+      "loss": 1.369,
+      "step": 1993
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.210603434693556e-06,
+      "loss": 1.4334,
+      "step": 1994
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.209843233023475e-06,
+      "loss": 1.2505,
+      "step": 1995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.209083031353395e-06,
+      "loss": 1.2847,
+      "step": 1996
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.208322829683315e-06,
+      "loss": 1.4569,
+      "step": 1997
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.207562628013235e-06,
+      "loss": 1.4784,
+      "step": 1998
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.206802426343155e-06,
+      "loss": 1.3742,
+      "step": 1999
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.206042224673075e-06,
+      "loss": 1.4369,
+      "step": 2000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.205282023002995e-06,
+      "loss": 1.2111,
+      "step": 2001
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.204521821332914e-06,
+      "loss": 1.1964,
+      "step": 2002
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.203761619662834e-06,
+      "loss": 1.3417,
+      "step": 2003
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.203001417992754e-06,
+      "loss": 1.2636,
+      "step": 2004
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.202241216322672e-06,
+      "loss": 1.3653,
+      "step": 2005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.201481014652592e-06,
+      "loss": 1.54,
+      "step": 2006
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.200720812982512e-06,
+      "loss": 1.5481,
+      "step": 2007
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.199960611312432e-06,
+      "loss": 1.3767,
+      "step": 2008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.199200409642352e-06,
+      "loss": 1.2291,
+      "step": 2009
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.19844020797227e-06,
+      "loss": 1.4797,
+      "step": 2010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.19768000630219e-06,
+      "loss": 1.3699,
+      "step": 2011
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.19691980463211e-06,
+      "loss": 1.5709,
+      "step": 2012
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.19615960296203e-06,
+      "loss": 1.4292,
+      "step": 2013
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.19539940129195e-06,
+      "loss": 1.1301,
+      "step": 2014
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.19463919962187e-06,
+      "loss": 1.3803,
+      "step": 2015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.193878997951789e-06,
+      "loss": 1.5824,
+      "step": 2016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.193118796281709e-06,
+      "loss": 1.3434,
+      "step": 2017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.192358594611629e-06,
+      "loss": 1.4552,
+      "step": 2018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.191598392941549e-06,
+      "loss": 1.1781,
+      "step": 2019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.190838191271468e-06,
+      "loss": 1.4657,
+      "step": 2020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.190077989601387e-06,
+      "loss": 1.36,
+      "step": 2021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.189317787931306e-06,
+      "loss": 1.4667,
+      "step": 2022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.188557586261226e-06,
+      "loss": 1.4395,
+      "step": 2023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.187797384591146e-06,
+      "loss": 1.3262,
+      "step": 2024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.187037182921066e-06,
+      "loss": 1.2519,
+      "step": 2025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.186276981250984e-06,
+      "loss": 1.3585,
+      "step": 2026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.185516779580904e-06,
+      "loss": 1.3679,
+      "step": 2027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.184756577910824e-06,
+      "loss": 1.545,
+      "step": 2028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.183996376240744e-06,
+      "loss": 1.5523,
+      "step": 2029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.183236174570664e-06,
+      "loss": 1.3951,
+      "step": 2030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.182475972900583e-06,
+      "loss": 1.5759,
+      "step": 2031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.181715771230503e-06,
+      "loss": 1.314,
+      "step": 2032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.180955569560423e-06,
+      "loss": 1.5222,
+      "step": 2033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.180195367890343e-06,
+      "loss": 1.392,
+      "step": 2034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.179435166220263e-06,
+      "loss": 1.4214,
+      "step": 2035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.178674964550183e-06,
+      "loss": 1.3944,
+      "step": 2036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.177914762880101e-06,
+      "loss": 1.3858,
+      "step": 2037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.17715456121002e-06,
+      "loss": 1.3565,
+      "step": 2038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.17639435953994e-06,
+      "loss": 1.5445,
+      "step": 2039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.17563415786986e-06,
+      "loss": 1.4424,
+      "step": 2040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.17487395619978e-06,
+      "loss": 1.3964,
+      "step": 2041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.174113754529699e-06,
+      "loss": 1.3372,
+      "step": 2042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.173353552859618e-06,
+      "loss": 1.482,
+      "step": 2043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.172593351189538e-06,
+      "loss": 1.3838,
+      "step": 2044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.171833149519458e-06,
+      "loss": 1.3161,
+      "step": 2045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.171072947849378e-06,
+      "loss": 1.3731,
+      "step": 2046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.170312746179298e-06,
+      "loss": 1.2931,
+      "step": 2047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.169552544509218e-06,
+      "loss": 1.2749,
+      "step": 2048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.168792342839137e-06,
+      "loss": 1.4976,
+      "step": 2049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.168032141169057e-06,
+      "loss": 1.1565,
+      "step": 2050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.167271939498977e-06,
+      "loss": 1.3688,
+      "step": 2051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.166511737828897e-06,
+      "loss": 1.4599,
+      "step": 2052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.165751536158815e-06,
+      "loss": 1.3849,
+      "step": 2053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.164991334488735e-06,
+      "loss": 1.4893,
+      "step": 2054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.164231132818655e-06,
+      "loss": 1.2358,
+      "step": 2055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.163470931148575e-06,
+      "loss": 1.319,
+      "step": 2056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.162710729478495e-06,
+      "loss": 1.4412,
+      "step": 2057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.161950527808415e-06,
+      "loss": 1.3388,
+      "step": 2058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.161190326138334e-06,
+      "loss": 1.2471,
+      "step": 2059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.160430124468254e-06,
+      "loss": 1.2636,
+      "step": 2060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.159669922798174e-06,
+      "loss": 1.392,
+      "step": 2061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.158909721128094e-06,
+      "loss": 1.2096,
+      "step": 2062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.158149519458012e-06,
+      "loss": 1.2603,
+      "step": 2063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.157389317787932e-06,
+      "loss": 1.3264,
+      "step": 2064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.156629116117852e-06,
+      "loss": 1.4542,
+      "step": 2065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.155868914447772e-06,
+      "loss": 1.2679,
+      "step": 2066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.155108712777692e-06,
+      "loss": 1.3412,
+      "step": 2067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.154348511107611e-06,
+      "loss": 1.5138,
+      "step": 2068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.15358830943753e-06,
+      "loss": 1.3104,
+      "step": 2069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.15282810776745e-06,
+      "loss": 1.6975,
+      "step": 2070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.15206790609737e-06,
+      "loss": 1.2239,
+      "step": 2071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.151307704427289e-06,
+      "loss": 1.3312,
+      "step": 2072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.150547502757209e-06,
+      "loss": 1.439,
+      "step": 2073
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.149787301087129e-06,
+      "loss": 1.4009,
+      "step": 2074
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.149027099417049e-06,
+      "loss": 1.1069,
+      "step": 2075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.148266897746969e-06,
+      "loss": 1.4112,
+      "step": 2076
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.147506696076888e-06,
+      "loss": 1.2895,
+      "step": 2077
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.146746494406808e-06,
+      "loss": 1.4832,
+      "step": 2078
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.145986292736726e-06,
+      "loss": 1.2084,
+      "step": 2079
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.145226091066646e-06,
+      "loss": 1.2231,
+      "step": 2080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.144465889396566e-06,
+      "loss": 1.2608,
+      "step": 2081
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.143705687726486e-06,
+      "loss": 1.2016,
+      "step": 2082
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.142945486056406e-06,
+      "loss": 1.3037,
+      "step": 2083
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.142185284386326e-06,
+      "loss": 1.4702,
+      "step": 2084
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.141425082716244e-06,
+      "loss": 1.388,
+      "step": 2085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.140664881046164e-06,
+      "loss": 1.2988,
+      "step": 2086
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.139904679376084e-06,
+      "loss": 1.3411,
+      "step": 2087
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.139144477706003e-06,
+      "loss": 1.318,
+      "step": 2088
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.138384276035923e-06,
+      "loss": 1.3231,
+      "step": 2089
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.137624074365843e-06,
+      "loss": 1.285,
+      "step": 2090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.136863872695763e-06,
+      "loss": 1.3605,
+      "step": 2091
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.136103671025683e-06,
+      "loss": 1.2947,
+      "step": 2092
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.135343469355603e-06,
+      "loss": 1.1917,
+      "step": 2093
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.134583267685523e-06,
+      "loss": 1.0927,
+      "step": 2094
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.13382306601544e-06,
+      "loss": 1.2348,
+      "step": 2095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.13306286434536e-06,
+      "loss": 1.3597,
+      "step": 2096
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.13230266267528e-06,
+      "loss": 1.4607,
+      "step": 2097
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.1315424610052e-06,
+      "loss": 1.4654,
+      "step": 2098
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.13078225933512e-06,
+      "loss": 1.6892,
+      "step": 2099
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.13002205766504e-06,
+      "loss": 1.3399,
+      "step": 2100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.129261855994958e-06,
+      "loss": 1.3446,
+      "step": 2101
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.128501654324878e-06,
+      "loss": 1.5882,
+      "step": 2102
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.127741452654798e-06,
+      "loss": 1.3942,
+      "step": 2103
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.126981250984718e-06,
+      "loss": 1.3377,
+      "step": 2104
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.126221049314638e-06,
+      "loss": 1.3468,
+      "step": 2105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.125460847644557e-06,
+      "loss": 1.3988,
+      "step": 2106
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.124700645974477e-06,
+      "loss": 1.3281,
+      "step": 2107
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.123940444304397e-06,
+      "loss": 1.3629,
+      "step": 2108
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.123180242634317e-06,
+      "loss": 1.2755,
+      "step": 2109
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.122420040964237e-06,
+      "loss": 1.34,
+      "step": 2110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.121659839294155e-06,
+      "loss": 1.3882,
+      "step": 2111
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.120899637624075e-06,
+      "loss": 1.3837,
+      "step": 2112
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.120139435953995e-06,
+      "loss": 1.3754,
+      "step": 2113
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.119379234283915e-06,
+      "loss": 1.3098,
+      "step": 2114
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.118619032613834e-06,
+      "loss": 1.433,
+      "step": 2115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.117858830943754e-06,
+      "loss": 1.1674,
+      "step": 2116
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.117098629273672e-06,
+      "loss": 1.4731,
+      "step": 2117
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.116338427603592e-06,
+      "loss": 1.6077,
+      "step": 2118
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.115578225933512e-06,
+      "loss": 1.2814,
+      "step": 2119
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.114818024263432e-06,
+      "loss": 1.3438,
+      "step": 2120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.114057822593352e-06,
+      "loss": 1.329,
+      "step": 2121
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.113297620923272e-06,
+      "loss": 1.5597,
+      "step": 2122
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.112537419253192e-06,
+      "loss": 1.3584,
+      "step": 2123
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.111777217583111e-06,
+      "loss": 1.433,
+      "step": 2124
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.111017015913031e-06,
+      "loss": 1.1786,
+      "step": 2125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.110256814242951e-06,
+      "loss": 1.3013,
+      "step": 2126
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.10949661257287e-06,
+      "loss": 1.2978,
+      "step": 2127
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.10873641090279e-06,
+      "loss": 1.2728,
+      "step": 2128
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.107976209232709e-06,
+      "loss": 1.2089,
+      "step": 2129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.107216007562629e-06,
+      "loss": 1.439,
+      "step": 2130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.106455805892549e-06,
+      "loss": 1.4377,
+      "step": 2131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.105695604222469e-06,
+      "loss": 1.1394,
+      "step": 2132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.104935402552387e-06,
+      "loss": 1.1947,
+      "step": 2133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.104175200882307e-06,
+      "loss": 1.3558,
+      "step": 2134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.103414999212226e-06,
+      "loss": 1.2515,
+      "step": 2135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.102654797542146e-06,
+      "loss": 1.2297,
+      "step": 2136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.101894595872066e-06,
+      "loss": 1.3478,
+      "step": 2137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.101134394201986e-06,
+      "loss": 1.2761,
+      "step": 2138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.100374192531906e-06,
+      "loss": 1.2644,
+      "step": 2139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.099613990861826e-06,
+      "loss": 1.5993,
+      "step": 2140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.098853789191746e-06,
+      "loss": 1.254,
+      "step": 2141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.098093587521665e-06,
+      "loss": 1.1557,
+      "step": 2142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.097333385851584e-06,
+      "loss": 1.3499,
+      "step": 2143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.096573184181503e-06,
+      "loss": 1.4913,
+      "step": 2144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.095812982511423e-06,
+      "loss": 1.2711,
+      "step": 2145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.095052780841343e-06,
+      "loss": 1.2033,
+      "step": 2146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.094292579171263e-06,
+      "loss": 1.4673,
+      "step": 2147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.093532377501183e-06,
+      "loss": 1.2509,
+      "step": 2148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.092772175831101e-06,
+      "loss": 1.3839,
+      "step": 2149
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.092011974161021e-06,
+      "loss": 1.4052,
+      "step": 2150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.09125177249094e-06,
+      "loss": 1.5396,
+      "step": 2151
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.09049157082086e-06,
+      "loss": 1.3151,
+      "step": 2152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.08973136915078e-06,
+      "loss": 1.0761,
+      "step": 2153
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.0889711674807e-06,
+      "loss": 1.2945,
+      "step": 2154
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.08821096581062e-06,
+      "loss": 1.462,
+      "step": 2155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.08745076414054e-06,
+      "loss": 1.3117,
+      "step": 2156
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.08669056247046e-06,
+      "loss": 1.3485,
+      "step": 2157
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.08593036080038e-06,
+      "loss": 1.4444,
+      "step": 2158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.085170159130298e-06,
+      "loss": 1.2051,
+      "step": 2159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.084409957460218e-06,
+      "loss": 1.3392,
+      "step": 2160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.083649755790138e-06,
+      "loss": 1.3428,
+      "step": 2161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.082889554120058e-06,
+      "loss": 1.2396,
+      "step": 2162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.082129352449977e-06,
+      "loss": 1.3442,
+      "step": 2163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.081369150779897e-06,
+      "loss": 1.4279,
+      "step": 2164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.080608949109815e-06,
+      "loss": 1.1699,
+      "step": 2165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.079848747439735e-06,
+      "loss": 1.4042,
+      "step": 2166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.079088545769655e-06,
+      "loss": 1.27,
+      "step": 2167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.078328344099575e-06,
+      "loss": 1.6316,
+      "step": 2168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.077568142429495e-06,
+      "loss": 1.3348,
+      "step": 2169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.076807940759415e-06,
+      "loss": 1.0702,
+      "step": 2170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.076047739089335e-06,
+      "loss": 1.2604,
+      "step": 2171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.075287537419254e-06,
+      "loss": 1.3183,
+      "step": 2172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.074527335749174e-06,
+      "loss": 1.5163,
+      "step": 2173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.073767134079094e-06,
+      "loss": 1.4735,
+      "step": 2174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.073006932409012e-06,
+      "loss": 1.292,
+      "step": 2175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.072246730738932e-06,
+      "loss": 1.402,
+      "step": 2176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.071486529068852e-06,
+      "loss": 1.3227,
+      "step": 2177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.070726327398772e-06,
+      "loss": 1.4059,
+      "step": 2178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.069966125728692e-06,
+      "loss": 1.4191,
+      "step": 2179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.069205924058612e-06,
+      "loss": 1.2121,
+      "step": 2180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.06844572238853e-06,
+      "loss": 1.1055,
+      "step": 2181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.06768552071845e-06,
+      "loss": 1.3191,
+      "step": 2182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.06692531904837e-06,
+      "loss": 1.5599,
+      "step": 2183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.06616511737829e-06,
+      "loss": 1.32,
+      "step": 2184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.065404915708209e-06,
+      "loss": 1.7162,
+      "step": 2185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.064644714038129e-06,
+      "loss": 1.4288,
+      "step": 2186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.063884512368049e-06,
+      "loss": 1.4018,
+      "step": 2187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.063124310697969e-06,
+      "loss": 1.5241,
+      "step": 2188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.062364109027889e-06,
+      "loss": 1.3979,
+      "step": 2189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.061603907357808e-06,
+      "loss": 1.303,
+      "step": 2190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.060843705687727e-06,
+      "loss": 1.2826,
+      "step": 2191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.060083504017646e-06,
+      "loss": 1.5239,
+      "step": 2192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.059323302347566e-06,
+      "loss": 1.3978,
+      "step": 2193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.058563100677486e-06,
+      "loss": 1.2512,
+      "step": 2194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.057802899007406e-06,
+      "loss": 1.2974,
+      "step": 2195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.057042697337326e-06,
+      "loss": 1.4006,
+      "step": 2196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.056282495667244e-06,
+      "loss": 1.2717,
+      "step": 2197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.055522293997164e-06,
+      "loss": 1.1899,
+      "step": 2198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.054762092327084e-06,
+      "loss": 1.1517,
+      "step": 2199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.054001890657004e-06,
+      "loss": 1.382,
+      "step": 2200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.053241688986923e-06,
+      "loss": 1.2289,
+      "step": 2201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.052481487316843e-06,
+      "loss": 1.31,
+      "step": 2202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.051721285646763e-06,
+      "loss": 1.3021,
+      "step": 2203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.050961083976683e-06,
+      "loss": 1.3795,
+      "step": 2204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.050200882306603e-06,
+      "loss": 1.4557,
+      "step": 2205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.049440680636523e-06,
+      "loss": 1.4454,
+      "step": 2206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.048680478966441e-06,
+      "loss": 1.1719,
+      "step": 2207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.04792027729636e-06,
+      "loss": 1.446,
+      "step": 2208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.04716007562628e-06,
+      "loss": 1.3886,
+      "step": 2209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.0463998739562e-06,
+      "loss": 1.3273,
+      "step": 2210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.04563967228612e-06,
+      "loss": 1.2794,
+      "step": 2211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.04487947061604e-06,
+      "loss": 1.3093,
+      "step": 2212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.044119268945958e-06,
+      "loss": 1.269,
+      "step": 2213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.043359067275878e-06,
+      "loss": 1.219,
+      "step": 2214
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.042598865605798e-06,
+      "loss": 1.5066,
+      "step": 2215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.041838663935718e-06,
+      "loss": 1.4388,
+      "step": 2216
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.041078462265638e-06,
+      "loss": 1.6323,
+      "step": 2217
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.040318260595558e-06,
+      "loss": 1.2441,
+      "step": 2218
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.039558058925477e-06,
+      "loss": 1.3962,
+      "step": 2219
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.038797857255397e-06,
+      "loss": 1.3491,
+      "step": 2220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.038037655585317e-06,
+      "loss": 1.3307,
+      "step": 2221
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.037277453915237e-06,
+      "loss": 1.3208,
+      "step": 2222
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.036517252245155e-06,
+      "loss": 1.5717,
+      "step": 2223
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.035757050575075e-06,
+      "loss": 1.2714,
+      "step": 2224
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.034996848904995e-06,
+      "loss": 1.5189,
+      "step": 2225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.034236647234915e-06,
+      "loss": 1.5161,
+      "step": 2226
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.033476445564835e-06,
+      "loss": 1.6292,
+      "step": 2227
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.032716243894754e-06,
+      "loss": 1.581,
+      "step": 2228
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.031956042224674e-06,
+      "loss": 1.1936,
+      "step": 2229
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.031195840554594e-06,
+      "loss": 1.2373,
+      "step": 2230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.030435638884512e-06,
+      "loss": 1.3647,
+      "step": 2231
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.029675437214432e-06,
+      "loss": 1.334,
+      "step": 2232
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.028915235544352e-06,
+      "loss": 1.1152,
+      "step": 2233
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.028155033874272e-06,
+      "loss": 1.4677,
+      "step": 2234
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.027394832204192e-06,
+      "loss": 1.4028,
+      "step": 2235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.026634630534112e-06,
+      "loss": 1.2304,
+      "step": 2236
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.025874428864031e-06,
+      "loss": 1.2266,
+      "step": 2237
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.025114227193951e-06,
+      "loss": 1.4456,
+      "step": 2238
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.02435402552387e-06,
+      "loss": 1.3732,
+      "step": 2239
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.02359382385379e-06,
+      "loss": 1.1488,
+      "step": 2240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.02283362218371e-06,
+      "loss": 1.6584,
+      "step": 2241
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.022073420513629e-06,
+      "loss": 1.3177,
+      "step": 2242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.021313218843549e-06,
+      "loss": 1.5137,
+      "step": 2243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.020553017173469e-06,
+      "loss": 1.4068,
+      "step": 2244
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.019792815503389e-06,
+      "loss": 1.3244,
+      "step": 2245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.019032613833308e-06,
+      "loss": 1.4268,
+      "step": 2246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.018272412163228e-06,
+      "loss": 1.4272,
+      "step": 2247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.017512210493148e-06,
+      "loss": 1.3452,
+      "step": 2248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.016752008823068e-06,
+      "loss": 1.4644,
+      "step": 2249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.015991807152986e-06,
+      "loss": 1.3256,
+      "step": 2250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.015231605482906e-06,
+      "loss": 1.3451,
+      "step": 2251
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.014471403812826e-06,
+      "loss": 1.2984,
+      "step": 2252
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.013711202142746e-06,
+      "loss": 1.3927,
+      "step": 2253
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.012951000472666e-06,
+      "loss": 1.4865,
+      "step": 2254
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.012190798802584e-06,
+      "loss": 1.4414,
+      "step": 2255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.011430597132504e-06,
+      "loss": 1.6368,
+      "step": 2256
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.010670395462423e-06,
+      "loss": 1.4781,
+      "step": 2257
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.009910193792343e-06,
+      "loss": 1.2476,
+      "step": 2258
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.009149992122263e-06,
+      "loss": 1.3293,
+      "step": 2259
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.008389790452183e-06,
+      "loss": 1.2515,
+      "step": 2260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.007629588782103e-06,
+      "loss": 1.2791,
+      "step": 2261
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.006869387112023e-06,
+      "loss": 1.3868,
+      "step": 2262
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.006109185441943e-06,
+      "loss": 1.3432,
+      "step": 2263
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.005348983771862e-06,
+      "loss": 1.4526,
+      "step": 2264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.004588782101782e-06,
+      "loss": 1.2828,
+      "step": 2265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.0038285804317e-06,
+      "loss": 1.3324,
+      "step": 2266
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.00306837876162e-06,
+      "loss": 1.4084,
+      "step": 2267
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.00230817709154e-06,
+      "loss": 1.5327,
+      "step": 2268
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.00154797542146e-06,
+      "loss": 1.1757,
+      "step": 2269
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.00078777375138e-06,
+      "loss": 1.3464,
+      "step": 2270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.000027572081298e-06,
+      "loss": 1.387,
+      "step": 2271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.999267370411218e-06,
+      "loss": 1.2764,
+      "step": 2272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.998507168741138e-06,
+      "loss": 1.3416,
+      "step": 2273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.997746967071058e-06,
+      "loss": 1.511,
+      "step": 2274
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.996986765400978e-06,
+      "loss": 1.3034,
+      "step": 2275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.996226563730897e-06,
+      "loss": 1.3956,
+      "step": 2276
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.995466362060817e-06,
+      "loss": 1.3372,
+      "step": 2277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.994706160390737e-06,
+      "loss": 1.4131,
+      "step": 2278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.993945958720657e-06,
+      "loss": 1.4306,
+      "step": 2279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.993185757050577e-06,
+      "loss": 1.0532,
+      "step": 2280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.992425555380497e-06,
+      "loss": 1.2955,
+      "step": 2281
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.991665353710415e-06,
+      "loss": 1.3289,
+      "step": 2282
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.990905152040335e-06,
+      "loss": 1.106,
+      "step": 2283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.990144950370255e-06,
+      "loss": 1.4762,
+      "step": 2284
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.989384748700174e-06,
+      "loss": 1.4986,
+      "step": 2285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.988624547030094e-06,
+      "loss": 1.0984,
+      "step": 2286
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.987864345360012e-06,
+      "loss": 1.2115,
+      "step": 2287
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.987104143689932e-06,
+      "loss": 1.3522,
+      "step": 2288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.986343942019852e-06,
+      "loss": 1.2539,
+      "step": 2289
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.985583740349772e-06,
+      "loss": 0.9926,
+      "step": 2290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.984823538679692e-06,
+      "loss": 1.5445,
+      "step": 2291
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.984063337009612e-06,
+      "loss": 1.2201,
+      "step": 2292
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.983303135339532e-06,
+      "loss": 1.1739,
+      "step": 2293
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.982542933669451e-06,
+      "loss": 1.5071,
+      "step": 2294
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.981782731999371e-06,
+      "loss": 1.3283,
+      "step": 2295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.981022530329291e-06,
+      "loss": 1.3116,
+      "step": 2296
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.980262328659211e-06,
+      "loss": 1.3047,
+      "step": 2297
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.979502126989129e-06,
+      "loss": 1.5149,
+      "step": 2298
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.978741925319049e-06,
+      "loss": 1.2594,
+      "step": 2299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.977981723648969e-06,
+      "loss": 1.2243,
+      "step": 2300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.977221521978889e-06,
+      "loss": 1.3056,
+      "step": 2301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.976461320308809e-06,
+      "loss": 1.4033,
+      "step": 2302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.975701118638727e-06,
+      "loss": 1.2511,
+      "step": 2303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.974940916968647e-06,
+      "loss": 1.2935,
+      "step": 2304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.974180715298566e-06,
+      "loss": 1.2304,
+      "step": 2305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.973420513628486e-06,
+      "loss": 1.421,
+      "step": 2306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.972660311958406e-06,
+      "loss": 1.4561,
+      "step": 2307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.971900110288326e-06,
+      "loss": 1.3109,
+      "step": 2308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.971139908618246e-06,
+      "loss": 1.3937,
+      "step": 2309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.970379706948166e-06,
+      "loss": 1.5565,
+      "step": 2310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.969619505278086e-06,
+      "loss": 1.4695,
+      "step": 2311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.968859303608005e-06,
+      "loss": 1.4785,
+      "step": 2312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.968099101937925e-06,
+      "loss": 1.4231,
+      "step": 2313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.967338900267843e-06,
+      "loss": 1.2802,
+      "step": 2314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.966578698597763e-06,
+      "loss": 1.3007,
+      "step": 2315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.965818496927683e-06,
+      "loss": 1.2284,
+      "step": 2316
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.965058295257603e-06,
+      "loss": 1.3431,
+      "step": 2317
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.964298093587523e-06,
+      "loss": 1.3909,
+      "step": 2318
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.963537891917441e-06,
+      "loss": 1.5728,
+      "step": 2319
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.962777690247361e-06,
+      "loss": 1.4311,
+      "step": 2320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.96201748857728e-06,
+      "loss": 1.1652,
+      "step": 2321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.9612572869072e-06,
+      "loss": 1.4018,
+      "step": 2322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.96049708523712e-06,
+      "loss": 1.1765,
+      "step": 2323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.95973688356704e-06,
+      "loss": 1.3084,
+      "step": 2324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.95897668189696e-06,
+      "loss": 1.5093,
+      "step": 2325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.95821648022688e-06,
+      "loss": 1.2778,
+      "step": 2326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.9574562785568e-06,
+      "loss": 1.2938,
+      "step": 2327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.95669607688672e-06,
+      "loss": 1.4622,
+      "step": 2328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.95593587521664e-06,
+      "loss": 1.4456,
+      "step": 2329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.955175673546558e-06,
+      "loss": 1.2655,
+      "step": 2330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.954415471876478e-06,
+      "loss": 1.2761,
+      "step": 2331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.953655270206397e-06,
+      "loss": 1.3942,
+      "step": 2332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.952895068536317e-06,
+      "loss": 1.3552,
+      "step": 2333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.952134866866237e-06,
+      "loss": 1.3534,
+      "step": 2334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.951374665196155e-06,
+      "loss": 1.3849,
+      "step": 2335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.950614463526075e-06,
+      "loss": 1.2964,
+      "step": 2336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.949854261855995e-06,
+      "loss": 1.5687,
+      "step": 2337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.949094060185915e-06,
+      "loss": 1.4965,
+      "step": 2338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.948333858515835e-06,
+      "loss": 1.2729,
+      "step": 2339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.947573656845755e-06,
+      "loss": 1.3657,
+      "step": 2340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.946813455175674e-06,
+      "loss": 1.3248,
+      "step": 2341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.946053253505594e-06,
+      "loss": 1.2444,
+      "step": 2342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.945293051835514e-06,
+      "loss": 1.3115,
+      "step": 2343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.944532850165434e-06,
+      "loss": 1.3636,
+      "step": 2344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.943772648495354e-06,
+      "loss": 1.2165,
+      "step": 2345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.943012446825272e-06,
+      "loss": 1.2839,
+      "step": 2346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.942252245155192e-06,
+      "loss": 1.3041,
+      "step": 2347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.941492043485112e-06,
+      "loss": 1.6134,
+      "step": 2348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.940731841815032e-06,
+      "loss": 1.5113,
+      "step": 2349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.939971640144951e-06,
+      "loss": 1.2437,
+      "step": 2350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.93921143847487e-06,
+      "loss": 1.4031,
+      "step": 2351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.93845123680479e-06,
+      "loss": 1.301,
+      "step": 2352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.93769103513471e-06,
+      "loss": 1.2701,
+      "step": 2353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.93693083346463e-06,
+      "loss": 1.4304,
+      "step": 2354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.936170631794549e-06,
+      "loss": 1.3769,
+      "step": 2355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.935410430124469e-06,
+      "loss": 1.3486,
+      "step": 2356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.934650228454389e-06,
+      "loss": 1.3639,
+      "step": 2357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.933890026784309e-06,
+      "loss": 1.2784,
+      "step": 2358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.933129825114228e-06,
+      "loss": 1.2615,
+      "step": 2359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.932369623444148e-06,
+      "loss": 1.2911,
+      "step": 2360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.931609421774068e-06,
+      "loss": 1.1983,
+      "step": 2361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.930849220103986e-06,
+      "loss": 1.4157,
+      "step": 2362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.930089018433906e-06,
+      "loss": 1.3367,
+      "step": 2363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.929328816763826e-06,
+      "loss": 1.4842,
+      "step": 2364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.928568615093746e-06,
+      "loss": 1.5366,
+      "step": 2365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.927808413423666e-06,
+      "loss": 1.4457,
+      "step": 2366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.927048211753584e-06,
+      "loss": 1.4688,
+      "step": 2367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.926288010083504e-06,
+      "loss": 1.6681,
+      "step": 2368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.925527808413424e-06,
+      "loss": 1.0182,
+      "step": 2369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.924767606743344e-06,
+      "loss": 1.3122,
+      "step": 2370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.924007405073263e-06,
+      "loss": 1.2763,
+      "step": 2371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.923247203403183e-06,
+      "loss": 1.496,
+      "step": 2372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.922487001733103e-06,
+      "loss": 1.2319,
+      "step": 2373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.921726800063023e-06,
+      "loss": 1.3748,
+      "step": 2374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.920966598392943e-06,
+      "loss": 1.2189,
+      "step": 2375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.920206396722863e-06,
+      "loss": 1.3445,
+      "step": 2376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.919446195052783e-06,
+      "loss": 1.3546,
+      "step": 2377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.9186859933827e-06,
+      "loss": 1.401,
+      "step": 2378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.91792579171262e-06,
+      "loss": 1.1836,
+      "step": 2379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.91716559004254e-06,
+      "loss": 1.1419,
+      "step": 2380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.91640538837246e-06,
+      "loss": 1.4476,
+      "step": 2381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.91564518670238e-06,
+      "loss": 1.212,
+      "step": 2382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.914884985032298e-06,
+      "loss": 1.3672,
+      "step": 2383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.914124783362218e-06,
+      "loss": 1.4012,
+      "step": 2384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.913364581692138e-06,
+      "loss": 1.283,
+      "step": 2385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.912604380022058e-06,
+      "loss": 1.6359,
+      "step": 2386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.911844178351978e-06,
+      "loss": 1.348,
+      "step": 2387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.911083976681898e-06,
+      "loss": 1.339,
+      "step": 2388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.910323775011817e-06,
+      "loss": 1.6808,
+      "step": 2389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.909563573341737e-06,
+      "loss": 1.3179,
+      "step": 2390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.908803371671657e-06,
+      "loss": 1.1662,
+      "step": 2391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.908043170001577e-06,
+      "loss": 1.3276,
+      "step": 2392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.907282968331497e-06,
+      "loss": 1.3718,
+      "step": 2393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.906522766661415e-06,
+      "loss": 1.3077,
+      "step": 2394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.905762564991335e-06,
+      "loss": 1.2985,
+      "step": 2395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.905002363321255e-06,
+      "loss": 1.1781,
+      "step": 2396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.904242161651175e-06,
+      "loss": 1.3839,
+      "step": 2397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.903481959981094e-06,
+      "loss": 1.3962,
+      "step": 2398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.902721758311013e-06,
+      "loss": 1.2318,
+      "step": 2399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.901961556640932e-06,
+      "loss": 1.2529,
+      "step": 2400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.901201354970852e-06,
+      "loss": 1.1572,
+      "step": 2401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.900441153300772e-06,
+      "loss": 1.297,
+      "step": 2402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.899680951630692e-06,
+      "loss": 1.3336,
+      "step": 2403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.898920749960612e-06,
+      "loss": 1.4143,
+      "step": 2404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.898160548290532e-06,
+      "loss": 1.3526,
+      "step": 2405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.897400346620452e-06,
+      "loss": 1.1937,
+      "step": 2406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.896640144950371e-06,
+      "loss": 1.2784,
+      "step": 2407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.895879943280291e-06,
+      "loss": 1.493,
+      "step": 2408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.895119741610211e-06,
+      "loss": 1.3569,
+      "step": 2409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.89435953994013e-06,
+      "loss": 1.3307,
+      "step": 2410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.893599338270049e-06,
+      "loss": 1.3295,
+      "step": 2411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.892839136599969e-06,
+      "loss": 1.525,
+      "step": 2412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.892078934929889e-06,
+      "loss": 1.3493,
+      "step": 2413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.891318733259809e-06,
+      "loss": 1.4703,
+      "step": 2414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.890558531589729e-06,
+      "loss": 1.1709,
+      "step": 2415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.889798329919648e-06,
+      "loss": 1.2194,
+      "step": 2416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.889038128249568e-06,
+      "loss": 1.3474,
+      "step": 2417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.888277926579488e-06,
+      "loss": 1.2514,
+      "step": 2418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.887517724909406e-06,
+      "loss": 1.4444,
+      "step": 2419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.886757523239326e-06,
+      "loss": 1.3715,
+      "step": 2420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.885997321569246e-06,
+      "loss": 1.4126,
+      "step": 2421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.885237119899166e-06,
+      "loss": 1.4036,
+      "step": 2422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.884476918229086e-06,
+      "loss": 1.1463,
+      "step": 2423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.883716716559006e-06,
+      "loss": 1.5627,
+      "step": 2424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.882956514888925e-06,
+      "loss": 1.3087,
+      "step": 2425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.882196313218844e-06,
+      "loss": 1.3904,
+      "step": 2426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.881436111548763e-06,
+      "loss": 1.535,
+      "step": 2427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.880675909878683e-06,
+      "loss": 1.4751,
+      "step": 2428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.879915708208603e-06,
+      "loss": 1.2503,
+      "step": 2429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.879155506538523e-06,
+      "loss": 1.2956,
+      "step": 2430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.878395304868443e-06,
+      "loss": 1.1707,
+      "step": 2431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.877635103198363e-06,
+      "loss": 1.3444,
+      "step": 2432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.876874901528283e-06,
+      "loss": 1.1624,
+      "step": 2433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.876114699858202e-06,
+      "loss": 1.2499,
+      "step": 2434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.875354498188122e-06,
+      "loss": 1.2753,
+      "step": 2435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.87459429651804e-06,
+      "loss": 1.4705,
+      "step": 2436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.87383409484796e-06,
+      "loss": 1.3876,
+      "step": 2437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.87307389317788e-06,
+      "loss": 1.2711,
+      "step": 2438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.8723136915078e-06,
+      "loss": 1.5086,
+      "step": 2439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.87155348983772e-06,
+      "loss": 1.2649,
+      "step": 2440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.870793288167638e-06,
+      "loss": 1.3488,
+      "step": 2441
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.870033086497558e-06,
+      "loss": 1.46,
+      "step": 2442
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.869272884827478e-06,
+      "loss": 1.1858,
+      "step": 2443
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.868512683157398e-06,
+      "loss": 1.2729,
+      "step": 2444
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.867752481487317e-06,
+      "loss": 1.3902,
+      "step": 2445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.866992279817237e-06,
+      "loss": 1.4781,
+      "step": 2446
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.866232078147157e-06,
+      "loss": 1.2817,
+      "step": 2447
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.865471876477077e-06,
+      "loss": 1.3124,
+      "step": 2448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.864711674806997e-06,
+      "loss": 1.2948,
+      "step": 2449
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.863951473136917e-06,
+      "loss": 1.5763,
+      "step": 2450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.863191271466837e-06,
+      "loss": 1.324,
+      "step": 2451
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.862431069796755e-06,
+      "loss": 1.4309,
+      "step": 2452
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.861670868126675e-06,
+      "loss": 1.3012,
+      "step": 2453
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.860910666456594e-06,
+      "loss": 1.1049,
+      "step": 2454
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.860150464786514e-06,
+      "loss": 1.3869,
+      "step": 2455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.859390263116434e-06,
+      "loss": 1.4639,
+      "step": 2456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.858630061446352e-06,
+      "loss": 1.3115,
+      "step": 2457
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.857869859776272e-06,
+      "loss": 1.182,
+      "step": 2458
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.857109658106192e-06,
+      "loss": 1.3372,
+      "step": 2459
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.856349456436112e-06,
+      "loss": 1.4519,
+      "step": 2460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.855589254766032e-06,
+      "loss": 1.2971,
+      "step": 2461
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.854829053095952e-06,
+      "loss": 1.1372,
+      "step": 2462
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.854068851425871e-06,
+      "loss": 1.4217,
+      "step": 2463
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.853308649755791e-06,
+      "loss": 1.2401,
+      "step": 2464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.852548448085711e-06,
+      "loss": 1.4927,
+      "step": 2465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.851788246415631e-06,
+      "loss": 1.4529,
+      "step": 2466
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.851028044745551e-06,
+      "loss": 1.3744,
+      "step": 2467
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.850267843075469e-06,
+      "loss": 1.4589,
+      "step": 2468
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.849507641405389e-06,
+      "loss": 1.2257,
+      "step": 2469
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.848747439735309e-06,
+      "loss": 1.1384,
+      "step": 2470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.847987238065229e-06,
+      "loss": 1.1312,
+      "step": 2471
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.847227036395148e-06,
+      "loss": 1.3309,
+      "step": 2472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.846466834725067e-06,
+      "loss": 1.4542,
+      "step": 2473
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.845706633054987e-06,
+      "loss": 1.2598,
+      "step": 2474
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.844946431384906e-06,
+      "loss": 1.3932,
+      "step": 2475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.844186229714826e-06,
+      "loss": 1.4368,
+      "step": 2476
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.843426028044746e-06,
+      "loss": 1.4433,
+      "step": 2477
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.842665826374666e-06,
+      "loss": 1.3076,
+      "step": 2478
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.841905624704586e-06,
+      "loss": 1.324,
+      "step": 2479
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.841145423034506e-06,
+      "loss": 1.3075,
+      "step": 2480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.840385221364426e-06,
+      "loss": 1.3565,
+      "step": 2481
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.839625019694345e-06,
+      "loss": 1.3433,
+      "step": 2482
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.838864818024265e-06,
+      "loss": 1.2788,
+      "step": 2483
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.838104616354183e-06,
+      "loss": 1.24,
+      "step": 2484
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.837344414684103e-06,
+      "loss": 1.3086,
+      "step": 2485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.836584213014023e-06,
+      "loss": 1.3005,
+      "step": 2486
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.835824011343943e-06,
+      "loss": 1.4353,
+      "step": 2487
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.835063809673863e-06,
+      "loss": 1.4588,
+      "step": 2488
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.834303608003781e-06,
+      "loss": 1.2476,
+      "step": 2489
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.8335434063337e-06,
+      "loss": 1.2125,
+      "step": 2490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.83278320466362e-06,
+      "loss": 1.5871,
+      "step": 2491
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.83202300299354e-06,
+      "loss": 1.3308,
+      "step": 2492
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.83126280132346e-06,
+      "loss": 1.1853,
+      "step": 2493
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.83050259965338e-06,
+      "loss": 1.4798,
+      "step": 2494
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.8297423979833e-06,
+      "loss": 1.3938,
+      "step": 2495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.82898219631322e-06,
+      "loss": 1.2477,
+      "step": 2496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.82822199464314e-06,
+      "loss": 1.39,
+      "step": 2497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.82746179297306e-06,
+      "loss": 1.3756,
+      "step": 2498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.82670159130298e-06,
+      "loss": 1.2937,
+      "step": 2499
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.825941389632898e-06,
+      "loss": 1.1713,
+      "step": 2500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.825181187962818e-06,
+      "loss": 1.3093,
+      "step": 2501
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.824420986292737e-06,
+      "loss": 1.2794,
+      "step": 2502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.823660784622657e-06,
+      "loss": 1.2925,
+      "step": 2503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.822900582952577e-06,
+      "loss": 1.4099,
+      "step": 2504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.822140381282495e-06,
+      "loss": 1.4382,
+      "step": 2505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.821380179612415e-06,
+      "loss": 1.2552,
+      "step": 2506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.820619977942335e-06,
+      "loss": 1.2068,
+      "step": 2507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.819859776272255e-06,
+      "loss": 1.441,
+      "step": 2508
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.819099574602175e-06,
+      "loss": 1.4253,
+      "step": 2509
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.818339372932095e-06,
+      "loss": 1.4259,
+      "step": 2510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.817579171262014e-06,
+      "loss": 1.2536,
+      "step": 2511
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.816818969591934e-06,
+      "loss": 1.568,
+      "step": 2512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.816058767921854e-06,
+      "loss": 1.1044,
+      "step": 2513
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.815298566251774e-06,
+      "loss": 1.4183,
+      "step": 2514
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.814538364581694e-06,
+      "loss": 1.3211,
+      "step": 2515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.813778162911612e-06,
+      "loss": 1.4865,
+      "step": 2516
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.813017961241532e-06,
+      "loss": 1.3647,
+      "step": 2517
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.812257759571452e-06,
+      "loss": 1.2236,
+      "step": 2518
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.811497557901372e-06,
+      "loss": 1.389,
+      "step": 2519
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.810737356231291e-06,
+      "loss": 1.4843,
+      "step": 2520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.80997715456121e-06,
+      "loss": 1.3713,
+      "step": 2521
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.80921695289113e-06,
+      "loss": 1.2978,
+      "step": 2522
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.80845675122105e-06,
+      "loss": 1.2945,
+      "step": 2523
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.807696549550969e-06,
+      "loss": 1.4143,
+      "step": 2524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.806936347880889e-06,
+      "loss": 1.3124,
+      "step": 2525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.806176146210809e-06,
+      "loss": 1.0617,
+      "step": 2526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.805415944540729e-06,
+      "loss": 1.4535,
+      "step": 2527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.804655742870649e-06,
+      "loss": 1.1995,
+      "step": 2528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.803895541200568e-06,
+      "loss": 1.3005,
+      "step": 2529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.803135339530488e-06,
+      "loss": 1.2008,
+      "step": 2530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.802375137860408e-06,
+      "loss": 1.1813,
+      "step": 2531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.801614936190326e-06,
+      "loss": 1.2506,
+      "step": 2532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.800854734520246e-06,
+      "loss": 1.1106,
+      "step": 2533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.800094532850166e-06,
+      "loss": 1.3265,
+      "step": 2534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.799334331180086e-06,
+      "loss": 1.2218,
+      "step": 2535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.798574129510006e-06,
+      "loss": 1.3512,
+      "step": 2536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.797813927839924e-06,
+      "loss": 1.4706,
+      "step": 2537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.797053726169844e-06,
+      "loss": 1.3859,
+      "step": 2538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.796293524499764e-06,
+      "loss": 1.1849,
+      "step": 2539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.795533322829683e-06,
+      "loss": 1.4227,
+      "step": 2540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.794773121159603e-06,
+      "loss": 1.3978,
+      "step": 2541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.794012919489523e-06,
+      "loss": 1.1189,
+      "step": 2542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.793252717819443e-06,
+      "loss": 1.4258,
+      "step": 2543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.792492516149363e-06,
+      "loss": 1.3791,
+      "step": 2544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.791732314479283e-06,
+      "loss": 1.3035,
+      "step": 2545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.790972112809203e-06,
+      "loss": 1.4289,
+      "step": 2546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.790211911139122e-06,
+      "loss": 1.3837,
+      "step": 2547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.78945170946904e-06,
+      "loss": 1.1891,
+      "step": 2548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.78869150779896e-06,
+      "loss": 1.2864,
+      "step": 2549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.78793130612888e-06,
+      "loss": 1.2995,
+      "step": 2550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.7871711044588e-06,
+      "loss": 1.2785,
+      "step": 2551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.78641090278872e-06,
+      "loss": 1.5416,
+      "step": 2552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.785650701118638e-06,
+      "loss": 1.3614,
+      "step": 2553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.784890499448558e-06,
+      "loss": 1.1882,
+      "step": 2554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.784130297778478e-06,
+      "loss": 1.2771,
+      "step": 2555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.783370096108398e-06,
+      "loss": 1.3582,
+      "step": 2556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.782609894438318e-06,
+      "loss": 1.5888,
+      "step": 2557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.781849692768237e-06,
+      "loss": 1.4255,
+      "step": 2558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.781089491098157e-06,
+      "loss": 1.1033,
+      "step": 2559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.780329289428077e-06,
+      "loss": 1.5372,
+      "step": 2560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.779569087757997e-06,
+      "loss": 1.3524,
+      "step": 2561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.778808886087917e-06,
+      "loss": 1.3127,
+      "step": 2562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.778048684417837e-06,
+      "loss": 1.3412,
+      "step": 2563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.777288482747755e-06,
+      "loss": 1.491,
+      "step": 2564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.776528281077675e-06,
+      "loss": 1.6197,
+      "step": 2565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.775768079407595e-06,
+      "loss": 1.512,
+      "step": 2566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.775007877737514e-06,
+      "loss": 1.2194,
+      "step": 2567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.774247676067434e-06,
+      "loss": 1.4061,
+      "step": 2568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.773487474397353e-06,
+      "loss": 1.1562,
+      "step": 2569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.772727272727272e-06,
+      "loss": 1.3291,
+      "step": 2570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.771967071057192e-06,
+      "loss": 1.2623,
+      "step": 2571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.771206869387112e-06,
+      "loss": 1.3093,
+      "step": 2572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.770446667717032e-06,
+      "loss": 1.4497,
+      "step": 2573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.769686466046952e-06,
+      "loss": 1.3728,
+      "step": 2574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.768926264376872e-06,
+      "loss": 1.4845,
+      "step": 2575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.768166062706791e-06,
+      "loss": 1.6585,
+      "step": 2576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.767405861036711e-06,
+      "loss": 1.0497,
+      "step": 2577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.766645659366631e-06,
+      "loss": 1.3488,
+      "step": 2578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.765885457696551e-06,
+      "loss": 1.1454,
+      "step": 2579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.76512525602647e-06,
+      "loss": 1.2475,
+      "step": 2580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.764365054356389e-06,
+      "loss": 1.2913,
+      "step": 2581
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.763604852686309e-06,
+      "loss": 1.1509,
+      "step": 2582
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.762844651016229e-06,
+      "loss": 1.432,
+      "step": 2583
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.762084449346149e-06,
+      "loss": 1.2555,
+      "step": 2584
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.761324247676067e-06,
+      "loss": 1.2766,
+      "step": 2585
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.760564046005987e-06,
+      "loss": 1.342,
+      "step": 2586
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.759803844335907e-06,
+      "loss": 1.4055,
+      "step": 2587
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.759043642665826e-06,
+      "loss": 1.2725,
+      "step": 2588
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.758283440995746e-06,
+      "loss": 1.2276,
+      "step": 2589
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.757523239325666e-06,
+      "loss": 1.3497,
+      "step": 2590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.756763037655586e-06,
+      "loss": 1.17,
+      "step": 2591
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.756002835985506e-06,
+      "loss": 1.5634,
+      "step": 2592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.755242634315426e-06,
+      "loss": 1.5979,
+      "step": 2593
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.754482432645346e-06,
+      "loss": 1.2595,
+      "step": 2594
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.753722230975265e-06,
+      "loss": 1.3776,
+      "step": 2595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.752962029305184e-06,
+      "loss": 1.2065,
+      "step": 2596
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.752201827635103e-06,
+      "loss": 1.2567,
+      "step": 2597
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.751441625965023e-06,
+      "loss": 1.3697,
+      "step": 2598
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.750681424294943e-06,
+      "loss": 1.1381,
+      "step": 2599
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.749921222624863e-06,
+      "loss": 1.3163,
+      "step": 2600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.749161020954783e-06,
+      "loss": 1.3876,
+      "step": 2601
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.748400819284703e-06,
+      "loss": 1.2658,
+      "step": 2602
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.747640617614623e-06,
+      "loss": 1.4627,
+      "step": 2603
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.746880415944542e-06,
+      "loss": 1.3213,
+      "step": 2604
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.746120214274462e-06,
+      "loss": 1.2965,
+      "step": 2605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.745360012604382e-06,
+      "loss": 1.2698,
+      "step": 2606
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.7445998109343e-06,
+      "loss": 1.1505,
+      "step": 2607
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.74383960926422e-06,
+      "loss": 1.2442,
+      "step": 2608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.74307940759414e-06,
+      "loss": 1.3999,
+      "step": 2609
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.74231920592406e-06,
+      "loss": 1.3767,
+      "step": 2610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.74155900425398e-06,
+      "loss": 1.2821,
+      "step": 2611
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.740798802583898e-06,
+      "loss": 1.2269,
+      "step": 2612
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.740038600913818e-06,
+      "loss": 1.157,
+      "step": 2613
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.739278399243738e-06,
+      "loss": 1.4477,
+      "step": 2614
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.738518197573657e-06,
+      "loss": 1.3135,
+      "step": 2615
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.737757995903577e-06,
+      "loss": 1.2194,
+      "step": 2616
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.736997794233497e-06,
+      "loss": 1.2843,
+      "step": 2617
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.736237592563417e-06,
+      "loss": 1.2151,
+      "step": 2618
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.735477390893337e-06,
+      "loss": 1.3308,
+      "step": 2619
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.734717189223257e-06,
+      "loss": 1.3091,
+      "step": 2620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.733956987553177e-06,
+      "loss": 1.388,
+      "step": 2621
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.733196785883096e-06,
+      "loss": 1.0773,
+      "step": 2622
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.732436584213015e-06,
+      "loss": 1.4438,
+      "step": 2623
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.731676382542934e-06,
+      "loss": 1.4501,
+      "step": 2624
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.730916180872854e-06,
+      "loss": 1.2841,
+      "step": 2625
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.730155979202774e-06,
+      "loss": 1.3972,
+      "step": 2626
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.729395777532694e-06,
+      "loss": 1.1332,
+      "step": 2627
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.728635575862612e-06,
+      "loss": 1.0225,
+      "step": 2628
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.727875374192532e-06,
+      "loss": 1.4306,
+      "step": 2629
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.727115172522452e-06,
+      "loss": 1.3233,
+      "step": 2630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.726354970852372e-06,
+      "loss": 1.2366,
+      "step": 2631
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.725594769182292e-06,
+      "loss": 1.3691,
+      "step": 2632
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.724834567512211e-06,
+      "loss": 1.3155,
+      "step": 2633
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.724074365842131e-06,
+      "loss": 1.3466,
+      "step": 2634
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.723314164172051e-06,
+      "loss": 1.323,
+      "step": 2635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.722553962501971e-06,
+      "loss": 1.3289,
+      "step": 2636
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.721793760831891e-06,
+      "loss": 1.3926,
+      "step": 2637
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.72103355916181e-06,
+      "loss": 1.4376,
+      "step": 2638
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.720273357491729e-06,
+      "loss": 1.4638,
+      "step": 2639
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.719513155821649e-06,
+      "loss": 1.3897,
+      "step": 2640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.718752954151569e-06,
+      "loss": 1.4143,
+      "step": 2641
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.717992752481488e-06,
+      "loss": 1.1983,
+      "step": 2642
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.717232550811408e-06,
+      "loss": 1.3563,
+      "step": 2643
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.716472349141326e-06,
+      "loss": 1.3799,
+      "step": 2644
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.715712147471246e-06,
+      "loss": 1.3038,
+      "step": 2645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.714951945801166e-06,
+      "loss": 1.1816,
+      "step": 2646
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.714191744131086e-06,
+      "loss": 1.4944,
+      "step": 2647
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.713431542461006e-06,
+      "loss": 1.3964,
+      "step": 2648
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.712671340790926e-06,
+      "loss": 1.374,
+      "step": 2649
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.711911139120846e-06,
+      "loss": 1.2101,
+      "step": 2650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.711150937450765e-06,
+      "loss": 1.4922,
+      "step": 2651
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.710390735780685e-06,
+      "loss": 1.453,
+      "step": 2652
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.709630534110605e-06,
+      "loss": 1.3888,
+      "step": 2653
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.708870332440523e-06,
+      "loss": 1.4241,
+      "step": 2654
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.708110130770443e-06,
+      "loss": 1.2427,
+      "step": 2655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.707349929100363e-06,
+      "loss": 1.2893,
+      "step": 2656
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.706589727430283e-06,
+      "loss": 1.282,
+      "step": 2657
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.705829525760203e-06,
+      "loss": 1.1242,
+      "step": 2658
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.705069324090123e-06,
+      "loss": 1.3997,
+      "step": 2659
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.70430912242004e-06,
+      "loss": 1.3606,
+      "step": 2660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.70354892074996e-06,
+      "loss": 1.2163,
+      "step": 2661
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.70278871907988e-06,
+      "loss": 1.4111,
+      "step": 2662
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.7020285174098e-06,
+      "loss": 1.2166,
+      "step": 2663
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.70126831573972e-06,
+      "loss": 1.2907,
+      "step": 2664
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.70050811406964e-06,
+      "loss": 1.3624,
+      "step": 2665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.69974791239956e-06,
+      "loss": 1.3018,
+      "step": 2666
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.69898771072948e-06,
+      "loss": 1.3919,
+      "step": 2667
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.6982275090594e-06,
+      "loss": 1.1591,
+      "step": 2668
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.69746730738932e-06,
+      "loss": 1.4574,
+      "step": 2669
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.696707105719238e-06,
+      "loss": 1.3215,
+      "step": 2670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.695946904049157e-06,
+      "loss": 1.4018,
+      "step": 2671
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.695186702379077e-06,
+      "loss": 1.354,
+      "step": 2672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.694426500708997e-06,
+      "loss": 1.3703,
+      "step": 2673
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.693666299038917e-06,
+      "loss": 1.3586,
+      "step": 2674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.692906097368837e-06,
+      "loss": 1.4316,
+      "step": 2675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.692145895698755e-06,
+      "loss": 1.3167,
+      "step": 2676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.691385694028675e-06,
+      "loss": 1.3799,
+      "step": 2677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.690625492358595e-06,
+      "loss": 1.278,
+      "step": 2678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.689865290688515e-06,
+      "loss": 1.2405,
+      "step": 2679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.689105089018434e-06,
+      "loss": 1.4142,
+      "step": 2680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.688344887348354e-06,
+      "loss": 1.3918,
+      "step": 2681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.687584685678274e-06,
+      "loss": 1.263,
+      "step": 2682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.686824484008194e-06,
+      "loss": 1.4972,
+      "step": 2683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.686064282338114e-06,
+      "loss": 1.252,
+      "step": 2684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.685304080668034e-06,
+      "loss": 1.2098,
+      "step": 2685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.684543878997952e-06,
+      "loss": 1.3323,
+      "step": 2686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.683783677327872e-06,
+      "loss": 1.2018,
+      "step": 2687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.683023475657792e-06,
+      "loss": 1.3811,
+      "step": 2688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.682263273987712e-06,
+      "loss": 1.5903,
+      "step": 2689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.681503072317631e-06,
+      "loss": 1.1603,
+      "step": 2690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.680742870647551e-06,
+      "loss": 1.291,
+      "step": 2691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.67998266897747e-06,
+      "loss": 1.1419,
+      "step": 2692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.67922246730739e-06,
+      "loss": 1.1448,
+      "step": 2693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.678462265637309e-06,
+      "loss": 1.2992,
+      "step": 2694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.677702063967229e-06,
+      "loss": 1.184,
+      "step": 2695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.676941862297149e-06,
+      "loss": 1.4001,
+      "step": 2696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.676181660627069e-06,
+      "loss": 1.5073,
+      "step": 2697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.675421458956989e-06,
+      "loss": 1.2569,
+      "step": 2698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.674661257286908e-06,
+      "loss": 1.2188,
+      "step": 2699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.673901055616828e-06,
+      "loss": 1.305,
+      "step": 2700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.673140853946748e-06,
+      "loss": 1.0829,
+      "step": 2701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.672380652276666e-06,
+      "loss": 1.1447,
+      "step": 2702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.671620450606586e-06,
+      "loss": 1.3548,
+      "step": 2703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.670860248936506e-06,
+      "loss": 1.1686,
+      "step": 2704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.670100047266426e-06,
+      "loss": 1.3193,
+      "step": 2705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.669339845596346e-06,
+      "loss": 1.1984,
+      "step": 2706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.668579643926266e-06,
+      "loss": 1.3327,
+      "step": 2707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.667819442256184e-06,
+      "loss": 1.0941,
+      "step": 2708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.667059240586104e-06,
+      "loss": 1.4636,
+      "step": 2709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.666299038916023e-06,
+      "loss": 1.2366,
+      "step": 2710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.665538837245943e-06,
+      "loss": 1.1553,
+      "step": 2711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.664778635575863e-06,
+      "loss": 1.308,
+      "step": 2712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.664018433905783e-06,
+      "loss": 1.1757,
+      "step": 2713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.663258232235703e-06,
+      "loss": 1.2003,
+      "step": 2714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.662498030565623e-06,
+      "loss": 1.3237,
+      "step": 2715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.661737828895543e-06,
+      "loss": 1.1776,
+      "step": 2716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.660977627225462e-06,
+      "loss": 1.2412,
+      "step": 2717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.66021742555538e-06,
+      "loss": 1.2567,
+      "step": 2718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.6594572238853e-06,
+      "loss": 1.2256,
+      "step": 2719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.65869702221522e-06,
+      "loss": 1.3353,
+      "step": 2720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.65793682054514e-06,
+      "loss": 1.3803,
+      "step": 2721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.65717661887506e-06,
+      "loss": 1.2455,
+      "step": 2722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.65641641720498e-06,
+      "loss": 1.4309,
+      "step": 2723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.655656215534898e-06,
+      "loss": 1.3679,
+      "step": 2724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.654896013864818e-06,
+      "loss": 1.2834,
+      "step": 2725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.654135812194738e-06,
+      "loss": 1.1233,
+      "step": 2726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.653375610524658e-06,
+      "loss": 1.3091,
+      "step": 2727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.652615408854577e-06,
+      "loss": 1.1477,
+      "step": 2728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.651855207184497e-06,
+      "loss": 1.4159,
+      "step": 2729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.651095005514417e-06,
+      "loss": 1.5237,
+      "step": 2730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.650334803844337e-06,
+      "loss": 1.1988,
+      "step": 2731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.649574602174257e-06,
+      "loss": 1.2857,
+      "step": 2732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.648814400504177e-06,
+      "loss": 1.1248,
+      "step": 2733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.648054198834095e-06,
+      "loss": 1.3468,
+      "step": 2734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.647293997164015e-06,
+      "loss": 1.4358,
+      "step": 2735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.646533795493935e-06,
+      "loss": 1.2319,
+      "step": 2736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.645773593823854e-06,
+      "loss": 1.3359,
+      "step": 2737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.645013392153774e-06,
+      "loss": 1.4101,
+      "step": 2738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.644253190483694e-06,
+      "loss": 1.1131,
+      "step": 2739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.643492988813612e-06,
+      "loss": 1.3552,
+      "step": 2740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.642732787143532e-06,
+      "loss": 1.1557,
+      "step": 2741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.641972585473452e-06,
+      "loss": 1.388,
+      "step": 2742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.641212383803372e-06,
+      "loss": 1.4294,
+      "step": 2743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.640452182133292e-06,
+      "loss": 1.1498,
+      "step": 2744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.639691980463212e-06,
+      "loss": 1.3625,
+      "step": 2745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.638931778793131e-06,
+      "loss": 1.2666,
+      "step": 2746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.638171577123051e-06,
+      "loss": 1.3487,
+      "step": 2747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.637411375452971e-06,
+      "loss": 1.5168,
+      "step": 2748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.636651173782891e-06,
+      "loss": 1.0977,
+      "step": 2749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.63589097211281e-06,
+      "loss": 1.5913,
+      "step": 2750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.635130770442729e-06,
+      "loss": 1.6156,
+      "step": 2751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.634370568772649e-06,
+      "loss": 1.4907,
+      "step": 2752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.633610367102569e-06,
+      "loss": 1.3067,
+      "step": 2753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.632850165432489e-06,
+      "loss": 1.2294,
+      "step": 2754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.632089963762408e-06,
+      "loss": 1.425,
+      "step": 2755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.631329762092327e-06,
+      "loss": 1.3106,
+      "step": 2756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.630569560422246e-06,
+      "loss": 1.2754,
+      "step": 2757
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.629809358752166e-06,
+      "loss": 1.1749,
+      "step": 2758
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.629049157082086e-06,
+      "loss": 1.4026,
+      "step": 2759
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.628288955412006e-06,
+      "loss": 1.2025,
+      "step": 2760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.627528753741926e-06,
+      "loss": 1.28,
+      "step": 2761
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.626768552071846e-06,
+      "loss": 1.2236,
+      "step": 2762
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.626008350401766e-06,
+      "loss": 1.317,
+      "step": 2763
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.625248148731685e-06,
+      "loss": 1.5175,
+      "step": 2764
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.624487947061605e-06,
+      "loss": 1.3281,
+      "step": 2765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.623727745391523e-06,
+      "loss": 1.353,
+      "step": 2766
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.622967543721443e-06,
+      "loss": 1.2175,
+      "step": 2767
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.622207342051363e-06,
+      "loss": 1.1498,
+      "step": 2768
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.621447140381283e-06,
+      "loss": 1.4528,
+      "step": 2769
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.620686938711203e-06,
+      "loss": 1.3722,
+      "step": 2770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.619926737041123e-06,
+      "loss": 1.302,
+      "step": 2771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.619166535371043e-06,
+      "loss": 1.1853,
+      "step": 2772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.61840633370096e-06,
+      "loss": 1.527,
+      "step": 2773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.61764613203088e-06,
+      "loss": 1.3912,
+      "step": 2774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.6168859303608e-06,
+      "loss": 1.3107,
+      "step": 2775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.61612572869072e-06,
+      "loss": 1.1323,
+      "step": 2776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.61536552702064e-06,
+      "loss": 1.257,
+      "step": 2777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.61460532535056e-06,
+      "loss": 1.3915,
+      "step": 2778
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.61384512368048e-06,
+      "loss": 1.1641,
+      "step": 2779
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.6130849220104e-06,
+      "loss": 1.2199,
+      "step": 2780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.61232472034032e-06,
+      "loss": 1.376,
+      "step": 2781
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.611564518670238e-06,
+      "loss": 1.3421,
+      "step": 2782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.610804317000158e-06,
+      "loss": 1.3995,
+      "step": 2783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.610044115330077e-06,
+      "loss": 1.2517,
+      "step": 2784
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.609283913659997e-06,
+      "loss": 1.262,
+      "step": 2785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.608523711989917e-06,
+      "loss": 1.4443,
+      "step": 2786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.607763510319837e-06,
+      "loss": 1.3952,
+      "step": 2787
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.607003308649757e-06,
+      "loss": 1.3976,
+      "step": 2788
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.606243106979677e-06,
+      "loss": 1.4427,
+      "step": 2789
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.605482905309597e-06,
+      "loss": 1.268,
+      "step": 2790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.604722703639516e-06,
+      "loss": 1.2599,
+      "step": 2791
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.603962501969436e-06,
+      "loss": 1.5654,
+      "step": 2792
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.603202300299355e-06,
+      "loss": 1.3659,
+      "step": 2793
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.602442098629274e-06,
+      "loss": 1.2457,
+      "step": 2794
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.601681896959194e-06,
+      "loss": 1.5349,
+      "step": 2795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.600921695289114e-06,
+      "loss": 1.2835,
+      "step": 2796
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.600161493619034e-06,
+      "loss": 1.2335,
+      "step": 2797
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.599401291948952e-06,
+      "loss": 1.5983,
+      "step": 2798
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.598641090278872e-06,
+      "loss": 1.3727,
+      "step": 2799
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.597880888608792e-06,
+      "loss": 1.3761,
+      "step": 2800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.597120686938712e-06,
+      "loss": 1.066,
+      "step": 2801
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.596360485268632e-06,
+      "loss": 1.0794,
+      "step": 2802
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.595600283598551e-06,
+      "loss": 1.7888,
+      "step": 2803
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.594840081928471e-06,
+      "loss": 1.505,
+      "step": 2804
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.594079880258391e-06,
+      "loss": 1.5784,
+      "step": 2805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.593319678588311e-06,
+      "loss": 1.2415,
+      "step": 2806
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.59255947691823e-06,
+      "loss": 1.3163,
+      "step": 2807
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.59179927524815e-06,
+      "loss": 1.4594,
+      "step": 2808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.591039073578069e-06,
+      "loss": 1.3453,
+      "step": 2809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.590278871907989e-06,
+      "loss": 1.1001,
+      "step": 2810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.589518670237909e-06,
+      "loss": 1.256,
+      "step": 2811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.588758468567828e-06,
+      "loss": 1.3237,
+      "step": 2812
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.587998266897748e-06,
+      "loss": 1.3946,
+      "step": 2813
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.587238065227666e-06,
+      "loss": 1.2025,
+      "step": 2814
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.586477863557586e-06,
+      "loss": 1.3825,
+      "step": 2815
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.585717661887506e-06,
+      "loss": 1.3521,
+      "step": 2816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.584957460217426e-06,
+      "loss": 1.4409,
+      "step": 2817
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.584197258547346e-06,
+      "loss": 1.2147,
+      "step": 2818
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.583437056877266e-06,
+      "loss": 1.3353,
+      "step": 2819
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.582676855207186e-06,
+      "loss": 1.2615,
+      "step": 2820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.581916653537105e-06,
+      "loss": 1.272,
+      "step": 2821
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.581156451867025e-06,
+      "loss": 1.0595,
+      "step": 2822
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.580396250196945e-06,
+      "loss": 1.0038,
+      "step": 2823
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.579636048526865e-06,
+      "loss": 1.1592,
+      "step": 2824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.578875846856783e-06,
+      "loss": 1.1599,
+      "step": 2825
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.578115645186703e-06,
+      "loss": 1.0975,
+      "step": 2826
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.577355443516623e-06,
+      "loss": 1.0866,
+      "step": 2827
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.576595241846543e-06,
+      "loss": 1.1993,
+      "step": 2828
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.575835040176463e-06,
+      "loss": 0.865,
+      "step": 2829
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.57507483850638e-06,
+      "loss": 1.323,
+      "step": 2830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.5743146368363e-06,
+      "loss": 1.2258,
+      "step": 2831
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.57355443516622e-06,
+      "loss": 0.9457,
+      "step": 2832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.57279423349614e-06,
+      "loss": 1.1339,
+      "step": 2833
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.57203403182606e-06,
+      "loss": 1.0348,
+      "step": 2834
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.57127383015598e-06,
+      "loss": 1.0518,
+      "step": 2835
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.5705136284859e-06,
+      "loss": 1.0238,
+      "step": 2836
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.56975342681582e-06,
+      "loss": 1.1035,
+      "step": 2837
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.56899322514574e-06,
+      "loss": 1.2359,
+      "step": 2838
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.56823302347566e-06,
+      "loss": 0.8801,
+      "step": 2839
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.56747282180558e-06,
+      "loss": 1.3123,
+      "step": 2840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.566712620135497e-06,
+      "loss": 0.8103,
+      "step": 2841
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.565952418465417e-06,
+      "loss": 0.9346,
+      "step": 2842
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.565192216795337e-06,
+      "loss": 1.2288,
+      "step": 2843
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.564432015125257e-06,
+      "loss": 0.9897,
+      "step": 2844
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.563671813455177e-06,
+      "loss": 1.225,
+      "step": 2845
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.562911611785095e-06,
+      "loss": 1.0632,
+      "step": 2846
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.562151410115015e-06,
+      "loss": 1.0606,
+      "step": 2847
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.561391208444935e-06,
+      "loss": 1.0544,
+      "step": 2848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.560631006774855e-06,
+      "loss": 0.9285,
+      "step": 2849
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.559870805104774e-06,
+      "loss": 1.0808,
+      "step": 2850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.559110603434694e-06,
+      "loss": 1.1544,
+      "step": 2851
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.558350401764614e-06,
+      "loss": 1.0114,
+      "step": 2852
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.557590200094534e-06,
+      "loss": 0.9802,
+      "step": 2853
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.556829998424454e-06,
+      "loss": 1.1057,
+      "step": 2854
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.556069796754374e-06,
+      "loss": 1.0698,
+      "step": 2855
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.555309595084294e-06,
+      "loss": 0.9613,
+      "step": 2856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.554549393414212e-06,
+      "loss": 1.1671,
+      "step": 2857
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.553789191744132e-06,
+      "loss": 1.1594,
+      "step": 2858
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.553028990074051e-06,
+      "loss": 1.2431,
+      "step": 2859
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.552268788403971e-06,
+      "loss": 1.0466,
+      "step": 2860
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.551508586733891e-06,
+      "loss": 1.0397,
+      "step": 2861
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.55074838506381e-06,
+      "loss": 1.1033,
+      "step": 2862
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.54998818339373e-06,
+      "loss": 0.9687,
+      "step": 2863
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.549227981723649e-06,
+      "loss": 0.9661,
+      "step": 2864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.548467780053569e-06,
+      "loss": 1.0093,
+      "step": 2865
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.547707578383489e-06,
+      "loss": 1.0252,
+      "step": 2866
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.546947376713409e-06,
+      "loss": 1.0818,
+      "step": 2867
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.546187175043328e-06,
+      "loss": 1.0777,
+      "step": 2868
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.545426973373248e-06,
+      "loss": 1.1708,
+      "step": 2869
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.544666771703168e-06,
+      "loss": 1.1968,
+      "step": 2870
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.543906570033088e-06,
+      "loss": 1.0763,
+      "step": 2871
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.543146368363008e-06,
+      "loss": 1.0529,
+      "step": 2872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.542386166692926e-06,
+      "loss": 0.9736,
+      "step": 2873
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.541625965022846e-06,
+      "loss": 1.1759,
+      "step": 2874
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.540865763352766e-06,
+      "loss": 1.2521,
+      "step": 2875
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.540105561682686e-06,
+      "loss": 1.1547,
+      "step": 2876
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.539345360012605e-06,
+      "loss": 0.9792,
+      "step": 2877
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.538585158342524e-06,
+      "loss": 1.207,
+      "step": 2878
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.537824956672443e-06,
+      "loss": 1.2295,
+      "step": 2879
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.537064755002363e-06,
+      "loss": 1.0285,
+      "step": 2880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.536304553332283e-06,
+      "loss": 0.9787,
+      "step": 2881
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.535544351662203e-06,
+      "loss": 1.1072,
+      "step": 2882
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.534784149992123e-06,
+      "loss": 1.1316,
+      "step": 2883
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.534023948322043e-06,
+      "loss": 1.0561,
+      "step": 2884
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.533263746651963e-06,
+      "loss": 1.2019,
+      "step": 2885
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.532503544981882e-06,
+      "loss": 1.1591,
+      "step": 2886
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.531743343311802e-06,
+      "loss": 1.0362,
+      "step": 2887
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.530983141641722e-06,
+      "loss": 1.2928,
+      "step": 2888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.53022293997164e-06,
+      "loss": 1.0345,
+      "step": 2889
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.52946273830156e-06,
+      "loss": 1.1775,
+      "step": 2890
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.52870253663148e-06,
+      "loss": 0.9236,
+      "step": 2891
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.5279423349614e-06,
+      "loss": 1.2382,
+      "step": 2892
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.52718213329132e-06,
+      "loss": 1.0494,
+      "step": 2893
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.526421931621238e-06,
+      "loss": 1.11,
+      "step": 2894
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.525661729951158e-06,
+      "loss": 1.0526,
+      "step": 2895
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.524901528281078e-06,
+      "loss": 1.1921,
+      "step": 2896
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.524141326610998e-06,
+      "loss": 1.3295,
+      "step": 2897
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.523381124940917e-06,
+      "loss": 1.0687,
+      "step": 2898
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.522620923270837e-06,
+      "loss": 0.9282,
+      "step": 2899
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.521860721600757e-06,
+      "loss": 0.9095,
+      "step": 2900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.521100519930677e-06,
+      "loss": 1.2092,
+      "step": 2901
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.520340318260597e-06,
+      "loss": 1.2147,
+      "step": 2902
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.519580116590517e-06,
+      "loss": 0.9093,
+      "step": 2903
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.518819914920437e-06,
+      "loss": 1.1205,
+      "step": 2904
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.518059713250355e-06,
+      "loss": 1.3478,
+      "step": 2905
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.517299511580275e-06,
+      "loss": 1.1239,
+      "step": 2906
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.516539309910194e-06,
+      "loss": 1.0389,
+      "step": 2907
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.515779108240114e-06,
+      "loss": 1.1716,
+      "step": 2908
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.515018906570034e-06,
+      "loss": 0.9876,
+      "step": 2909
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.514258704899952e-06,
+      "loss": 1.2706,
+      "step": 2910
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.513498503229872e-06,
+      "loss": 1.0573,
+      "step": 2911
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.512738301559792e-06,
+      "loss": 0.963,
+      "step": 2912
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.511978099889712e-06,
+      "loss": 1.1778,
+      "step": 2913
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.511217898219632e-06,
+      "loss": 1.1102,
+      "step": 2914
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.510457696549552e-06,
+      "loss": 1.1505,
+      "step": 2915
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.509697494879471e-06,
+      "loss": 1.2155,
+      "step": 2916
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.508937293209391e-06,
+      "loss": 1.1532,
+      "step": 2917
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.508177091539311e-06,
+      "loss": 0.9009,
+      "step": 2918
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.507416889869231e-06,
+      "loss": 1.1358,
+      "step": 2919
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.50665668819915e-06,
+      "loss": 1.0177,
+      "step": 2920
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.505896486529069e-06,
+      "loss": 0.917,
+      "step": 2921
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.505136284858989e-06,
+      "loss": 1.0053,
+      "step": 2922
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.504376083188909e-06,
+      "loss": 1.2559,
+      "step": 2923
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.503615881518829e-06,
+      "loss": 0.9567,
+      "step": 2924
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.502855679848748e-06,
+      "loss": 1.1215,
+      "step": 2925
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.502095478178667e-06,
+      "loss": 1.0218,
+      "step": 2926
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.501335276508586e-06,
+      "loss": 0.9914,
+      "step": 2927
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.500575074838506e-06,
+      "loss": 1.0203,
+      "step": 2928
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.499814873168426e-06,
+      "loss": 1.1313,
+      "step": 2929
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.499054671498346e-06,
+      "loss": 1.0243,
+      "step": 2930
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.498294469828266e-06,
+      "loss": 1.025,
+      "step": 2931
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.497534268158186e-06,
+      "loss": 1.1205,
+      "step": 2932
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.496774066488106e-06,
+      "loss": 1.0932,
+      "step": 2933
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.496013864818025e-06,
+      "loss": 0.9468,
+      "step": 2934
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.495253663147945e-06,
+      "loss": 1.0139,
+      "step": 2935
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.494493461477865e-06,
+      "loss": 1.0422,
+      "step": 2936
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.493733259807783e-06,
+      "loss": 1.2155,
+      "step": 2937
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.492973058137703e-06,
+      "loss": 1.2381,
+      "step": 2938
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.492212856467623e-06,
+      "loss": 0.86,
+      "step": 2939
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.491452654797543e-06,
+      "loss": 1.0747,
+      "step": 2940
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.490692453127463e-06,
+      "loss": 0.9257,
+      "step": 2941
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.489932251457381e-06,
+      "loss": 1.032,
+      "step": 2942
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.4891720497873e-06,
+      "loss": 1.0476,
+      "step": 2943
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.48841184811722e-06,
+      "loss": 1.2708,
+      "step": 2944
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.48765164644714e-06,
+      "loss": 0.9322,
+      "step": 2945
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.48689144477706e-06,
+      "loss": 0.9916,
+      "step": 2946
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.48613124310698e-06,
+      "loss": 1.0537,
+      "step": 2947
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.4853710414369e-06,
+      "loss": 1.1046,
+      "step": 2948
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.48461083976682e-06,
+      "loss": 0.9985,
+      "step": 2949
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.48385063809674e-06,
+      "loss": 0.9892,
+      "step": 2950
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.48309043642666e-06,
+      "loss": 1.0903,
+      "step": 2951
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.48233023475658e-06,
+      "loss": 1.0907,
+      "step": 2952
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.481570033086498e-06,
+      "loss": 1.0318,
+      "step": 2953
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.480809831416417e-06,
+      "loss": 1.1468,
+      "step": 2954
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.480049629746337e-06,
+      "loss": 0.8967,
+      "step": 2955
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.479289428076257e-06,
+      "loss": 1.3047,
+      "step": 2956
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.478529226406177e-06,
+      "loss": 1.068,
+      "step": 2957
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.477769024736097e-06,
+      "loss": 1.0798,
+      "step": 2958
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.477008823066017e-06,
+      "loss": 1.2119,
+      "step": 2959
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.476248621395937e-06,
+      "loss": 1.1416,
+      "step": 2960
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.475488419725856e-06,
+      "loss": 1.0075,
+      "step": 2961
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.474728218055775e-06,
+      "loss": 1.0262,
+      "step": 2962
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.473968016385694e-06,
+      "loss": 1.2099,
+      "step": 2963
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.473207814715614e-06,
+      "loss": 1.2765,
+      "step": 2964
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.472447613045534e-06,
+      "loss": 1.2154,
+      "step": 2965
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.471687411375454e-06,
+      "loss": 1.0855,
+      "step": 2966
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.470927209705374e-06,
+      "loss": 1.1051,
+      "step": 2967
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.470167008035294e-06,
+      "loss": 1.0118,
+      "step": 2968
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.469406806365212e-06,
+      "loss": 0.8761,
+      "step": 2969
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.468646604695132e-06,
+      "loss": 1.0045,
+      "step": 2970
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.467886403025052e-06,
+      "loss": 1.2826,
+      "step": 2971
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.467126201354971e-06,
+      "loss": 1.3944,
+      "step": 2972
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.466365999684891e-06,
+      "loss": 0.951,
+      "step": 2973
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.465605798014811e-06,
+      "loss": 1.1709,
+      "step": 2974
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.464845596344731e-06,
+      "loss": 1.0279,
+      "step": 2975
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.464085394674651e-06,
+      "loss": 1.0844,
+      "step": 2976
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.46332519300457e-06,
+      "loss": 1.2314,
+      "step": 2977
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.46256499133449e-06,
+      "loss": 1.1846,
+      "step": 2978
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.461804789664409e-06,
+      "loss": 1.2669,
+      "step": 2979
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.461044587994329e-06,
+      "loss": 1.2303,
+      "step": 2980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.460284386324248e-06,
+      "loss": 1.0524,
+      "step": 2981
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.459524184654168e-06,
+      "loss": 1.1061,
+      "step": 2982
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.458763982984088e-06,
+      "loss": 1.0979,
+      "step": 2983
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.458003781314008e-06,
+      "loss": 1.0451,
+      "step": 2984
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.457243579643926e-06,
+      "loss": 1.0363,
+      "step": 2985
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.456483377973846e-06,
+      "loss": 1.3339,
+      "step": 2986
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.455723176303766e-06,
+      "loss": 0.9891,
+      "step": 2987
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.454962974633686e-06,
+      "loss": 1.1585,
+      "step": 2988
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.454202772963606e-06,
+      "loss": 1.1204,
+      "step": 2989
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.453442571293525e-06,
+      "loss": 0.9182,
+      "step": 2990
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.452682369623445e-06,
+      "loss": 0.8641,
+      "step": 2991
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.451922167953365e-06,
+      "loss": 1.054,
+      "step": 2992
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.451161966283285e-06,
+      "loss": 1.0941,
+      "step": 2993
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.450401764613205e-06,
+      "loss": 1.0908,
+      "step": 2994
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.449641562943123e-06,
+      "loss": 0.8834,
+      "step": 2995
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.448881361273043e-06,
+      "loss": 1.0381,
+      "step": 2996
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.448121159602963e-06,
+      "loss": 1.0344,
+      "step": 2997
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.447360957932883e-06,
+      "loss": 1.0625,
+      "step": 2998
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.446600756262802e-06,
+      "loss": 1.0518,
+      "step": 2999
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.445840554592722e-06,
+      "loss": 1.0181,
+      "step": 3000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.44508035292264e-06,
+      "loss": 0.9129,
+      "step": 3001
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.44432015125256e-06,
+      "loss": 0.9873,
+      "step": 3002
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.44355994958248e-06,
+      "loss": 1.0043,
+      "step": 3003
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.4427997479124e-06,
+      "loss": 0.8103,
+      "step": 3004
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.44203954624232e-06,
+      "loss": 0.8024,
+      "step": 3005
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.44127934457224e-06,
+      "loss": 1.164,
+      "step": 3006
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.44051914290216e-06,
+      "loss": 0.9337,
+      "step": 3007
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.43975894123208e-06,
+      "loss": 1.1035,
+      "step": 3008
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.438998739562e-06,
+      "loss": 1.1831,
+      "step": 3009
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.43823853789192e-06,
+      "loss": 1.1512,
+      "step": 3010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.437478336221837e-06,
+      "loss": 1.27,
+      "step": 3011
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.436718134551757e-06,
+      "loss": 1.1877,
+      "step": 3012
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.435957932881677e-06,
+      "loss": 0.9544,
+      "step": 3013
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.435197731211597e-06,
+      "loss": 1.0967,
+      "step": 3014
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.434437529541517e-06,
+      "loss": 1.0752,
+      "step": 3015
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.433677327871437e-06,
+      "loss": 0.89,
+      "step": 3016
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.432917126201355e-06,
+      "loss": 1.3047,
+      "step": 3017
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.432156924531275e-06,
+      "loss": 1.2,
+      "step": 3018
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.431396722861195e-06,
+      "loss": 1.064,
+      "step": 3019
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.430636521191114e-06,
+      "loss": 1.0481,
+      "step": 3020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.429876319521034e-06,
+      "loss": 0.9964,
+      "step": 3021
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.429116117850954e-06,
+      "loss": 1.0764,
+      "step": 3022
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.428355916180874e-06,
+      "loss": 0.9052,
+      "step": 3023
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.427595714510794e-06,
+      "loss": 1.14,
+      "step": 3024
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.426835512840714e-06,
+      "loss": 1.0134,
+      "step": 3025
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.426075311170634e-06,
+      "loss": 1.2012,
+      "step": 3026
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.425315109500552e-06,
+      "loss": 0.9865,
+      "step": 3027
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.424554907830472e-06,
+      "loss": 1.006,
+      "step": 3028
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.423794706160391e-06,
+      "loss": 1.0409,
+      "step": 3029
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.423034504490311e-06,
+      "loss": 1.2033,
+      "step": 3030
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.422274302820231e-06,
+      "loss": 0.9519,
+      "step": 3031
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.421514101150151e-06,
+      "loss": 1.217,
+      "step": 3032
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.420753899480069e-06,
+      "loss": 1.2102,
+      "step": 3033
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.419993697809989e-06,
+      "loss": 1.0089,
+      "step": 3034
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.419233496139909e-06,
+      "loss": 1.1578,
+      "step": 3035
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.418473294469829e-06,
+      "loss": 1.0736,
+      "step": 3036
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.417713092799749e-06,
+      "loss": 1.1406,
+      "step": 3037
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.416952891129668e-06,
+      "loss": 1.1414,
+      "step": 3038
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.416192689459588e-06,
+      "loss": 0.8218,
+      "step": 3039
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.415432487789508e-06,
+      "loss": 1.1563,
+      "step": 3040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.414672286119428e-06,
+      "loss": 1.1867,
+      "step": 3041
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.413912084449348e-06,
+      "loss": 1.0013,
+      "step": 3042
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.413151882779266e-06,
+      "loss": 1.2325,
+      "step": 3043
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.412391681109186e-06,
+      "loss": 0.9533,
+      "step": 3044
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.411631479439106e-06,
+      "loss": 1.0037,
+      "step": 3045
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.410871277769026e-06,
+      "loss": 1.0541,
+      "step": 3046
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.410111076098945e-06,
+      "loss": 0.9985,
+      "step": 3047
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.409350874428865e-06,
+      "loss": 1.1026,
+      "step": 3048
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.408590672758783e-06,
+      "loss": 1.312,
+      "step": 3049
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.407830471088703e-06,
+      "loss": 1.1257,
+      "step": 3050
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.407070269418623e-06,
+      "loss": 1.0568,
+      "step": 3051
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.406310067748543e-06,
+      "loss": 1.212,
+      "step": 3052
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.405549866078463e-06,
+      "loss": 1.1321,
+      "step": 3053
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.404789664408383e-06,
+      "loss": 1.1847,
+      "step": 3054
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.404029462738303e-06,
+      "loss": 1.0773,
+      "step": 3055
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.403269261068222e-06,
+      "loss": 0.9967,
+      "step": 3056
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.402509059398142e-06,
+      "loss": 1.1695,
+      "step": 3057
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.401748857728062e-06,
+      "loss": 1.1735,
+      "step": 3058
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.40098865605798e-06,
+      "loss": 1.1076,
+      "step": 3059
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.4002284543879e-06,
+      "loss": 1.1835,
+      "step": 3060
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.39946825271782e-06,
+      "loss": 1.1344,
+      "step": 3061
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.39870805104774e-06,
+      "loss": 1.0496,
+      "step": 3062
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.39794784937766e-06,
+      "loss": 1.0456,
+      "step": 3063
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.397187647707578e-06,
+      "loss": 1.0815,
+      "step": 3064
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.396427446037498e-06,
+      "loss": 1.0693,
+      "step": 3065
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.395667244367418e-06,
+      "loss": 1.0143,
+      "step": 3066
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.394907042697337e-06,
+      "loss": 1.0875,
+      "step": 3067
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.394146841027257e-06,
+      "loss": 1.0674,
+      "step": 3068
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.393386639357177e-06,
+      "loss": 0.9801,
+      "step": 3069
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.392626437687097e-06,
+      "loss": 1.0773,
+      "step": 3070
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.391866236017017e-06,
+      "loss": 1.2258,
+      "step": 3071
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.391106034346937e-06,
+      "loss": 0.928,
+      "step": 3072
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.390345832676857e-06,
+      "loss": 1.01,
+      "step": 3073
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.389585631006776e-06,
+      "loss": 0.9982,
+      "step": 3074
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.388825429336695e-06,
+      "loss": 1.1475,
+      "step": 3075
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.388065227666614e-06,
+      "loss": 1.0743,
+      "step": 3076
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.387305025996534e-06,
+      "loss": 1.1056,
+      "step": 3077
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.386544824326454e-06,
+      "loss": 1.0864,
+      "step": 3078
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.385784622656374e-06,
+      "loss": 1.0116,
+      "step": 3079
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.385024420986292e-06,
+      "loss": 1.1313,
+      "step": 3080
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.384264219316212e-06,
+      "loss": 1.0967,
+      "step": 3081
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.383504017646132e-06,
+      "loss": 1.216,
+      "step": 3082
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.382743815976052e-06,
+      "loss": 1.1127,
+      "step": 3083
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.381983614305972e-06,
+      "loss": 1.1219,
+      "step": 3084
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.381223412635891e-06,
+      "loss": 1.0684,
+      "step": 3085
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.380463210965811e-06,
+      "loss": 0.9774,
+      "step": 3086
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.379703009295731e-06,
+      "loss": 0.9602,
+      "step": 3087
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.378942807625651e-06,
+      "loss": 1.3086,
+      "step": 3088
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.378182605955571e-06,
+      "loss": 0.8552,
+      "step": 3089
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.37742240428549e-06,
+      "loss": 1.0304,
+      "step": 3090
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.376662202615409e-06,
+      "loss": 1.0323,
+      "step": 3091
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.375902000945329e-06,
+      "loss": 1.0072,
+      "step": 3092
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.375141799275249e-06,
+      "loss": 1.1228,
+      "step": 3093
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.374381597605168e-06,
+      "loss": 0.9672,
+      "step": 3094
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.373621395935088e-06,
+      "loss": 1.0973,
+      "step": 3095
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.372861194265007e-06,
+      "loss": 0.9067,
+      "step": 3096
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.372100992594926e-06,
+      "loss": 1.1027,
+      "step": 3097
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.371340790924846e-06,
+      "loss": 1.1298,
+      "step": 3098
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.370580589254766e-06,
+      "loss": 1.2093,
+      "step": 3099
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.369820387584686e-06,
+      "loss": 1.1439,
+      "step": 3100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.369060185914606e-06,
+      "loss": 1.0739,
+      "step": 3101
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.368299984244526e-06,
+      "loss": 1.1492,
+      "step": 3102
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.367539782574445e-06,
+      "loss": 1.0967,
+      "step": 3103
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.366779580904365e-06,
+      "loss": 1.0075,
+      "step": 3104
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.366019379234285e-06,
+      "loss": 1.2526,
+      "step": 3105
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.365259177564205e-06,
+      "loss": 0.9473,
+      "step": 3106
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.364498975894123e-06,
+      "loss": 1.0322,
+      "step": 3107
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.363738774224043e-06,
+      "loss": 1.1546,
+      "step": 3108
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.362978572553963e-06,
+      "loss": 1.0011,
+      "step": 3109
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.362218370883883e-06,
+      "loss": 0.9755,
+      "step": 3110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.361458169213803e-06,
+      "loss": 1.0896,
+      "step": 3111
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.36069796754372e-06,
+      "loss": 1.1548,
+      "step": 3112
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.35993776587364e-06,
+      "loss": 1.2367,
+      "step": 3113
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.35917756420356e-06,
+      "loss": 0.9033,
+      "step": 3114
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.35841736253348e-06,
+      "loss": 1.2022,
+      "step": 3115
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.3576571608634e-06,
+      "loss": 1.1329,
+      "step": 3116
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.35689695919332e-06,
+      "loss": 0.9341,
+      "step": 3117
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.35613675752324e-06,
+      "loss": 1.1447,
+      "step": 3118
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.35537655585316e-06,
+      "loss": 1.3997,
+      "step": 3119
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.35461635418308e-06,
+      "loss": 1.1167,
+      "step": 3120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.353856152513e-06,
+      "loss": 1.0408,
+      "step": 3121
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.35309595084292e-06,
+      "loss": 1.1621,
+      "step": 3122
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.352335749172838e-06,
+      "loss": 0.9445,
+      "step": 3123
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.351575547502757e-06,
+      "loss": 1.1949,
+      "step": 3124
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.350815345832677e-06,
+      "loss": 0.9529,
+      "step": 3125
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.350055144162597e-06,
+      "loss": 1.2183,
+      "step": 3126
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.349294942492517e-06,
+      "loss": 1.1068,
+      "step": 3127
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.348534740822435e-06,
+      "loss": 1.2122,
+      "step": 3128
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.347774539152355e-06,
+      "loss": 1.0282,
+      "step": 3129
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.347014337482275e-06,
+      "loss": 1.0045,
+      "step": 3130
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.346254135812195e-06,
+      "loss": 1.11,
+      "step": 3131
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.345493934142115e-06,
+      "loss": 1.0817,
+      "step": 3132
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.344733732472034e-06,
+      "loss": 1.0172,
+      "step": 3133
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.343973530801954e-06,
+      "loss": 1.1934,
+      "step": 3134
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.343213329131874e-06,
+      "loss": 1.2732,
+      "step": 3135
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.342453127461794e-06,
+      "loss": 1.193,
+      "step": 3136
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.341692925791714e-06,
+      "loss": 1.1436,
+      "step": 3137
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.340932724121634e-06,
+      "loss": 0.9239,
+      "step": 3138
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.340172522451552e-06,
+      "loss": 1.1438,
+      "step": 3139
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.339412320781472e-06,
+      "loss": 1.2396,
+      "step": 3140
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.338652119111392e-06,
+      "loss": 1.4534,
+      "step": 3141
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.337891917441311e-06,
+      "loss": 1.1701,
+      "step": 3142
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.337131715771231e-06,
+      "loss": 1.073,
+      "step": 3143
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.336371514101151e-06,
+      "loss": 1.0563,
+      "step": 3144
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.335611312431071e-06,
+      "loss": 1.1245,
+      "step": 3145
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.33485111076099e-06,
+      "loss": 1.3191,
+      "step": 3146
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.33409090909091e-06,
+      "loss": 1.0838,
+      "step": 3147
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.33333070742083e-06,
+      "loss": 1.0812,
+      "step": 3148
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.33257050575075e-06,
+      "loss": 1.2348,
+      "step": 3149
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.331810304080669e-06,
+      "loss": 1.1143,
+      "step": 3150
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.331050102410588e-06,
+      "loss": 0.9975,
+      "step": 3151
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.330289900740508e-06,
+      "loss": 1.1423,
+      "step": 3152
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.329529699070428e-06,
+      "loss": 1.1308,
+      "step": 3153
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.328769497400348e-06,
+      "loss": 1.0463,
+      "step": 3154
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.328009295730266e-06,
+      "loss": 1.0532,
+      "step": 3155
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.327249094060186e-06,
+      "loss": 1.1137,
+      "step": 3156
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.326488892390106e-06,
+      "loss": 1.1263,
+      "step": 3157
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.325728690720026e-06,
+      "loss": 1.0954,
+      "step": 3158
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.324968489049946e-06,
+      "loss": 1.0999,
+      "step": 3159
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.324208287379865e-06,
+      "loss": 1.023,
+      "step": 3160
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.323448085709785e-06,
+      "loss": 1.0177,
+      "step": 3161
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.322687884039705e-06,
+      "loss": 1.3737,
+      "step": 3162
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.321927682369625e-06,
+      "loss": 1.0688,
+      "step": 3163
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.321167480699545e-06,
+      "loss": 1.2151,
+      "step": 3164
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.320407279029465e-06,
+      "loss": 0.9752,
+      "step": 3165
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.319647077359383e-06,
+      "loss": 1.1601,
+      "step": 3166
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.318886875689303e-06,
+      "loss": 1.0999,
+      "step": 3167
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.318126674019223e-06,
+      "loss": 0.9721,
+      "step": 3168
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.317366472349142e-06,
+      "loss": 0.9137,
+      "step": 3169
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.316606270679062e-06,
+      "loss": 1.0344,
+      "step": 3170
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.31584606900898e-06,
+      "loss": 0.9076,
+      "step": 3171
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.3150858673389e-06,
+      "loss": 1.0499,
+      "step": 3172
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.31432566566882e-06,
+      "loss": 1.1068,
+      "step": 3173
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.31356546399874e-06,
+      "loss": 1.0665,
+      "step": 3174
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.31280526232866e-06,
+      "loss": 1.0245,
+      "step": 3175
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.31204506065858e-06,
+      "loss": 1.2876,
+      "step": 3176
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.3112848589885e-06,
+      "loss": 1.1313,
+      "step": 3177
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.31052465731842e-06,
+      "loss": 0.9875,
+      "step": 3178
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.30976445564834e-06,
+      "loss": 1.0603,
+      "step": 3179
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.30900425397826e-06,
+      "loss": 1.0853,
+      "step": 3180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.308244052308179e-06,
+      "loss": 1.1056,
+      "step": 3181
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.307483850638097e-06,
+      "loss": 1.1038,
+      "step": 3182
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.306723648968017e-06,
+      "loss": 1.2537,
+      "step": 3183
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.305963447297937e-06,
+      "loss": 1.1925,
+      "step": 3184
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.305203245627857e-06,
+      "loss": 1.2327,
+      "step": 3185
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.304443043957777e-06,
+      "loss": 0.9735,
+      "step": 3186
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.303682842287695e-06,
+      "loss": 0.8743,
+      "step": 3187
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.302922640617615e-06,
+      "loss": 1.1932,
+      "step": 3188
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.302162438947534e-06,
+      "loss": 1.1709,
+      "step": 3189
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.301402237277454e-06,
+      "loss": 1.2398,
+      "step": 3190
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.300642035607374e-06,
+      "loss": 1.009,
+      "step": 3191
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.299881833937294e-06,
+      "loss": 0.9707,
+      "step": 3192
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.299121632267214e-06,
+      "loss": 0.9499,
+      "step": 3193
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.298361430597134e-06,
+      "loss": 1.066,
+      "step": 3194
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.297601228927054e-06,
+      "loss": 0.9628,
+      "step": 3195
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.296841027256973e-06,
+      "loss": 1.2268,
+      "step": 3196
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.296080825586893e-06,
+      "loss": 0.9228,
+      "step": 3197
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.295320623916811e-06,
+      "loss": 0.8618,
+      "step": 3198
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.294560422246731e-06,
+      "loss": 0.9071,
+      "step": 3199
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.293800220576651e-06,
+      "loss": 1.1669,
+      "step": 3200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.293040018906571e-06,
+      "loss": 1.0565,
+      "step": 3201
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.292279817236491e-06,
+      "loss": 1.145,
+      "step": 3202
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.291519615566409e-06,
+      "loss": 1.137,
+      "step": 3203
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.290759413896329e-06,
+      "loss": 1.0577,
+      "step": 3204
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.289999212226249e-06,
+      "loss": 1.2628,
+      "step": 3205
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.289239010556169e-06,
+      "loss": 1.226,
+      "step": 3206
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.288478808886088e-06,
+      "loss": 0.9208,
+      "step": 3207
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.287718607216008e-06,
+      "loss": 0.9884,
+      "step": 3208
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.286958405545928e-06,
+      "loss": 1.0843,
+      "step": 3209
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.286198203875848e-06,
+      "loss": 1.1809,
+      "step": 3210
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.285438002205768e-06,
+      "loss": 1.0142,
+      "step": 3211
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.284677800535688e-06,
+      "loss": 0.933,
+      "step": 3212
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.283917598865608e-06,
+      "loss": 1.0861,
+      "step": 3213
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.283157397195526e-06,
+      "loss": 1.2304,
+      "step": 3214
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.282397195525446e-06,
+      "loss": 1.095,
+      "step": 3215
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.281636993855366e-06,
+      "loss": 1.0693,
+      "step": 3216
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.280876792185285e-06,
+      "loss": 1.1242,
+      "step": 3217
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.280116590515205e-06,
+      "loss": 1.1512,
+      "step": 3218
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.279356388845123e-06,
+      "loss": 1.0277,
+      "step": 3219
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.278596187175043e-06,
+      "loss": 1.2488,
+      "step": 3220
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.277835985504963e-06,
+      "loss": 1.0444,
+      "step": 3221
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.277075783834883e-06,
+      "loss": 0.9522,
+      "step": 3222
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.276315582164803e-06,
+      "loss": 0.9168,
+      "step": 3223
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.275555380494723e-06,
+      "loss": 0.9915,
+      "step": 3224
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.274795178824643e-06,
+      "loss": 0.9707,
+      "step": 3225
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.274034977154562e-06,
+      "loss": 1.05,
+      "step": 3226
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.273274775484482e-06,
+      "loss": 1.0803,
+      "step": 3227
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.272514573814402e-06,
+      "loss": 1.1526,
+      "step": 3228
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.271754372144322e-06,
+      "loss": 0.8452,
+      "step": 3229
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.27099417047424e-06,
+      "loss": 1.0015,
+      "step": 3230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.27023396880416e-06,
+      "loss": 0.8788,
+      "step": 3231
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.26947376713408e-06,
+      "loss": 1.0627,
+      "step": 3232
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.268713565464e-06,
+      "loss": 0.9188,
+      "step": 3233
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.26795336379392e-06,
+      "loss": 1.0622,
+      "step": 3234
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.267193162123838e-06,
+      "loss": 0.817,
+      "step": 3235
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.266432960453758e-06,
+      "loss": 1.1562,
+      "step": 3236
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.265672758783677e-06,
+      "loss": 1.0745,
+      "step": 3237
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.264912557113597e-06,
+      "loss": 1.2598,
+      "step": 3238
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.264152355443517e-06,
+      "loss": 1.222,
+      "step": 3239
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.263392153773437e-06,
+      "loss": 1.0936,
+      "step": 3240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.262631952103357e-06,
+      "loss": 1.0233,
+      "step": 3241
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.261871750433277e-06,
+      "loss": 1.0482,
+      "step": 3242
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.261111548763197e-06,
+      "loss": 1.1518,
+      "step": 3243
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.260351347093116e-06,
+      "loss": 1.2418,
+      "step": 3244
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.259591145423036e-06,
+      "loss": 0.9445,
+      "step": 3245
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.258830943752954e-06,
+      "loss": 1.0476,
+      "step": 3246
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.258070742082874e-06,
+      "loss": 1.0486,
+      "step": 3247
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.257310540412794e-06,
+      "loss": 1.0797,
+      "step": 3248
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.256550338742714e-06,
+      "loss": 1.082,
+      "step": 3249
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.255790137072634e-06,
+      "loss": 1.0846,
+      "step": 3250
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.255029935402552e-06,
+      "loss": 0.8821,
+      "step": 3251
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.254269733732472e-06,
+      "loss": 1.1297,
+      "step": 3252
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.253509532062392e-06,
+      "loss": 1.2394,
+      "step": 3253
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.252749330392312e-06,
+      "loss": 0.9917,
+      "step": 3254
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.251989128722231e-06,
+      "loss": 1.2135,
+      "step": 3255
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.251228927052151e-06,
+      "loss": 1.1813,
+      "step": 3256
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.250468725382071e-06,
+      "loss": 1.036,
+      "step": 3257
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.249708523711991e-06,
+      "loss": 1.0794,
+      "step": 3258
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.24894832204191e-06,
+      "loss": 0.9226,
+      "step": 3259
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.24818812037183e-06,
+      "loss": 1.1888,
+      "step": 3260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.247427918701749e-06,
+      "loss": 1.0574,
+      "step": 3261
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.246667717031669e-06,
+      "loss": 1.0462,
+      "step": 3262
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.245907515361589e-06,
+      "loss": 1.1655,
+      "step": 3263
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.245147313691508e-06,
+      "loss": 1.2776,
+      "step": 3264
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.244387112021428e-06,
+      "loss": 1.0016,
+      "step": 3265
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.243626910351348e-06,
+      "loss": 1.2882,
+      "step": 3266
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.242866708681266e-06,
+      "loss": 1.1933,
+      "step": 3267
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.242106507011186e-06,
+      "loss": 0.9914,
+      "step": 3268
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.241346305341106e-06,
+      "loss": 1.0915,
+      "step": 3269
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.240586103671026e-06,
+      "loss": 1.1664,
+      "step": 3270
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.239825902000946e-06,
+      "loss": 1.1133,
+      "step": 3271
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.239065700330866e-06,
+      "loss": 1.2124,
+      "step": 3272
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.238305498660785e-06,
+      "loss": 1.084,
+      "step": 3273
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.237545296990705e-06,
+      "loss": 0.9432,
+      "step": 3274
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.236785095320625e-06,
+      "loss": 1.245,
+      "step": 3275
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.236024893650545e-06,
+      "loss": 1.0891,
+      "step": 3276
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.235264691980463e-06,
+      "loss": 1.0031,
+      "step": 3277
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.234504490310383e-06,
+      "loss": 1.1715,
+      "step": 3278
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.233744288640303e-06,
+      "loss": 1.1722,
+      "step": 3279
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.232984086970223e-06,
+      "loss": 1.1254,
+      "step": 3280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.232223885300143e-06,
+      "loss": 1.1185,
+      "step": 3281
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.231463683630062e-06,
+      "loss": 1.119,
+      "step": 3282
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.23070348195998e-06,
+      "loss": 1.1615,
+      "step": 3283
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.2299432802899e-06,
+      "loss": 0.983,
+      "step": 3284
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.22918307861982e-06,
+      "loss": 1.0506,
+      "step": 3285
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.22842287694974e-06,
+      "loss": 1.2133,
+      "step": 3286
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.22766267527966e-06,
+      "loss": 0.9309,
+      "step": 3287
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.22690247360958e-06,
+      "loss": 1.051,
+      "step": 3288
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.2261422719395e-06,
+      "loss": 1.1939,
+      "step": 3289
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.22538207026942e-06,
+      "loss": 0.9443,
+      "step": 3290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.22462186859934e-06,
+      "loss": 1.157,
+      "step": 3291
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.22386166692926e-06,
+      "loss": 1.0804,
+      "step": 3292
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.223101465259177e-06,
+      "loss": 1.1846,
+      "step": 3293
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.222341263589097e-06,
+      "loss": 1.0817,
+      "step": 3294
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.221581061919017e-06,
+      "loss": 1.1441,
+      "step": 3295
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.220820860248937e-06,
+      "loss": 1.0522,
+      "step": 3296
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.220060658578857e-06,
+      "loss": 1.162,
+      "step": 3297
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.219300456908777e-06,
+      "loss": 1.1642,
+      "step": 3298
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.218540255238695e-06,
+      "loss": 0.9833,
+      "step": 3299
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.217780053568615e-06,
+      "loss": 1.2381,
+      "step": 3300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.217019851898535e-06,
+      "loss": 1.3637,
+      "step": 3301
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.216259650228454e-06,
+      "loss": 1.0941,
+      "step": 3302
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.215499448558374e-06,
+      "loss": 1.071,
+      "step": 3303
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.214739246888294e-06,
+      "loss": 1.2039,
+      "step": 3304
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.213979045218214e-06,
+      "loss": 1.0361,
+      "step": 3305
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.213218843548134e-06,
+      "loss": 1.1387,
+      "step": 3306
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.212458641878054e-06,
+      "loss": 1.066,
+      "step": 3307
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.211698440207974e-06,
+      "loss": 1.1047,
+      "step": 3308
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.210938238537892e-06,
+      "loss": 1.056,
+      "step": 3309
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.210178036867812e-06,
+      "loss": 1.1516,
+      "step": 3310
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.209417835197731e-06,
+      "loss": 1.0115,
+      "step": 3311
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.208657633527651e-06,
+      "loss": 1.1657,
+      "step": 3312
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.207897431857571e-06,
+      "loss": 1.0705,
+      "step": 3313
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.207137230187491e-06,
+      "loss": 1.0012,
+      "step": 3314
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.206377028517411e-06,
+      "loss": 1.0953,
+      "step": 3315
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.205616826847329e-06,
+      "loss": 0.9724,
+      "step": 3316
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.204856625177249e-06,
+      "loss": 0.9618,
+      "step": 3317
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.204096423507169e-06,
+      "loss": 1.0545,
+      "step": 3318
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.203336221837089e-06,
+      "loss": 1.037,
+      "step": 3319
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.202576020167009e-06,
+      "loss": 1.108,
+      "step": 3320
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.201815818496928e-06,
+      "loss": 1.2221,
+      "step": 3321
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.201055616826848e-06,
+      "loss": 1.0602,
+      "step": 3322
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.200295415156768e-06,
+      "loss": 1.0677,
+      "step": 3323
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.199535213486688e-06,
+      "loss": 1.1178,
+      "step": 3324
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.198775011816606e-06,
+      "loss": 1.0716,
+      "step": 3325
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.198014810146526e-06,
+      "loss": 1.2637,
+      "step": 3326
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.197254608476446e-06,
+      "loss": 1.0826,
+      "step": 3327
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.196494406806366e-06,
+      "loss": 0.9951,
+      "step": 3328
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.195734205136286e-06,
+      "loss": 1.2559,
+      "step": 3329
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.194974003466205e-06,
+      "loss": 0.8996,
+      "step": 3330
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.194213801796125e-06,
+      "loss": 1.1503,
+      "step": 3331
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.193453600126045e-06,
+      "loss": 1.1557,
+      "step": 3332
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.192693398455965e-06,
+      "loss": 1.0838,
+      "step": 3333
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.191933196785885e-06,
+      "loss": 1.0219,
+      "step": 3334
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.191172995115805e-06,
+      "loss": 1.0286,
+      "step": 3335
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.190412793445723e-06,
+      "loss": 1.1473,
+      "step": 3336
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.189652591775643e-06,
+      "loss": 0.91,
+      "step": 3337
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.188892390105563e-06,
+      "loss": 1.2566,
+      "step": 3338
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.188132188435482e-06,
+      "loss": 1.0889,
+      "step": 3339
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.187371986765402e-06,
+      "loss": 1.0482,
+      "step": 3340
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.18661178509532e-06,
+      "loss": 1.203,
+      "step": 3341
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.18585158342524e-06,
+      "loss": 1.3334,
+      "step": 3342
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.18509138175516e-06,
+      "loss": 0.9673,
+      "step": 3343
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.18433118008508e-06,
+      "loss": 1.1481,
+      "step": 3344
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.183570978415e-06,
+      "loss": 0.9926,
+      "step": 3345
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.18281077674492e-06,
+      "loss": 1.2182,
+      "step": 3346
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.18205057507484e-06,
+      "loss": 0.9154,
+      "step": 3347
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.18129037340476e-06,
+      "loss": 1.0884,
+      "step": 3348
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.18053017173468e-06,
+      "loss": 1.3345,
+      "step": 3349
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.179769970064599e-06,
+      "loss": 1.0885,
+      "step": 3350
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.179009768394519e-06,
+      "loss": 0.9805,
+      "step": 3351
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.178249566724437e-06,
+      "loss": 1.0459,
+      "step": 3352
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.177489365054357e-06,
+      "loss": 1.0916,
+      "step": 3353
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.176729163384277e-06,
+      "loss": 1.0645,
+      "step": 3354
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.175968961714197e-06,
+      "loss": 1.0953,
+      "step": 3355
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.175208760044117e-06,
+      "loss": 1.1314,
+      "step": 3356
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.174448558374035e-06,
+      "loss": 1.0901,
+      "step": 3357
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.173688356703955e-06,
+      "loss": 1.225,
+      "step": 3358
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.172928155033874e-06,
+      "loss": 1.0065,
+      "step": 3359
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.172167953363794e-06,
+      "loss": 1.1374,
+      "step": 3360
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.171407751693714e-06,
+      "loss": 1.2705,
+      "step": 3361
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.170647550023634e-06,
+      "loss": 1.1686,
+      "step": 3362
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.169887348353554e-06,
+      "loss": 1.1308,
+      "step": 3363
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.169127146683474e-06,
+      "loss": 1.1509,
+      "step": 3364
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.168366945013394e-06,
+      "loss": 0.9999,
+      "step": 3365
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.167606743343313e-06,
+      "loss": 1.1439,
+      "step": 3366
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.166846541673233e-06,
+      "loss": 1.0065,
+      "step": 3367
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.166086340003151e-06,
+      "loss": 1.0622,
+      "step": 3368
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.165326138333071e-06,
+      "loss": 1.144,
+      "step": 3369
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.164565936662991e-06,
+      "loss": 0.9233,
+      "step": 3370
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.163805734992911e-06,
+      "loss": 1.0926,
+      "step": 3371
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.163045533322831e-06,
+      "loss": 0.9259,
+      "step": 3372
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.162285331652749e-06,
+      "loss": 1.0616,
+      "step": 3373
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.161525129982669e-06,
+      "loss": 0.9134,
+      "step": 3374
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.160764928312589e-06,
+      "loss": 0.9664,
+      "step": 3375
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.160004726642509e-06,
+      "loss": 1.013,
+      "step": 3376
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.159244524972428e-06,
+      "loss": 1.1301,
+      "step": 3377
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.158484323302348e-06,
+      "loss": 1.0569,
+      "step": 3378
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.157724121632268e-06,
+      "loss": 1.0187,
+      "step": 3379
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.156963919962188e-06,
+      "loss": 1.1372,
+      "step": 3380
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.156203718292108e-06,
+      "loss": 1.1857,
+      "step": 3381
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.155443516622028e-06,
+      "loss": 1.169,
+      "step": 3382
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.154683314951948e-06,
+      "loss": 0.955,
+      "step": 3383
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.153923113281866e-06,
+      "loss": 1.1351,
+      "step": 3384
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.153162911611786e-06,
+      "loss": 1.1696,
+      "step": 3385
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.152402709941705e-06,
+      "loss": 1.1055,
+      "step": 3386
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.151642508271625e-06,
+      "loss": 1.1934,
+      "step": 3387
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.150882306601545e-06,
+      "loss": 1.2142,
+      "step": 3388
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.150122104931463e-06,
+      "loss": 1.2727,
+      "step": 3389
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.149361903261383e-06,
+      "loss": 1.0431,
+      "step": 3390
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.148601701591303e-06,
+      "loss": 0.9033,
+      "step": 3391
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.147841499921223e-06,
+      "loss": 1.0814,
+      "step": 3392
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.147081298251143e-06,
+      "loss": 1.0955,
+      "step": 3393
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.146321096581063e-06,
+      "loss": 1.0815,
+      "step": 3394
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.145560894910982e-06,
+      "loss": 1.123,
+      "step": 3395
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.144800693240902e-06,
+      "loss": 1.0131,
+      "step": 3396
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.144040491570822e-06,
+      "loss": 1.3795,
+      "step": 3397
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.143280289900742e-06,
+      "loss": 1.0113,
+      "step": 3398
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.142520088230662e-06,
+      "loss": 1.0973,
+      "step": 3399
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.14175988656058e-06,
+      "loss": 0.8978,
+      "step": 3400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.1409996848905e-06,
+      "loss": 0.9553,
+      "step": 3401
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.14023948322042e-06,
+      "loss": 1.2721,
+      "step": 3402
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.13947928155034e-06,
+      "loss": 0.9917,
+      "step": 3403
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.13871907988026e-06,
+      "loss": 1.0999,
+      "step": 3404
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.137958878210178e-06,
+      "loss": 0.9861,
+      "step": 3405
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.137198676540097e-06,
+      "loss": 1.0649,
+      "step": 3406
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.136438474870017e-06,
+      "loss": 1.1742,
+      "step": 3407
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.135678273199937e-06,
+      "loss": 1.1217,
+      "step": 3408
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.134918071529857e-06,
+      "loss": 1.1943,
+      "step": 3409
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.134157869859777e-06,
+      "loss": 0.9811,
+      "step": 3410
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.133397668189697e-06,
+      "loss": 1.0611,
+      "step": 3411
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.132637466519617e-06,
+      "loss": 1.0795,
+      "step": 3412
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.131877264849536e-06,
+      "loss": 1.2845,
+      "step": 3413
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.131117063179456e-06,
+      "loss": 1.0775,
+      "step": 3414
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.130356861509376e-06,
+      "loss": 0.993,
+      "step": 3415
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.129596659839294e-06,
+      "loss": 1.2036,
+      "step": 3416
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.128836458169214e-06,
+      "loss": 0.8979,
+      "step": 3417
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.128076256499134e-06,
+      "loss": 1.1416,
+      "step": 3418
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.127316054829054e-06,
+      "loss": 1.0271,
+      "step": 3419
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.126555853158974e-06,
+      "loss": 1.2844,
+      "step": 3420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.125795651488892e-06,
+      "loss": 0.9679,
+      "step": 3421
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.125035449818812e-06,
+      "loss": 1.0445,
+      "step": 3422
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.124275248148732e-06,
+      "loss": 0.7847,
+      "step": 3423
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.123515046478652e-06,
+      "loss": 1.1756,
+      "step": 3424
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.122754844808571e-06,
+      "loss": 1.1055,
+      "step": 3425
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.121994643138491e-06,
+      "loss": 1.0944,
+      "step": 3426
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.121234441468411e-06,
+      "loss": 1.0927,
+      "step": 3427
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.120474239798331e-06,
+      "loss": 1.1532,
+      "step": 3428
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.11971403812825e-06,
+      "loss": 1.1882,
+      "step": 3429
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.11895383645817e-06,
+      "loss": 1.131,
+      "step": 3430
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.11819363478809e-06,
+      "loss": 1.2673,
+      "step": 3431
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.117433433118009e-06,
+      "loss": 1.0335,
+      "step": 3432
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.116673231447929e-06,
+      "loss": 0.9971,
+      "step": 3433
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.115913029777848e-06,
+      "loss": 1.1896,
+      "step": 3434
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.115152828107768e-06,
+      "loss": 1.1732,
+      "step": 3435
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.114392626437688e-06,
+      "loss": 0.9878,
+      "step": 3436
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.113632424767606e-06,
+      "loss": 1.0809,
+      "step": 3437
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.112872223097526e-06,
+      "loss": 1.2183,
+      "step": 3438
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.112112021427446e-06,
+      "loss": 1.0786,
+      "step": 3439
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.111351819757366e-06,
+      "loss": 1.1301,
+      "step": 3440
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.110591618087286e-06,
+      "loss": 1.0991,
+      "step": 3441
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.109831416417206e-06,
+      "loss": 0.9395,
+      "step": 3442
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.109071214747125e-06,
+      "loss": 1.3627,
+      "step": 3443
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.108311013077045e-06,
+      "loss": 1.1174,
+      "step": 3444
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.107550811406965e-06,
+      "loss": 1.0816,
+      "step": 3445
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.106790609736885e-06,
+      "loss": 1.1267,
+      "step": 3446
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.106030408066805e-06,
+      "loss": 1.0982,
+      "step": 3447
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.105270206396723e-06,
+      "loss": 1.0424,
+      "step": 3448
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.104510004726643e-06,
+      "loss": 1.1903,
+      "step": 3449
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.103749803056563e-06,
+      "loss": 1.088,
+      "step": 3450
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.102989601386483e-06,
+      "loss": 0.9546,
+      "step": 3451
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.102229399716402e-06,
+      "loss": 1.3357,
+      "step": 3452
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.10146919804632e-06,
+      "loss": 1.1306,
+      "step": 3453
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.10070899637624e-06,
+      "loss": 1.0712,
+      "step": 3454
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.09994879470616e-06,
+      "loss": 1.088,
+      "step": 3455
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.09918859303608e-06,
+      "loss": 0.9703,
+      "step": 3456
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.098428391366e-06,
+      "loss": 1.0786,
+      "step": 3457
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.09766818969592e-06,
+      "loss": 1.1469,
+      "step": 3458
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.09690798802584e-06,
+      "loss": 0.9655,
+      "step": 3459
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.09614778635576e-06,
+      "loss": 0.9336,
+      "step": 3460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.09538758468568e-06,
+      "loss": 1.085,
+      "step": 3461
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.0946273830156e-06,
+      "loss": 1.2772,
+      "step": 3462
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.093867181345519e-06,
+      "loss": 1.2466,
+      "step": 3463
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.093106979675437e-06,
+      "loss": 0.9311,
+      "step": 3464
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.092346778005357e-06,
+      "loss": 1.1048,
+      "step": 3465
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.091586576335277e-06,
+      "loss": 1.1044,
+      "step": 3466
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.090826374665197e-06,
+      "loss": 1.1289,
+      "step": 3467
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.090066172995117e-06,
+      "loss": 1.121,
+      "step": 3468
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.089305971325035e-06,
+      "loss": 1.2395,
+      "step": 3469
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.088545769654955e-06,
+      "loss": 1.2113,
+      "step": 3470
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.087785567984875e-06,
+      "loss": 1.0904,
+      "step": 3471
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.087025366314794e-06,
+      "loss": 0.9701,
+      "step": 3472
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.086265164644714e-06,
+      "loss": 1.0633,
+      "step": 3473
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.085504962974634e-06,
+      "loss": 1.1068,
+      "step": 3474
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.084744761304554e-06,
+      "loss": 1.221,
+      "step": 3475
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.083984559634474e-06,
+      "loss": 1.0689,
+      "step": 3476
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.083224357964394e-06,
+      "loss": 0.9032,
+      "step": 3477
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.082464156294314e-06,
+      "loss": 1.0629,
+      "step": 3478
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.081703954624233e-06,
+      "loss": 1.1307,
+      "step": 3479
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.080943752954152e-06,
+      "loss": 1.107,
+      "step": 3480
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.080183551284071e-06,
+      "loss": 1.1194,
+      "step": 3481
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.079423349613991e-06,
+      "loss": 1.2527,
+      "step": 3482
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.078663147943911e-06,
+      "loss": 1.3116,
+      "step": 3483
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.077902946273831e-06,
+      "loss": 1.067,
+      "step": 3484
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.07714274460375e-06,
+      "loss": 1.172,
+      "step": 3485
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.076382542933669e-06,
+      "loss": 1.2896,
+      "step": 3486
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.075622341263589e-06,
+      "loss": 1.1064,
+      "step": 3487
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.074862139593509e-06,
+      "loss": 1.2763,
+      "step": 3488
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.074101937923429e-06,
+      "loss": 1.1045,
+      "step": 3489
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.073341736253348e-06,
+      "loss": 0.9514,
+      "step": 3490
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.072581534583268e-06,
+      "loss": 1.3641,
+      "step": 3491
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.071821332913188e-06,
+      "loss": 0.8832,
+      "step": 3492
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.071061131243108e-06,
+      "loss": 0.93,
+      "step": 3493
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.070300929573028e-06,
+      "loss": 1.1046,
+      "step": 3494
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.069540727902948e-06,
+      "loss": 0.7435,
+      "step": 3495
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.068780526232866e-06,
+      "loss": 1.1851,
+      "step": 3496
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.068020324562786e-06,
+      "loss": 1.1789,
+      "step": 3497
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.067260122892706e-06,
+      "loss": 1.0943,
+      "step": 3498
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.066499921222625e-06,
+      "loss": 0.9556,
+      "step": 3499
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.065739719552545e-06,
+      "loss": 1.0787,
+      "step": 3500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.064979517882465e-06,
+      "loss": 0.9758,
+      "step": 3501
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.064219316212385e-06,
+      "loss": 1.0412,
+      "step": 3502
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.063459114542305e-06,
+      "loss": 0.8852,
+      "step": 3503
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.062698912872223e-06,
+      "loss": 1.2285,
+      "step": 3504
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.061938711202143e-06,
+      "loss": 1.1615,
+      "step": 3505
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.061178509532063e-06,
+      "loss": 1.0561,
+      "step": 3506
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.060418307861983e-06,
+      "loss": 1.1273,
+      "step": 3507
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.059658106191902e-06,
+      "loss": 0.9314,
+      "step": 3508
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.058897904521822e-06,
+      "loss": 1.3349,
+      "step": 3509
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.058137702851742e-06,
+      "loss": 0.9424,
+      "step": 3510
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.057377501181662e-06,
+      "loss": 0.9853,
+      "step": 3511
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.05661729951158e-06,
+      "loss": 1.2504,
+      "step": 3512
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.0558570978415e-06,
+      "loss": 1.2111,
+      "step": 3513
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.05509689617142e-06,
+      "loss": 1.154,
+      "step": 3514
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.05433669450134e-06,
+      "loss": 0.9667,
+      "step": 3515
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.05357649283126e-06,
+      "loss": 1.1234,
+      "step": 3516
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.05281629116118e-06,
+      "loss": 1.1771,
+      "step": 3517
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.0520560894911e-06,
+      "loss": 1.304,
+      "step": 3518
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.05129588782102e-06,
+      "loss": 1.0267,
+      "step": 3519
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.050535686150939e-06,
+      "loss": 0.9167,
+      "step": 3520
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.049775484480859e-06,
+      "loss": 1.0834,
+      "step": 3521
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.049015282810777e-06,
+      "loss": 1.0965,
+      "step": 3522
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.048255081140697e-06,
+      "loss": 1.3348,
+      "step": 3523
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.047494879470617e-06,
+      "loss": 0.9555,
+      "step": 3524
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.046734677800537e-06,
+      "loss": 1.3055,
+      "step": 3525
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.045974476130456e-06,
+      "loss": 1.125,
+      "step": 3526
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.045214274460376e-06,
+      "loss": 1.1277,
+      "step": 3527
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.044454072790295e-06,
+      "loss": 1.2359,
+      "step": 3528
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.043693871120214e-06,
+      "loss": 1.0039,
+      "step": 3529
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.042933669450134e-06,
+      "loss": 1.0764,
+      "step": 3530
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.042173467780054e-06,
+      "loss": 1.1048,
+      "step": 3531
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.041413266109974e-06,
+      "loss": 1.0473,
+      "step": 3532
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.040653064439894e-06,
+      "loss": 1.0085,
+      "step": 3533
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.039892862769814e-06,
+      "loss": 0.9226,
+      "step": 3534
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.039132661099734e-06,
+      "loss": 1.1394,
+      "step": 3535
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.038372459429653e-06,
+      "loss": 0.8771,
+      "step": 3536
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.037612257759573e-06,
+      "loss": 1.301,
+      "step": 3537
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.036852056089491e-06,
+      "loss": 1.2258,
+      "step": 3538
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.036091854419411e-06,
+      "loss": 1.1577,
+      "step": 3539
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.035331652749331e-06,
+      "loss": 0.9022,
+      "step": 3540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.034571451079251e-06,
+      "loss": 1.0193,
+      "step": 3541
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.03381124940917e-06,
+      "loss": 0.7656,
+      "step": 3542
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.03305104773909e-06,
+      "loss": 1.0626,
+      "step": 3543
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.032290846069009e-06,
+      "loss": 1.0247,
+      "step": 3544
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.031530644398929e-06,
+      "loss": 1.0292,
+      "step": 3545
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.030770442728849e-06,
+      "loss": 1.1635,
+      "step": 3546
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.030010241058768e-06,
+      "loss": 1.064,
+      "step": 3547
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.029250039388688e-06,
+      "loss": 1.087,
+      "step": 3548
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.028489837718608e-06,
+      "loss": 0.831,
+      "step": 3549
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.027729636048528e-06,
+      "loss": 0.9867,
+      "step": 3550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.026969434378448e-06,
+      "loss": 0.9024,
+      "step": 3551
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.026209232708368e-06,
+      "loss": 1.1273,
+      "step": 3552
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.025449031038288e-06,
+      "loss": 1.1275,
+      "step": 3553
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.024688829368206e-06,
+      "loss": 1.0773,
+      "step": 3554
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.023928627698126e-06,
+      "loss": 1.0017,
+      "step": 3555
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.023168426028045e-06,
+      "loss": 1.1158,
+      "step": 3556
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.022408224357965e-06,
+      "loss": 1.0489,
+      "step": 3557
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.021648022687885e-06,
+      "loss": 1.1533,
+      "step": 3558
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.020887821017805e-06,
+      "loss": 1.3572,
+      "step": 3559
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.020127619347723e-06,
+      "loss": 0.9696,
+      "step": 3560
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.019367417677643e-06,
+      "loss": 0.9993,
+      "step": 3561
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.018607216007563e-06,
+      "loss": 1.3237,
+      "step": 3562
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.017847014337483e-06,
+      "loss": 1.0365,
+      "step": 3563
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.017086812667403e-06,
+      "loss": 1.1898,
+      "step": 3564
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.016326610997322e-06,
+      "loss": 1.0679,
+      "step": 3565
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.015566409327242e-06,
+      "loss": 1.1376,
+      "step": 3566
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.014806207657162e-06,
+      "loss": 1.042,
+      "step": 3567
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.014046005987082e-06,
+      "loss": 1.1384,
+      "step": 3568
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.013285804317002e-06,
+      "loss": 1.1542,
+      "step": 3569
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.01252560264692e-06,
+      "loss": 0.9674,
+      "step": 3570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.01176540097684e-06,
+      "loss": 1.0264,
+      "step": 3571
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.01100519930676e-06,
+      "loss": 1.3365,
+      "step": 3572
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.01024499763668e-06,
+      "loss": 1.1381,
+      "step": 3573
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.0094847959666e-06,
+      "loss": 1.0647,
+      "step": 3574
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.00872459429652e-06,
+      "loss": 1.1383,
+      "step": 3575
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.007964392626437e-06,
+      "loss": 1.1788,
+      "step": 3576
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.007204190956357e-06,
+      "loss": 1.2622,
+      "step": 3577
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.006443989286277e-06,
+      "loss": 1.2395,
+      "step": 3578
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.005683787616197e-06,
+      "loss": 1.0875,
+      "step": 3579
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.004923585946117e-06,
+      "loss": 1.2926,
+      "step": 3580
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.004163384276037e-06,
+      "loss": 1.178,
+      "step": 3581
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.003403182605957e-06,
+      "loss": 1.0966,
+      "step": 3582
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.002642980935876e-06,
+      "loss": 1.2873,
+      "step": 3583
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.001882779265796e-06,
+      "loss": 0.9881,
+      "step": 3584
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.001122577595716e-06,
+      "loss": 1.1296,
+      "step": 3585
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.000362375925634e-06,
+      "loss": 1.1042,
+      "step": 3586
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.999602174255554e-06,
+      "loss": 0.9309,
+      "step": 3587
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.998841972585474e-06,
+      "loss": 1.073,
+      "step": 3588
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.998081770915394e-06,
+      "loss": 1.0929,
+      "step": 3589
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.997321569245314e-06,
+      "loss": 1.1718,
+      "step": 3590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.996561367575234e-06,
+      "loss": 0.9543,
+      "step": 3591
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.995801165905152e-06,
+      "loss": 1.1434,
+      "step": 3592
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.995040964235072e-06,
+      "loss": 1.0776,
+      "step": 3593
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.994280762564991e-06,
+      "loss": 1.0902,
+      "step": 3594
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.993520560894911e-06,
+      "loss": 1.1728,
+      "step": 3595
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.992760359224831e-06,
+      "loss": 0.9952,
+      "step": 3596
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.992000157554751e-06,
+      "loss": 1.0382,
+      "step": 3597
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.991239955884671e-06,
+      "loss": 1.1521,
+      "step": 3598
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.99047975421459e-06,
+      "loss": 1.3547,
+      "step": 3599
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.98971955254451e-06,
+      "loss": 1.1298,
+      "step": 3600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.98895935087443e-06,
+      "loss": 1.165,
+      "step": 3601
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.988199149204349e-06,
+      "loss": 1.0319,
+      "step": 3602
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.987438947534268e-06,
+      "loss": 1.086,
+      "step": 3603
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.986678745864188e-06,
+      "loss": 1.0878,
+      "step": 3604
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.985918544194108e-06,
+      "loss": 1.0922,
+      "step": 3605
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.985158342524028e-06,
+      "loss": 1.0866,
+      "step": 3606
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.984398140853948e-06,
+      "loss": 1.0791,
+      "step": 3607
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.983637939183866e-06,
+      "loss": 1.1759,
+      "step": 3608
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.982877737513786e-06,
+      "loss": 1.0593,
+      "step": 3609
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.982117535843706e-06,
+      "loss": 1.2677,
+      "step": 3610
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.981357334173626e-06,
+      "loss": 0.9461,
+      "step": 3611
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.980597132503545e-06,
+      "loss": 1.051,
+      "step": 3612
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.979836930833465e-06,
+      "loss": 0.9783,
+      "step": 3613
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.979076729163385e-06,
+      "loss": 0.9376,
+      "step": 3614
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.978316527493305e-06,
+      "loss": 0.9532,
+      "step": 3615
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.977556325823225e-06,
+      "loss": 1.196,
+      "step": 3616
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.976796124153145e-06,
+      "loss": 0.8535,
+      "step": 3617
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.976035922483063e-06,
+      "loss": 0.9826,
+      "step": 3618
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.975275720812983e-06,
+      "loss": 1.0576,
+      "step": 3619
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.974515519142903e-06,
+      "loss": 1.091,
+      "step": 3620
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.973755317472822e-06,
+      "loss": 0.939,
+      "step": 3621
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.972995115802742e-06,
+      "loss": 1.2671,
+      "step": 3622
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.972234914132662e-06,
+      "loss": 1.1156,
+      "step": 3623
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.97147471246258e-06,
+      "loss": 1.238,
+      "step": 3624
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.9707145107925e-06,
+      "loss": 1.1087,
+      "step": 3625
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.96995430912242e-06,
+      "loss": 1.2005,
+      "step": 3626
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.96919410745234e-06,
+      "loss": 0.9546,
+      "step": 3627
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.96843390578226e-06,
+      "loss": 1.0773,
+      "step": 3628
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.96767370411218e-06,
+      "loss": 1.0835,
+      "step": 3629
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.9669135024421e-06,
+      "loss": 0.9978,
+      "step": 3630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.96615330077202e-06,
+      "loss": 0.7966,
+      "step": 3631
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.96539309910194e-06,
+      "loss": 1.1313,
+      "step": 3632
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.964632897431859e-06,
+      "loss": 1.229,
+      "step": 3633
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.963872695761777e-06,
+      "loss": 1.0742,
+      "step": 3634
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.963112494091697e-06,
+      "loss": 1.0033,
+      "step": 3635
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.962352292421617e-06,
+      "loss": 1.0059,
+      "step": 3636
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.961592090751537e-06,
+      "loss": 1.0678,
+      "step": 3637
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.960831889081457e-06,
+      "loss": 0.9301,
+      "step": 3638
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.960071687411377e-06,
+      "loss": 1.1584,
+      "step": 3639
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.959311485741295e-06,
+      "loss": 1.0646,
+      "step": 3640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.958551284071215e-06,
+      "loss": 1.1868,
+      "step": 3641
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.957791082401134e-06,
+      "loss": 1.0461,
+      "step": 3642
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.957030880731054e-06,
+      "loss": 1.1513,
+      "step": 3643
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.956270679060974e-06,
+      "loss": 1.2141,
+      "step": 3644
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.955510477390894e-06,
+      "loss": 1.2329,
+      "step": 3645
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.954750275720814e-06,
+      "loss": 0.9515,
+      "step": 3646
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.953990074050734e-06,
+      "loss": 0.9607,
+      "step": 3647
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.953229872380654e-06,
+      "loss": 1.295,
+      "step": 3648
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.952469670710573e-06,
+      "loss": 0.9939,
+      "step": 3649
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.951709469040492e-06,
+      "loss": 1.2727,
+      "step": 3650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.950949267370411e-06,
+      "loss": 0.9552,
+      "step": 3651
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.950189065700331e-06,
+      "loss": 0.9656,
+      "step": 3652
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.949428864030251e-06,
+      "loss": 0.9207,
+      "step": 3653
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.948668662360171e-06,
+      "loss": 0.9992,
+      "step": 3654
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.94790846069009e-06,
+      "loss": 1.2181,
+      "step": 3655
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.947148259020009e-06,
+      "loss": 0.9936,
+      "step": 3656
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.946388057349929e-06,
+      "loss": 1.087,
+      "step": 3657
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.945627855679849e-06,
+      "loss": 1.0627,
+      "step": 3658
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.944867654009769e-06,
+      "loss": 1.083,
+      "step": 3659
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.944107452339688e-06,
+      "loss": 0.9038,
+      "step": 3660
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.943347250669608e-06,
+      "loss": 1.2612,
+      "step": 3661
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.942587048999528e-06,
+      "loss": 1.0793,
+      "step": 3662
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.941826847329448e-06,
+      "loss": 1.2309,
+      "step": 3663
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.941066645659368e-06,
+      "loss": 0.9839,
+      "step": 3664
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.940306443989288e-06,
+      "loss": 1.0726,
+      "step": 3665
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.939546242319206e-06,
+      "loss": 0.9924,
+      "step": 3666
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.938786040649126e-06,
+      "loss": 1.0713,
+      "step": 3667
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.938025838979046e-06,
+      "loss": 0.9434,
+      "step": 3668
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.937265637308965e-06,
+      "loss": 1.0081,
+      "step": 3669
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.936505435638885e-06,
+      "loss": 1.238,
+      "step": 3670
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.935745233968803e-06,
+      "loss": 0.9372,
+      "step": 3671
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.934985032298723e-06,
+      "loss": 0.969,
+      "step": 3672
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.934224830628643e-06,
+      "loss": 1.0975,
+      "step": 3673
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.933464628958563e-06,
+      "loss": 1.0618,
+      "step": 3674
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.932704427288483e-06,
+      "loss": 0.9204,
+      "step": 3675
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.931944225618403e-06,
+      "loss": 1.1026,
+      "step": 3676
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.931184023948323e-06,
+      "loss": 1.1254,
+      "step": 3677
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.930423822278242e-06,
+      "loss": 1.1855,
+      "step": 3678
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.929663620608162e-06,
+      "loss": 0.895,
+      "step": 3679
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.928903418938082e-06,
+      "loss": 1.1201,
+      "step": 3680
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.928143217268002e-06,
+      "loss": 1.0898,
+      "step": 3681
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.92738301559792e-06,
+      "loss": 1.0446,
+      "step": 3682
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.92662281392784e-06,
+      "loss": 1.2431,
+      "step": 3683
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.92586261225776e-06,
+      "loss": 1.0728,
+      "step": 3684
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.92510241058768e-06,
+      "loss": 1.0746,
+      "step": 3685
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.9243422089176e-06,
+      "loss": 1.1837,
+      "step": 3686
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.92358200724752e-06,
+      "loss": 1.0586,
+      "step": 3687
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.92282180557744e-06,
+      "loss": 1.0988,
+      "step": 3688
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.922061603907359e-06,
+      "loss": 1.0509,
+      "step": 3689
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.921301402237279e-06,
+      "loss": 0.966,
+      "step": 3690
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.920541200567199e-06,
+      "loss": 1.1039,
+      "step": 3691
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.919780998897119e-06,
+      "loss": 1.0819,
+      "step": 3692
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.919020797227037e-06,
+      "loss": 1.0379,
+      "step": 3693
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.918260595556957e-06,
+      "loss": 0.9295,
+      "step": 3694
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.917500393886877e-06,
+      "loss": 0.9862,
+      "step": 3695
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.916740192216796e-06,
+      "loss": 1.1902,
+      "step": 3696
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.915979990546716e-06,
+      "loss": 1.1085,
+      "step": 3697
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.915219788876634e-06,
+      "loss": 1.0445,
+      "step": 3698
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.914459587206554e-06,
+      "loss": 1.1802,
+      "step": 3699
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.913699385536474e-06,
+      "loss": 1.1013,
+      "step": 3700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.912939183866394e-06,
+      "loss": 1.0752,
+      "step": 3701
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.912178982196314e-06,
+      "loss": 1.0315,
+      "step": 3702
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.911418780526234e-06,
+      "loss": 0.9527,
+      "step": 3703
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.910658578856154e-06,
+      "loss": 0.9173,
+      "step": 3704
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.909898377186073e-06,
+      "loss": 1.0827,
+      "step": 3705
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.909138175515993e-06,
+      "loss": 1.0067,
+      "step": 3706
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.908377973845913e-06,
+      "loss": 1.18,
+      "step": 3707
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.907617772175833e-06,
+      "loss": 0.9121,
+      "step": 3708
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.906857570505751e-06,
+      "loss": 1.0066,
+      "step": 3709
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.906097368835671e-06,
+      "loss": 1.0924,
+      "step": 3710
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.905337167165591e-06,
+      "loss": 1.1521,
+      "step": 3711
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.90457696549551e-06,
+      "loss": 1.1834,
+      "step": 3712
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.90381676382543e-06,
+      "loss": 1.0727,
+      "step": 3713
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.903056562155349e-06,
+      "loss": 1.1517,
+      "step": 3714
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.902296360485269e-06,
+      "loss": 0.9855,
+      "step": 3715
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.901536158815188e-06,
+      "loss": 1.2271,
+      "step": 3716
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.900775957145108e-06,
+      "loss": 1.2739,
+      "step": 3717
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.900015755475028e-06,
+      "loss": 1.1535,
+      "step": 3718
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.899255553804948e-06,
+      "loss": 1.2475,
+      "step": 3719
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.898495352134868e-06,
+      "loss": 1.0482,
+      "step": 3720
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.897735150464788e-06,
+      "loss": 1.1557,
+      "step": 3721
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.896974948794708e-06,
+      "loss": 1.1409,
+      "step": 3722
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.896214747124627e-06,
+      "loss": 1.0348,
+      "step": 3723
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.895454545454547e-06,
+      "loss": 1.299,
+      "step": 3724
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.894694343784465e-06,
+      "loss": 1.0114,
+      "step": 3725
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.893934142114385e-06,
+      "loss": 1.1333,
+      "step": 3726
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.893173940444305e-06,
+      "loss": 1.1275,
+      "step": 3727
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.892413738774225e-06,
+      "loss": 1.2216,
+      "step": 3728
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.891653537104145e-06,
+      "loss": 0.9999,
+      "step": 3729
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.890893335434063e-06,
+      "loss": 1.0721,
+      "step": 3730
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.890133133763983e-06,
+      "loss": 1.1456,
+      "step": 3731
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.889372932093903e-06,
+      "loss": 1.0431,
+      "step": 3732
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.888612730423823e-06,
+      "loss": 1.0402,
+      "step": 3733
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.887852528753742e-06,
+      "loss": 1.0759,
+      "step": 3734
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.887092327083662e-06,
+      "loss": 1.2731,
+      "step": 3735
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.886332125413582e-06,
+      "loss": 0.9917,
+      "step": 3736
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.885571923743502e-06,
+      "loss": 0.9954,
+      "step": 3737
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.884811722073422e-06,
+      "loss": 1.0805,
+      "step": 3738
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.884051520403342e-06,
+      "loss": 0.999,
+      "step": 3739
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.883291318733262e-06,
+      "loss": 1.1005,
+      "step": 3740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.88253111706318e-06,
+      "loss": 1.1847,
+      "step": 3741
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.8817709153931e-06,
+      "loss": 0.9014,
+      "step": 3742
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.88101071372302e-06,
+      "loss": 1.067,
+      "step": 3743
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.88025051205294e-06,
+      "loss": 1.3328,
+      "step": 3744
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.87949031038286e-06,
+      "loss": 1.0352,
+      "step": 3745
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.878730108712777e-06,
+      "loss": 0.948,
+      "step": 3746
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.877969907042697e-06,
+      "loss": 1.128,
+      "step": 3747
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.877209705372617e-06,
+      "loss": 1.2043,
+      "step": 3748
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.876449503702537e-06,
+      "loss": 0.8706,
+      "step": 3749
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.875689302032457e-06,
+      "loss": 1.1405,
+      "step": 3750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.874929100362377e-06,
+      "loss": 1.0839,
+      "step": 3751
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.874168898692297e-06,
+      "loss": 1.3416,
+      "step": 3752
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.873408697022216e-06,
+      "loss": 1.0781,
+      "step": 3753
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.872648495352136e-06,
+      "loss": 1.0014,
+      "step": 3754
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.871888293682056e-06,
+      "loss": 0.9825,
+      "step": 3755
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.871128092011976e-06,
+      "loss": 1.0933,
+      "step": 3756
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.870367890341894e-06,
+      "loss": 1.0377,
+      "step": 3757
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.869607688671814e-06,
+      "loss": 0.9226,
+      "step": 3758
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.868847487001734e-06,
+      "loss": 0.9372,
+      "step": 3759
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.868087285331654e-06,
+      "loss": 1.0402,
+      "step": 3760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.867327083661574e-06,
+      "loss": 0.8613,
+      "step": 3761
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.866566881991492e-06,
+      "loss": 0.897,
+      "step": 3762
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.865806680321412e-06,
+      "loss": 1.2003,
+      "step": 3763
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.865046478651331e-06,
+      "loss": 1.242,
+      "step": 3764
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.864286276981251e-06,
+      "loss": 1.1791,
+      "step": 3765
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.863526075311171e-06,
+      "loss": 0.872,
+      "step": 3766
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.862765873641091e-06,
+      "loss": 0.9929,
+      "step": 3767
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.86200567197101e-06,
+      "loss": 0.8273,
+      "step": 3768
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.86124547030093e-06,
+      "loss": 0.9004,
+      "step": 3769
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.86048526863085e-06,
+      "loss": 0.9023,
+      "step": 3770
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.85972506696077e-06,
+      "loss": 0.859,
+      "step": 3771
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.85896486529069e-06,
+      "loss": 1.1313,
+      "step": 3772
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.858204663620608e-06,
+      "loss": 1.3174,
+      "step": 3773
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.857444461950528e-06,
+      "loss": 1.2615,
+      "step": 3774
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.856684260280448e-06,
+      "loss": 1.1241,
+      "step": 3775
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.855924058610368e-06,
+      "loss": 1.1726,
+      "step": 3776
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.855163856940288e-06,
+      "loss": 1.1112,
+      "step": 3777
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.854403655270206e-06,
+      "loss": 0.8656,
+      "step": 3778
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.853643453600126e-06,
+      "loss": 0.9411,
+      "step": 3779
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.852883251930046e-06,
+      "loss": 1.0796,
+      "step": 3780
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.852123050259966e-06,
+      "loss": 1.0342,
+      "step": 3781
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.851362848589885e-06,
+      "loss": 1.0789,
+      "step": 3782
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.850602646919805e-06,
+      "loss": 1.1314,
+      "step": 3783
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.849842445249725e-06,
+      "loss": 0.8181,
+      "step": 3784
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.849082243579645e-06,
+      "loss": 1.1723,
+      "step": 3785
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.848322041909565e-06,
+      "loss": 1.2024,
+      "step": 3786
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.847561840239485e-06,
+      "loss": 1.1145,
+      "step": 3787
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.846801638569405e-06,
+      "loss": 0.9248,
+      "step": 3788
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.846041436899323e-06,
+      "loss": 1.0667,
+      "step": 3789
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.845281235229243e-06,
+      "loss": 1.0065,
+      "step": 3790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.844521033559162e-06,
+      "loss": 1.1515,
+      "step": 3791
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.843760831889082e-06,
+      "loss": 1.2509,
+      "step": 3792
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.843000630219002e-06,
+      "loss": 1.267,
+      "step": 3793
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.84224042854892e-06,
+      "loss": 0.9928,
+      "step": 3794
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.84148022687884e-06,
+      "loss": 0.9898,
+      "step": 3795
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.84072002520876e-06,
+      "loss": 1.0837,
+      "step": 3796
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.83995982353868e-06,
+      "loss": 1.1474,
+      "step": 3797
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.8391996218686e-06,
+      "loss": 1.1515,
+      "step": 3798
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.83843942019852e-06,
+      "loss": 1.0015,
+      "step": 3799
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.83767921852844e-06,
+      "loss": 1.1799,
+      "step": 3800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.83691901685836e-06,
+      "loss": 1.0017,
+      "step": 3801
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.83615881518828e-06,
+      "loss": 1.1865,
+      "step": 3802
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.835398613518199e-06,
+      "loss": 1.1034,
+      "step": 3803
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.834638411848119e-06,
+      "loss": 1.114,
+      "step": 3804
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.833878210178037e-06,
+      "loss": 1.1203,
+      "step": 3805
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.833118008507957e-06,
+      "loss": 1.0285,
+      "step": 3806
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.832357806837877e-06,
+      "loss": 1.1731,
+      "step": 3807
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.831597605167797e-06,
+      "loss": 1.1142,
+      "step": 3808
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.830837403497716e-06,
+      "loss": 1.0311,
+      "step": 3809
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.830077201827635e-06,
+      "loss": 1.201,
+      "step": 3810
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.829317000157554e-06,
+      "loss": 1.0316,
+      "step": 3811
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.828556798487474e-06,
+      "loss": 0.9688,
+      "step": 3812
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.827796596817394e-06,
+      "loss": 1.1908,
+      "step": 3813
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.827036395147314e-06,
+      "loss": 0.9397,
+      "step": 3814
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.826276193477234e-06,
+      "loss": 1.3851,
+      "step": 3815
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.825515991807154e-06,
+      "loss": 1.1176,
+      "step": 3816
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.824755790137074e-06,
+      "loss": 1.1508,
+      "step": 3817
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.823995588466993e-06,
+      "loss": 1.1297,
+      "step": 3818
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.823235386796913e-06,
+      "loss": 0.9882,
+      "step": 3819
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.822475185126833e-06,
+      "loss": 1.0108,
+      "step": 3820
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.821714983456751e-06,
+      "loss": 1.2161,
+      "step": 3821
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.820954781786671e-06,
+      "loss": 1.0927,
+      "step": 3822
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.820194580116591e-06,
+      "loss": 1.0121,
+      "step": 3823
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.819434378446511e-06,
+      "loss": 0.9496,
+      "step": 3824
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.81867417677643e-06,
+      "loss": 0.9853,
+      "step": 3825
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.817913975106349e-06,
+      "loss": 1.2035,
+      "step": 3826
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.817153773436269e-06,
+      "loss": 0.854,
+      "step": 3827
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.816393571766189e-06,
+      "loss": 1.0171,
+      "step": 3828
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.815633370096108e-06,
+      "loss": 1.0694,
+      "step": 3829
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.814873168426028e-06,
+      "loss": 1.0702,
+      "step": 3830
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.814112966755948e-06,
+      "loss": 0.9764,
+      "step": 3831
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.813352765085868e-06,
+      "loss": 1.0255,
+      "step": 3832
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.812592563415788e-06,
+      "loss": 1.1943,
+      "step": 3833
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.811832361745708e-06,
+      "loss": 0.9654,
+      "step": 3834
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.811072160075628e-06,
+      "loss": 1.0903,
+      "step": 3835
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.810311958405547e-06,
+      "loss": 1.0852,
+      "step": 3836
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.809551756735466e-06,
+      "loss": 1.013,
+      "step": 3837
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.808791555065385e-06,
+      "loss": 1.2913,
+      "step": 3838
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.808031353395305e-06,
+      "loss": 1.11,
+      "step": 3839
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.807271151725225e-06,
+      "loss": 1.0276,
+      "step": 3840
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.806510950055145e-06,
+      "loss": 1.0722,
+      "step": 3841
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.805750748385063e-06,
+      "loss": 1.0402,
+      "step": 3842
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.804990546714983e-06,
+      "loss": 1.1256,
+      "step": 3843
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.804230345044903e-06,
+      "loss": 0.9287,
+      "step": 3844
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.803470143374823e-06,
+      "loss": 0.8882,
+      "step": 3845
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.802709941704743e-06,
+      "loss": 1.1476,
+      "step": 3846
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.801949740034663e-06,
+      "loss": 1.1334,
+      "step": 3847
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.801189538364582e-06,
+      "loss": 1.1261,
+      "step": 3848
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.800429336694502e-06,
+      "loss": 1.0463,
+      "step": 3849
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.799669135024422e-06,
+      "loss": 1.1293,
+      "step": 3850
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.798908933354342e-06,
+      "loss": 1.1724,
+      "step": 3851
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.798148731684262e-06,
+      "loss": 1.076,
+      "step": 3852
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.79738853001418e-06,
+      "loss": 1.0992,
+      "step": 3853
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.7966283283441e-06,
+      "loss": 0.9614,
+      "step": 3854
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.79586812667402e-06,
+      "loss": 0.9979,
+      "step": 3855
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.79510792500394e-06,
+      "loss": 1.1398,
+      "step": 3856
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.79434772333386e-06,
+      "loss": 0.9856,
+      "step": 3857
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.79358752166378e-06,
+      "loss": 1.1477,
+      "step": 3858
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.792827319993697e-06,
+      "loss": 1.0146,
+      "step": 3859
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.792067118323617e-06,
+      "loss": 1.0669,
+      "step": 3860
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.791306916653537e-06,
+      "loss": 1.1507,
+      "step": 3861
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.790546714983457e-06,
+      "loss": 1.1744,
+      "step": 3862
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.789786513313377e-06,
+      "loss": 1.2152,
+      "step": 3863
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.789026311643297e-06,
+      "loss": 0.9696,
+      "step": 3864
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.788266109973217e-06,
+      "loss": 1.0846,
+      "step": 3865
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.787505908303136e-06,
+      "loss": 1.105,
+      "step": 3866
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.786745706633056e-06,
+      "loss": 1.2675,
+      "step": 3867
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.785985504962976e-06,
+      "loss": 1.0521,
+      "step": 3868
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.785225303292894e-06,
+      "loss": 0.8941,
+      "step": 3869
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.784465101622814e-06,
+      "loss": 1.0961,
+      "step": 3870
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.783704899952734e-06,
+      "loss": 0.9852,
+      "step": 3871
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.782944698282654e-06,
+      "loss": 1.1913,
+      "step": 3872
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.782184496612574e-06,
+      "loss": 1.0876,
+      "step": 3873
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.781424294942494e-06,
+      "loss": 0.9072,
+      "step": 3874
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.780664093272413e-06,
+      "loss": 1.2364,
+      "step": 3875
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.779903891602333e-06,
+      "loss": 1.1675,
+      "step": 3876
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.779143689932253e-06,
+      "loss": 1.1267,
+      "step": 3877
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.778383488262173e-06,
+      "loss": 1.1374,
+      "step": 3878
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.777623286592091e-06,
+      "loss": 0.8899,
+      "step": 3879
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.776863084922011e-06,
+      "loss": 1.1993,
+      "step": 3880
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.77610288325193e-06,
+      "loss": 1.1094,
+      "step": 3881
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.77534268158185e-06,
+      "loss": 1.0007,
+      "step": 3882
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.77458247991177e-06,
+      "loss": 1.0847,
+      "step": 3883
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.773822278241689e-06,
+      "loss": 1.2489,
+      "step": 3884
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.773062076571609e-06,
+      "loss": 0.9329,
+      "step": 3885
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.772301874901528e-06,
+      "loss": 1.0784,
+      "step": 3886
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.771541673231448e-06,
+      "loss": 1.0843,
+      "step": 3887
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.770781471561368e-06,
+      "loss": 0.9967,
+      "step": 3888
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.770021269891288e-06,
+      "loss": 1.1309,
+      "step": 3889
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.769261068221208e-06,
+      "loss": 1.0714,
+      "step": 3890
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.768500866551128e-06,
+      "loss": 1.0303,
+      "step": 3891
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.767740664881048e-06,
+      "loss": 1.2767,
+      "step": 3892
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.766980463210967e-06,
+      "loss": 1.034,
+      "step": 3893
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.766220261540887e-06,
+      "loss": 1.3017,
+      "step": 3894
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.765460059870805e-06,
+      "loss": 0.9705,
+      "step": 3895
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.764699858200725e-06,
+      "loss": 1.0002,
+      "step": 3896
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.763939656530645e-06,
+      "loss": 1.0799,
+      "step": 3897
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.763179454860565e-06,
+      "loss": 1.0409,
+      "step": 3898
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.762419253190485e-06,
+      "loss": 1.0005,
+      "step": 3899
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.761659051520403e-06,
+      "loss": 1.1016,
+      "step": 3900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.760898849850323e-06,
+      "loss": 1.2171,
+      "step": 3901
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.760138648180243e-06,
+      "loss": 1.1153,
+      "step": 3902
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.759378446510163e-06,
+      "loss": 1.1939,
+      "step": 3903
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.758618244840082e-06,
+      "loss": 1.0379,
+      "step": 3904
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.757858043170002e-06,
+      "loss": 1.0158,
+      "step": 3905
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.757097841499922e-06,
+      "loss": 1.1941,
+      "step": 3906
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.756337639829842e-06,
+      "loss": 1.0125,
+      "step": 3907
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.755577438159762e-06,
+      "loss": 1.1765,
+      "step": 3908
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.754817236489682e-06,
+      "loss": 1.1127,
+      "step": 3909
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.754057034819602e-06,
+      "loss": 1.0466,
+      "step": 3910
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.75329683314952e-06,
+      "loss": 1.0269,
+      "step": 3911
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.75253663147944e-06,
+      "loss": 1.0126,
+      "step": 3912
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.75177642980936e-06,
+      "loss": 1.1943,
+      "step": 3913
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.75101622813928e-06,
+      "loss": 1.0559,
+      "step": 3914
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.7502560264692e-06,
+      "loss": 1.2082,
+      "step": 3915
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.749495824799117e-06,
+      "loss": 1.2207,
+      "step": 3916
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.748735623129037e-06,
+      "loss": 1.0971,
+      "step": 3917
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.747975421458957e-06,
+      "loss": 1.3552,
+      "step": 3918
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.747215219788877e-06,
+      "loss": 1.0377,
+      "step": 3919
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.746455018118797e-06,
+      "loss": 1.0178,
+      "step": 3920
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.745694816448717e-06,
+      "loss": 0.9142,
+      "step": 3921
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.744934614778636e-06,
+      "loss": 1.0455,
+      "step": 3922
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.744174413108556e-06,
+      "loss": 0.9348,
+      "step": 3923
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.743414211438476e-06,
+      "loss": 1.047,
+      "step": 3924
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.742654009768396e-06,
+      "loss": 1.2996,
+      "step": 3925
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.741893808098316e-06,
+      "loss": 1.1201,
+      "step": 3926
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.741133606428234e-06,
+      "loss": 0.9698,
+      "step": 3927
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.740373404758154e-06,
+      "loss": 0.9032,
+      "step": 3928
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.739613203088074e-06,
+      "loss": 1.0318,
+      "step": 3929
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.738853001417994e-06,
+      "loss": 1.1508,
+      "step": 3930
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.738092799747913e-06,
+      "loss": 1.0746,
+      "step": 3931
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.737332598077832e-06,
+      "loss": 1.0033,
+      "step": 3932
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.736572396407751e-06,
+      "loss": 0.8995,
+      "step": 3933
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.735812194737671e-06,
+      "loss": 1.0565,
+      "step": 3934
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.735051993067591e-06,
+      "loss": 0.8242,
+      "step": 3935
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.734291791397511e-06,
+      "loss": 1.1979,
+      "step": 3936
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.733531589727431e-06,
+      "loss": 1.1168,
+      "step": 3937
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.73277138805735e-06,
+      "loss": 0.875,
+      "step": 3938
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.73201118638727e-06,
+      "loss": 0.9896,
+      "step": 3939
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.73125098471719e-06,
+      "loss": 1.097,
+      "step": 3940
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.73049078304711e-06,
+      "loss": 1.0681,
+      "step": 3941
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.72973058137703e-06,
+      "loss": 0.9429,
+      "step": 3942
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.728970379706948e-06,
+      "loss": 1.0286,
+      "step": 3943
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.728210178036868e-06,
+      "loss": 0.9539,
+      "step": 3944
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.727449976366788e-06,
+      "loss": 1.1195,
+      "step": 3945
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.726689774696708e-06,
+      "loss": 1.2708,
+      "step": 3946
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.725929573026628e-06,
+      "loss": 1.0462,
+      "step": 3947
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.725169371356546e-06,
+      "loss": 1.1321,
+      "step": 3948
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.724409169686466e-06,
+      "loss": 1.1547,
+      "step": 3949
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.723648968016386e-06,
+      "loss": 1.1688,
+      "step": 3950
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.722888766346306e-06,
+      "loss": 1.0038,
+      "step": 3951
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.722128564676225e-06,
+      "loss": 1.0593,
+      "step": 3952
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.721368363006145e-06,
+      "loss": 1.0704,
+      "step": 3953
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.720608161336065e-06,
+      "loss": 1.1391,
+      "step": 3954
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.719847959665985e-06,
+      "loss": 1.2106,
+      "step": 3955
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.719087757995905e-06,
+      "loss": 0.9479,
+      "step": 3956
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.718327556325825e-06,
+      "loss": 1.2366,
+      "step": 3957
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.717567354655745e-06,
+      "loss": 1.0339,
+      "step": 3958
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.716807152985663e-06,
+      "loss": 1.1139,
+      "step": 3959
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.716046951315583e-06,
+      "loss": 1.1226,
+      "step": 3960
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.715286749645502e-06,
+      "loss": 0.7389,
+      "step": 3961
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.714526547975422e-06,
+      "loss": 1.1184,
+      "step": 3962
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.713766346305342e-06,
+      "loss": 1.1401,
+      "step": 3963
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.71300614463526e-06,
+      "loss": 1.0407,
+      "step": 3964
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.71224594296518e-06,
+      "loss": 1.1334,
+      "step": 3965
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.7114857412951e-06,
+      "loss": 0.9255,
+      "step": 3966
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.71072553962502e-06,
+      "loss": 1.2043,
+      "step": 3967
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.70996533795494e-06,
+      "loss": 1.0472,
+      "step": 3968
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.70920513628486e-06,
+      "loss": 1.2166,
+      "step": 3969
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.70844493461478e-06,
+      "loss": 1.117,
+      "step": 3970
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.7076847329447e-06,
+      "loss": 1.0118,
+      "step": 3971
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.706924531274619e-06,
+      "loss": 1.0789,
+      "step": 3972
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.706164329604539e-06,
+      "loss": 1.1358,
+      "step": 3973
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.705404127934459e-06,
+      "loss": 1.0522,
+      "step": 3974
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.704643926264377e-06,
+      "loss": 0.9881,
+      "step": 3975
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.703883724594297e-06,
+      "loss": 1.0086,
+      "step": 3976
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.703123522924217e-06,
+      "loss": 0.9746,
+      "step": 3977
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.702363321254137e-06,
+      "loss": 1.1391,
+      "step": 3978
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.701603119584056e-06,
+      "loss": 0.9505,
+      "step": 3979
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.700842917913975e-06,
+      "loss": 1.1881,
+      "step": 3980
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.700082716243894e-06,
+      "loss": 0.958,
+      "step": 3981
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.699322514573814e-06,
+      "loss": 1.101,
+      "step": 3982
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.698562312903734e-06,
+      "loss": 1.064,
+      "step": 3983
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.697802111233654e-06,
+      "loss": 0.9362,
+      "step": 3984
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.697041909563574e-06,
+      "loss": 1.2086,
+      "step": 3985
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.696281707893494e-06,
+      "loss": 1.1249,
+      "step": 3986
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.695521506223414e-06,
+      "loss": 1.0042,
+      "step": 3987
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.694761304553333e-06,
+      "loss": 1.1507,
+      "step": 3988
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.694001102883253e-06,
+      "loss": 1.0003,
+      "step": 3989
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.693240901213173e-06,
+      "loss": 1.2047,
+      "step": 3990
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.692480699543091e-06,
+      "loss": 1.0969,
+      "step": 3991
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.691720497873011e-06,
+      "loss": 1.1416,
+      "step": 3992
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.690960296202931e-06,
+      "loss": 0.9544,
+      "step": 3993
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.690200094532851e-06,
+      "loss": 0.8992,
+      "step": 3994
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.68943989286277e-06,
+      "loss": 1.2029,
+      "step": 3995
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.688679691192689e-06,
+      "loss": 1.076,
+      "step": 3996
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.687919489522609e-06,
+      "loss": 1.0163,
+      "step": 3997
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.687159287852529e-06,
+      "loss": 1.1586,
+      "step": 3998
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.686399086182448e-06,
+      "loss": 1.1527,
+      "step": 3999
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.685638884512368e-06,
+      "loss": 1.0159,
+      "step": 4000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.684878682842288e-06,
+      "loss": 1.0688,
+      "step": 4001
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.684118481172208e-06,
+      "loss": 1.0989,
+      "step": 4002
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.683358279502128e-06,
+      "loss": 0.9632,
+      "step": 4003
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.682598077832048e-06,
+      "loss": 1.1413,
+      "step": 4004
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.681837876161968e-06,
+      "loss": 1.1687,
+      "step": 4005
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.681077674491887e-06,
+      "loss": 1.1153,
+      "step": 4006
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.680317472821806e-06,
+      "loss": 1.0858,
+      "step": 4007
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.679557271151725e-06,
+      "loss": 0.9796,
+      "step": 4008
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.678797069481645e-06,
+      "loss": 0.9215,
+      "step": 4009
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.678036867811565e-06,
+      "loss": 1.0776,
+      "step": 4010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.677276666141485e-06,
+      "loss": 1.1578,
+      "step": 4011
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.676516464471403e-06,
+      "loss": 1.0569,
+      "step": 4012
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.675756262801323e-06,
+      "loss": 1.1053,
+      "step": 4013
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.674996061131243e-06,
+      "loss": 1.0455,
+      "step": 4014
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.674235859461163e-06,
+      "loss": 1.1941,
+      "step": 4015
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.673475657791083e-06,
+      "loss": 1.0108,
+      "step": 4016
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.672715456121002e-06,
+      "loss": 0.9983,
+      "step": 4017
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.671955254450922e-06,
+      "loss": 1.0699,
+      "step": 4018
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.671195052780842e-06,
+      "loss": 1.168,
+      "step": 4019
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.670434851110762e-06,
+      "loss": 0.806,
+      "step": 4020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.669674649440682e-06,
+      "loss": 1.0036,
+      "step": 4021
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.668914447770602e-06,
+      "loss": 1.0724,
+      "step": 4022
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.66815424610052e-06,
+      "loss": 1.1449,
+      "step": 4023
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.66739404443044e-06,
+      "loss": 1.2481,
+      "step": 4024
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.66663384276036e-06,
+      "loss": 1.0126,
+      "step": 4025
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.66587364109028e-06,
+      "loss": 1.1807,
+      "step": 4026
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.6651134394202e-06,
+      "loss": 1.0768,
+      "step": 4027
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.664353237750117e-06,
+      "loss": 0.9495,
+      "step": 4028
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.663593036080037e-06,
+      "loss": 0.9749,
+      "step": 4029
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.662832834409957e-06,
+      "loss": 0.9951,
+      "step": 4030
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.662072632739877e-06,
+      "loss": 1.0812,
+      "step": 4031
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.661312431069797e-06,
+      "loss": 1.254,
+      "step": 4032
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.660552229399717e-06,
+      "loss": 1.1503,
+      "step": 4033
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.659792027729637e-06,
+      "loss": 1.0169,
+      "step": 4034
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.659031826059556e-06,
+      "loss": 0.9969,
+      "step": 4035
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.658271624389476e-06,
+      "loss": 1.0474,
+      "step": 4036
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.657511422719396e-06,
+      "loss": 1.0918,
+      "step": 4037
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.656751221049316e-06,
+      "loss": 1.2624,
+      "step": 4038
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.655991019379234e-06,
+      "loss": 1.1408,
+      "step": 4039
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.655230817709154e-06,
+      "loss": 1.0464,
+      "step": 4040
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.654470616039074e-06,
+      "loss": 0.9614,
+      "step": 4041
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.653710414368994e-06,
+      "loss": 1.1473,
+      "step": 4042
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.652950212698914e-06,
+      "loss": 1.066,
+      "step": 4043
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.652190011028833e-06,
+      "loss": 1.1469,
+      "step": 4044
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.651429809358753e-06,
+      "loss": 1.1347,
+      "step": 4045
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.650669607688673e-06,
+      "loss": 1.1856,
+      "step": 4046
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.649909406018591e-06,
+      "loss": 1.1406,
+      "step": 4047
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.649149204348511e-06,
+      "loss": 0.9137,
+      "step": 4048
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.648389002678431e-06,
+      "loss": 0.8765,
+      "step": 4049
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.647628801008351e-06,
+      "loss": 1.1684,
+      "step": 4050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.64686859933827e-06,
+      "loss": 1.0908,
+      "step": 4051
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.64610839766819e-06,
+      "loss": 0.9922,
+      "step": 4052
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.64534819599811e-06,
+      "loss": 1.0822,
+      "step": 4053
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.64458799432803e-06,
+      "loss": 1.1139,
+      "step": 4054
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.643827792657949e-06,
+      "loss": 1.2313,
+      "step": 4055
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.643067590987868e-06,
+      "loss": 0.8593,
+      "step": 4056
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.642307389317788e-06,
+      "loss": 1.0846,
+      "step": 4057
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.641547187647708e-06,
+      "loss": 1.13,
+      "step": 4058
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.640786985977628e-06,
+      "loss": 1.1101,
+      "step": 4059
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.640026784307548e-06,
+      "loss": 1.0605,
+      "step": 4060
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.639266582637468e-06,
+      "loss": 1.0193,
+      "step": 4061
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.638506380967388e-06,
+      "loss": 1.2264,
+      "step": 4062
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.637746179297307e-06,
+      "loss": 1.1194,
+      "step": 4063
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.636985977627227e-06,
+      "loss": 1.0076,
+      "step": 4064
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.636225775957147e-06,
+      "loss": 1.1425,
+      "step": 4065
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.635465574287065e-06,
+      "loss": 1.0146,
+      "step": 4066
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.634705372616985e-06,
+      "loss": 1.2265,
+      "step": 4067
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.633945170946905e-06,
+      "loss": 1.1214,
+      "step": 4068
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.633184969276825e-06,
+      "loss": 1.0005,
+      "step": 4069
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.632424767606745e-06,
+      "loss": 1.044,
+      "step": 4070
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.631664565936663e-06,
+      "loss": 0.8272,
+      "step": 4071
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.630904364266583e-06,
+      "loss": 1.1101,
+      "step": 4072
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.630144162596503e-06,
+      "loss": 1.0324,
+      "step": 4073
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.629383960926422e-06,
+      "loss": 1.0129,
+      "step": 4074
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.628623759256342e-06,
+      "loss": 1.0423,
+      "step": 4075
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.627863557586261e-06,
+      "loss": 1.1191,
+      "step": 4076
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.627103355916181e-06,
+      "loss": 0.9889,
+      "step": 4077
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.626343154246101e-06,
+      "loss": 1.1825,
+      "step": 4078
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.625582952576021e-06,
+      "loss": 1.1533,
+      "step": 4079
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.624822750905941e-06,
+      "loss": 1.0695,
+      "step": 4080
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.62406254923586e-06,
+      "loss": 0.9822,
+      "step": 4081
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.6233023475657795e-06,
+      "loss": 1.061,
+      "step": 4082
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.622542145895699e-06,
+      "loss": 1.1417,
+      "step": 4083
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.621781944225619e-06,
+      "loss": 1.1308,
+      "step": 4084
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.621021742555539e-06,
+      "loss": 1.0765,
+      "step": 4085
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.620261540885459e-06,
+      "loss": 1.1735,
+      "step": 4086
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.619501339215378e-06,
+      "loss": 1.1658,
+      "step": 4087
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.618741137545298e-06,
+      "loss": 1.1299,
+      "step": 4088
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.617980935875217e-06,
+      "loss": 1.0795,
+      "step": 4089
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.617220734205137e-06,
+      "loss": 1.1092,
+      "step": 4090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.6164605325350566e-06,
+      "loss": 0.9621,
+      "step": 4091
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.6157003308649756e-06,
+      "loss": 0.8928,
+      "step": 4092
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.614940129194895e-06,
+      "loss": 1.0915,
+      "step": 4093
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.614179927524815e-06,
+      "loss": 1.195,
+      "step": 4094
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.613419725854735e-06,
+      "loss": 1.1469,
+      "step": 4095
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.612659524184655e-06,
+      "loss": 1.0029,
+      "step": 4096
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.611899322514574e-06,
+      "loss": 0.9279,
+      "step": 4097
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.611139120844494e-06,
+      "loss": 0.9912,
+      "step": 4098
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.610378919174414e-06,
+      "loss": 0.9897,
+      "step": 4099
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.6096187175043336e-06,
+      "loss": 1.0242,
+      "step": 4100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.608858515834253e-06,
+      "loss": 1.3288,
+      "step": 4101
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.608098314164173e-06,
+      "loss": 1.0082,
+      "step": 4102
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.607338112494092e-06,
+      "loss": 1.2245,
+      "step": 4103
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.606577910824012e-06,
+      "loss": 0.9602,
+      "step": 4104
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.605817709153932e-06,
+      "loss": 1.0577,
+      "step": 4105
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.605057507483852e-06,
+      "loss": 1.1423,
+      "step": 4106
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.604297305813772e-06,
+      "loss": 1.0262,
+      "step": 4107
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.60353710414369e-06,
+      "loss": 1.1243,
+      "step": 4108
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.60277690247361e-06,
+      "loss": 1.1204,
+      "step": 4109
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.60201670080353e-06,
+      "loss": 1.0071,
+      "step": 4110
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.6012564991334494e-06,
+      "loss": 1.1682,
+      "step": 4111
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.600496297463369e-06,
+      "loss": 0.9609,
+      "step": 4112
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.599736095793288e-06,
+      "loss": 1.1018,
+      "step": 4113
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.598975894123208e-06,
+      "loss": 1.2302,
+      "step": 4114
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.598215692453128e-06,
+      "loss": 0.907,
+      "step": 4115
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.597455490783048e-06,
+      "loss": 1.1266,
+      "step": 4116
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.596695289112968e-06,
+      "loss": 1.1202,
+      "step": 4117
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.595935087442888e-06,
+      "loss": 0.9307,
+      "step": 4118
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.595174885772807e-06,
+      "loss": 0.967,
+      "step": 4119
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.5944146841027264e-06,
+      "loss": 0.9285,
+      "step": 4120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.593654482432646e-06,
+      "loss": 1.0514,
+      "step": 4121
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.592894280762566e-06,
+      "loss": 1.1501,
+      "step": 4122
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.592134079092486e-06,
+      "loss": 1.0533,
+      "step": 4123
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.591373877422404e-06,
+      "loss": 1.0314,
+      "step": 4124
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.590613675752324e-06,
+      "loss": 1.0114,
+      "step": 4125
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.589853474082244e-06,
+      "loss": 0.9911,
+      "step": 4126
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.589093272412164e-06,
+      "loss": 1.064,
+      "step": 4127
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.588333070742084e-06,
+      "loss": 1.1376,
+      "step": 4128
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.587572869072003e-06,
+      "loss": 1.1369,
+      "step": 4129
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.5868126674019225e-06,
+      "loss": 1.1123,
+      "step": 4130
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.586052465731842e-06,
+      "loss": 1.195,
+      "step": 4131
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.585292264061762e-06,
+      "loss": 0.9477,
+      "step": 4132
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.584532062391682e-06,
+      "loss": 1.1814,
+      "step": 4133
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.583771860721602e-06,
+      "loss": 0.9533,
+      "step": 4134
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.583011659051521e-06,
+      "loss": 1.211,
+      "step": 4135
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.582251457381441e-06,
+      "loss": 1.1322,
+      "step": 4136
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.581491255711361e-06,
+      "loss": 0.8686,
+      "step": 4137
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.5807310540412805e-06,
+      "loss": 0.8103,
+      "step": 4138
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.5799708523712e-06,
+      "loss": 1.0757,
+      "step": 4139
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.5792106507011185e-06,
+      "loss": 0.9853,
+      "step": 4140
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.578450449031038e-06,
+      "loss": 1.1673,
+      "step": 4141
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.577690247360958e-06,
+      "loss": 1.039,
+      "step": 4142
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.576930045690878e-06,
+      "loss": 1.1051,
+      "step": 4143
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.576169844020798e-06,
+      "loss": 1.0167,
+      "step": 4144
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.575409642350717e-06,
+      "loss": 1.2024,
+      "step": 4145
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.574649440680637e-06,
+      "loss": 1.2437,
+      "step": 4146
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.573889239010557e-06,
+      "loss": 0.9565,
+      "step": 4147
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.5731290373404765e-06,
+      "loss": 1.0504,
+      "step": 4148
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.572368835670396e-06,
+      "loss": 1.0085,
+      "step": 4149
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.571608634000316e-06,
+      "loss": 1.0141,
+      "step": 4150
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.570848432330235e-06,
+      "loss": 1.1666,
+      "step": 4151
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.570088230660155e-06,
+      "loss": 0.9962,
+      "step": 4152
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.569328028990075e-06,
+      "loss": 1.1383,
+      "step": 4153
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.568567827319995e-06,
+      "loss": 1.0623,
+      "step": 4154
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.567807625649915e-06,
+      "loss": 1.0724,
+      "step": 4155
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.567047423979833e-06,
+      "loss": 1.2747,
+      "step": 4156
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.566287222309753e-06,
+      "loss": 1.0421,
+      "step": 4157
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.5655270206396725e-06,
+      "loss": 1.1462,
+      "step": 4158
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.564766818969592e-06,
+      "loss": 1.0999,
+      "step": 4159
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.564006617299512e-06,
+      "loss": 1.0577,
+      "step": 4160
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.563246415629431e-06,
+      "loss": 1.2047,
+      "step": 4161
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.562486213959351e-06,
+      "loss": 1.1344,
+      "step": 4162
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.561726012289271e-06,
+      "loss": 1.0787,
+      "step": 4163
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.560965810619191e-06,
+      "loss": 1.3066,
+      "step": 4164
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.560205608949111e-06,
+      "loss": 1.0463,
+      "step": 4165
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.5594454072790305e-06,
+      "loss": 1.0185,
+      "step": 4166
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.5586852056089495e-06,
+      "loss": 1.0904,
+      "step": 4167
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.557925003938869e-06,
+      "loss": 0.8987,
+      "step": 4168
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.557164802268789e-06,
+      "loss": 1.1628,
+      "step": 4169
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.556404600598709e-06,
+      "loss": 1.1202,
+      "step": 4170
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.555644398928629e-06,
+      "loss": 0.8543,
+      "step": 4171
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.554884197258547e-06,
+      "loss": 1.0963,
+      "step": 4172
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.554123995588467e-06,
+      "loss": 0.9792,
+      "step": 4173
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.553363793918387e-06,
+      "loss": 1.0335,
+      "step": 4174
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.552603592248307e-06,
+      "loss": 0.9477,
+      "step": 4175
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.5518433905782265e-06,
+      "loss": 1.1242,
+      "step": 4176
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.5510831889081455e-06,
+      "loss": 1.1722,
+      "step": 4177
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.550322987238065e-06,
+      "loss": 1.0236,
+      "step": 4178
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.549562785567985e-06,
+      "loss": 1.2354,
+      "step": 4179
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.548802583897905e-06,
+      "loss": 0.9575,
+      "step": 4180
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.548042382227825e-06,
+      "loss": 1.1853,
+      "step": 4181
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.547282180557745e-06,
+      "loss": 0.9115,
+      "step": 4182
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.546521978887664e-06,
+      "loss": 0.8466,
+      "step": 4183
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.545761777217584e-06,
+      "loss": 0.9363,
+      "step": 4184
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.5450015755475035e-06,
+      "loss": 1.0599,
+      "step": 4185
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.544241373877423e-06,
+      "loss": 0.8598,
+      "step": 4186
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.543481172207343e-06,
+      "loss": 1.2815,
+      "step": 4187
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.542720970537262e-06,
+      "loss": 1.1401,
+      "step": 4188
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.541960768867182e-06,
+      "loss": 0.9966,
+      "step": 4189
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.541200567197102e-06,
+      "loss": 1.1345,
+      "step": 4190
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.540440365527022e-06,
+      "loss": 1.101,
+      "step": 4191
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.539680163856942e-06,
+      "loss": 1.185,
+      "step": 4192
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.53891996218686e-06,
+      "loss": 1.1023,
+      "step": 4193
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.53815976051678e-06,
+      "loss": 0.9226,
+      "step": 4194
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.5373995588466996e-06,
+      "loss": 1.0227,
+      "step": 4195
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.536639357176619e-06,
+      "loss": 1.11,
+      "step": 4196
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.535879155506539e-06,
+      "loss": 1.0523,
+      "step": 4197
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.535118953836459e-06,
+      "loss": 1.2154,
+      "step": 4198
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.534358752166378e-06,
+      "loss": 0.9062,
+      "step": 4199
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.533598550496298e-06,
+      "loss": 0.9916,
+      "step": 4200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.532838348826218e-06,
+      "loss": 1.1239,
+      "step": 4201
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.532078147156138e-06,
+      "loss": 0.9925,
+      "step": 4202
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.5313179454860576e-06,
+      "loss": 1.0188,
+      "step": 4203
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.5305577438159766e-06,
+      "loss": 1.1716,
+      "step": 4204
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.529797542145896e-06,
+      "loss": 1.0952,
+      "step": 4205
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.529037340475816e-06,
+      "loss": 0.8623,
+      "step": 4206
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.528277138805736e-06,
+      "loss": 0.9838,
+      "step": 4207
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.527516937135656e-06,
+      "loss": 1.212,
+      "step": 4208
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.526756735465574e-06,
+      "loss": 1.1218,
+      "step": 4209
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.525996533795494e-06,
+      "loss": 1.0272,
+      "step": 4210
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.525236332125414e-06,
+      "loss": 1.119,
+      "step": 4211
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.524476130455334e-06,
+      "loss": 0.9954,
+      "step": 4212
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.5237159287852536e-06,
+      "loss": 0.9107,
+      "step": 4213
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.5229557271151734e-06,
+      "loss": 1.113,
+      "step": 4214
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.5221955254450924e-06,
+      "loss": 1.1023,
+      "step": 4215
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.521435323775012e-06,
+      "loss": 1.1662,
+      "step": 4216
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.520675122104932e-06,
+      "loss": 1.0138,
+      "step": 4217
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.519914920434852e-06,
+      "loss": 1.0192,
+      "step": 4218
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.519154718764772e-06,
+      "loss": 1.0184,
+      "step": 4219
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.518394517094691e-06,
+      "loss": 1.1574,
+      "step": 4220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.517634315424611e-06,
+      "loss": 1.123,
+      "step": 4221
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.516874113754531e-06,
+      "loss": 1.1221,
+      "step": 4222
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.5161139120844504e-06,
+      "loss": 1.1715,
+      "step": 4223
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.51535371041437e-06,
+      "loss": 1.1051,
+      "step": 4224
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.5145935087442885e-06,
+      "loss": 1.065,
+      "step": 4225
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.513833307074208e-06,
+      "loss": 1.2629,
+      "step": 4226
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.513073105404128e-06,
+      "loss": 1.135,
+      "step": 4227
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.512312903734048e-06,
+      "loss": 1.1353,
+      "step": 4228
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.511552702063968e-06,
+      "loss": 1.0912,
+      "step": 4229
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.510792500393888e-06,
+      "loss": 1.0759,
+      "step": 4230
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.510032298723807e-06,
+      "loss": 1.2152,
+      "step": 4231
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.509272097053727e-06,
+      "loss": 1.0533,
+      "step": 4232
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.5085118953836465e-06,
+      "loss": 1.1467,
+      "step": 4233
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.507751693713566e-06,
+      "loss": 1.0064,
+      "step": 4234
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.506991492043486e-06,
+      "loss": 1.0807,
+      "step": 4235
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.506231290373405e-06,
+      "loss": 1.1941,
+      "step": 4236
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.505471088703325e-06,
+      "loss": 1.1505,
+      "step": 4237
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.504710887033245e-06,
+      "loss": 1.0399,
+      "step": 4238
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.503950685363165e-06,
+      "loss": 1.1588,
+      "step": 4239
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.503190483693085e-06,
+      "loss": 1.0784,
+      "step": 4240
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.502430282023003e-06,
+      "loss": 1.1045,
+      "step": 4241
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.501670080352923e-06,
+      "loss": 1.0324,
+      "step": 4242
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.5009098786828425e-06,
+      "loss": 1.1813,
+      "step": 4243
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.500149677012762e-06,
+      "loss": 1.3209,
+      "step": 4244
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.499389475342682e-06,
+      "loss": 1.0396,
+      "step": 4245
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.498629273672602e-06,
+      "loss": 1.0159,
+      "step": 4246
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.497869072002521e-06,
+      "loss": 1.0109,
+      "step": 4247
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.497108870332441e-06,
+      "loss": 1.1683,
+      "step": 4248
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.496348668662361e-06,
+      "loss": 1.0719,
+      "step": 4249
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.495588466992281e-06,
+      "loss": 1.0594,
+      "step": 4250
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.4948282653222005e-06,
+      "loss": 0.9899,
+      "step": 4251
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.4940680636521195e-06,
+      "loss": 1.1607,
+      "step": 4252
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.493307861982039e-06,
+      "loss": 0.8849,
+      "step": 4253
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.492547660311959e-06,
+      "loss": 1.091,
+      "step": 4254
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.491787458641879e-06,
+      "loss": 1.2144,
+      "step": 4255
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.491027256971799e-06,
+      "loss": 1.0105,
+      "step": 4256
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.490267055301717e-06,
+      "loss": 1.1057,
+      "step": 4257
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.489506853631637e-06,
+      "loss": 1.1929,
+      "step": 4258
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.488746651961557e-06,
+      "loss": 0.9273,
+      "step": 4259
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.487986450291477e-06,
+      "loss": 1.005,
+      "step": 4260
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.4872262486213965e-06,
+      "loss": 0.8775,
+      "step": 4261
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.486466046951316e-06,
+      "loss": 1.0398,
+      "step": 4262
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.485705845281235e-06,
+      "loss": 1.1674,
+      "step": 4263
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.484945643611155e-06,
+      "loss": 1.0384,
+      "step": 4264
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.484185441941075e-06,
+      "loss": 1.0702,
+      "step": 4265
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.483425240270995e-06,
+      "loss": 0.9926,
+      "step": 4266
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.482665038600915e-06,
+      "loss": 1.0787,
+      "step": 4267
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.481904836930834e-06,
+      "loss": 0.8879,
+      "step": 4268
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.481144635260754e-06,
+      "loss": 1.2158,
+      "step": 4269
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.4803844335906735e-06,
+      "loss": 1.082,
+      "step": 4270
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.479624231920593e-06,
+      "loss": 1.0759,
+      "step": 4271
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.478864030250513e-06,
+      "loss": 0.9869,
+      "step": 4272
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.478103828580432e-06,
+      "loss": 1.0794,
+      "step": 4273
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.477343626910352e-06,
+      "loss": 1.0566,
+      "step": 4274
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.476583425240272e-06,
+      "loss": 1.1156,
+      "step": 4275
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.475823223570192e-06,
+      "loss": 1.0541,
+      "step": 4276
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.475063021900111e-06,
+      "loss": 0.9999,
+      "step": 4277
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.474302820230031e-06,
+      "loss": 1.0656,
+      "step": 4278
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.47354261855995e-06,
+      "loss": 1.1471,
+      "step": 4279
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.4727824168898695e-06,
+      "loss": 1.1506,
+      "step": 4280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.472022215219789e-06,
+      "loss": 0.8885,
+      "step": 4281
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.471262013549709e-06,
+      "loss": 1.0869,
+      "step": 4282
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.470501811879629e-06,
+      "loss": 0.9004,
+      "step": 4283
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.469741610209548e-06,
+      "loss": 1.2129,
+      "step": 4284
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.468981408539468e-06,
+      "loss": 0.9864,
+      "step": 4285
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.468221206869388e-06,
+      "loss": 1.0796,
+      "step": 4286
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.467461005199308e-06,
+      "loss": 0.998,
+      "step": 4287
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.4667008035292275e-06,
+      "loss": 1.1827,
+      "step": 4288
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.4659406018591465e-06,
+      "loss": 1.057,
+      "step": 4289
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.465180400189066e-06,
+      "loss": 1.1068,
+      "step": 4290
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.464420198518986e-06,
+      "loss": 0.8567,
+      "step": 4291
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.463659996848906e-06,
+      "loss": 1.2422,
+      "step": 4292
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.462899795178826e-06,
+      "loss": 1.0216,
+      "step": 4293
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.462139593508744e-06,
+      "loss": 1.1973,
+      "step": 4294
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.461379391838664e-06,
+      "loss": 1.0982,
+      "step": 4295
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.460619190168584e-06,
+      "loss": 1.0151,
+      "step": 4296
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.459858988498504e-06,
+      "loss": 1.1324,
+      "step": 4297
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.4590987868284235e-06,
+      "loss": 1.0095,
+      "step": 4298
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.458338585158343e-06,
+      "loss": 0.8114,
+      "step": 4299
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.457578383488262e-06,
+      "loss": 1.1137,
+      "step": 4300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.456818181818182e-06,
+      "loss": 0.9887,
+      "step": 4301
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.456057980148102e-06,
+      "loss": 0.8974,
+      "step": 4302
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.455297778478022e-06,
+      "loss": 1.1848,
+      "step": 4303
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.454537576807942e-06,
+      "loss": 1.193,
+      "step": 4304
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.453777375137861e-06,
+      "loss": 0.9465,
+      "step": 4305
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.453017173467781e-06,
+      "loss": 1.2065,
+      "step": 4306
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.4522569717977006e-06,
+      "loss": 1.0285,
+      "step": 4307
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.45149677012762e-06,
+      "loss": 1.0085,
+      "step": 4308
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.45073656845754e-06,
+      "loss": 0.9235,
+      "step": 4309
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.4499763667874584e-06,
+      "loss": 1.0898,
+      "step": 4310
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.449216165117378e-06,
+      "loss": 1.2539,
+      "step": 4311
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.448455963447298e-06,
+      "loss": 1.14,
+      "step": 4312
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.447695761777218e-06,
+      "loss": 0.9919,
+      "step": 4313
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.446935560107138e-06,
+      "loss": 1.1141,
+      "step": 4314
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.446175358437058e-06,
+      "loss": 1.0247,
+      "step": 4315
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.445415156766977e-06,
+      "loss": 1.045,
+      "step": 4316
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.4446549550968966e-06,
+      "loss": 1.122,
+      "step": 4317
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.4438947534268164e-06,
+      "loss": 1.1022,
+      "step": 4318
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.443134551756736e-06,
+      "loss": 1.2457,
+      "step": 4319
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.442374350086656e-06,
+      "loss": 1.1524,
+      "step": 4320
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.441614148416575e-06,
+      "loss": 1.0104,
+      "step": 4321
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.440853946746495e-06,
+      "loss": 1.0932,
+      "step": 4322
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.440093745076415e-06,
+      "loss": 1.2436,
+      "step": 4323
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.439333543406335e-06,
+      "loss": 1.1446,
+      "step": 4324
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.438573341736255e-06,
+      "loss": 1.0544,
+      "step": 4325
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.437813140066173e-06,
+      "loss": 1.0678,
+      "step": 4326
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.437052938396093e-06,
+      "loss": 0.9798,
+      "step": 4327
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.4362927367260125e-06,
+      "loss": 1.0205,
+      "step": 4328
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.435532535055932e-06,
+      "loss": 1.0016,
+      "step": 4329
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.434772333385852e-06,
+      "loss": 1.242,
+      "step": 4330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.434012131715772e-06,
+      "loss": 1.0467,
+      "step": 4331
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.433251930045691e-06,
+      "loss": 1.0069,
+      "step": 4332
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.432491728375611e-06,
+      "loss": 0.8586,
+      "step": 4333
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.431731526705531e-06,
+      "loss": 1.423,
+      "step": 4334
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.430971325035451e-06,
+      "loss": 0.9538,
+      "step": 4335
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.4302111233653705e-06,
+      "loss": 1.0792,
+      "step": 4336
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.4294509216952895e-06,
+      "loss": 1.0613,
+      "step": 4337
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.428690720025209e-06,
+      "loss": 0.9488,
+      "step": 4338
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.427930518355129e-06,
+      "loss": 1.1962,
+      "step": 4339
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.427170316685049e-06,
+      "loss": 1.1268,
+      "step": 4340
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.426410115014969e-06,
+      "loss": 1.0495,
+      "step": 4341
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.425649913344887e-06,
+      "loss": 1.0708,
+      "step": 4342
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.424889711674807e-06,
+      "loss": 1.0483,
+      "step": 4343
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.424129510004727e-06,
+      "loss": 1.2986,
+      "step": 4344
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.423369308334647e-06,
+      "loss": 1.2047,
+      "step": 4345
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.4226091066645665e-06,
+      "loss": 0.9333,
+      "step": 4346
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.421848904994486e-06,
+      "loss": 1.1256,
+      "step": 4347
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.421088703324405e-06,
+      "loss": 1.1744,
+      "step": 4348
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.420328501654325e-06,
+      "loss": 0.9219,
+      "step": 4349
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.419568299984245e-06,
+      "loss": 1.1677,
+      "step": 4350
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.418808098314165e-06,
+      "loss": 0.9773,
+      "step": 4351
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.418047896644085e-06,
+      "loss": 0.9375,
+      "step": 4352
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.417287694974004e-06,
+      "loss": 1.1286,
+      "step": 4353
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.416527493303924e-06,
+      "loss": 1.1855,
+      "step": 4354
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.4157672916338435e-06,
+      "loss": 1.0348,
+      "step": 4355
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.415007089963763e-06,
+      "loss": 0.8892,
+      "step": 4356
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.414246888293683e-06,
+      "loss": 1.0635,
+      "step": 4357
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.413486686623602e-06,
+      "loss": 0.8683,
+      "step": 4358
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.412726484953522e-06,
+      "loss": 0.7415,
+      "step": 4359
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.411966283283441e-06,
+      "loss": 1.1466,
+      "step": 4360
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.411206081613361e-06,
+      "loss": 1.1589,
+      "step": 4361
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.410445879943281e-06,
+      "loss": 1.1186,
+      "step": 4362
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.409685678273201e-06,
+      "loss": 1.039,
+      "step": 4363
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.40892547660312e-06,
+      "loss": 1.0917,
+      "step": 4364
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.4081652749330395e-06,
+      "loss": 1.1813,
+      "step": 4365
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.407405073262959e-06,
+      "loss": 1.1258,
+      "step": 4366
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.406644871592879e-06,
+      "loss": 1.1231,
+      "step": 4367
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.405884669922799e-06,
+      "loss": 1.084,
+      "step": 4368
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.405124468252718e-06,
+      "loss": 0.958,
+      "step": 4369
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.404364266582638e-06,
+      "loss": 1.0814,
+      "step": 4370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.403604064912558e-06,
+      "loss": 1.0089,
+      "step": 4371
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.402843863242478e-06,
+      "loss": 1.0861,
+      "step": 4372
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.4020836615723975e-06,
+      "loss": 0.9398,
+      "step": 4373
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.4013234599023165e-06,
+      "loss": 1.2397,
+      "step": 4374
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.400563258232236e-06,
+      "loss": 0.9224,
+      "step": 4375
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.399803056562156e-06,
+      "loss": 1.0687,
+      "step": 4376
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.399042854892076e-06,
+      "loss": 1.0024,
+      "step": 4377
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.398282653221996e-06,
+      "loss": 1.1021,
+      "step": 4378
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.397522451551916e-06,
+      "loss": 1.042,
+      "step": 4379
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.396762249881834e-06,
+      "loss": 0.9896,
+      "step": 4380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.396002048211754e-06,
+      "loss": 1.0193,
+      "step": 4381
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.395241846541674e-06,
+      "loss": 0.8975,
+      "step": 4382
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.3944816448715935e-06,
+      "loss": 1.2832,
+      "step": 4383
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.393721443201513e-06,
+      "loss": 0.9599,
+      "step": 4384
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.392961241531432e-06,
+      "loss": 1.0361,
+      "step": 4385
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.392201039861352e-06,
+      "loss": 1.0944,
+      "step": 4386
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.391440838191272e-06,
+      "loss": 1.1471,
+      "step": 4387
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.390680636521192e-06,
+      "loss": 0.9374,
+      "step": 4388
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.389920434851112e-06,
+      "loss": 1.1431,
+      "step": 4389
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.389160233181031e-06,
+      "loss": 1.1983,
+      "step": 4390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.388400031510951e-06,
+      "loss": 0.9339,
+      "step": 4391
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.3876398298408705e-06,
+      "loss": 1.0082,
+      "step": 4392
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.38687962817079e-06,
+      "loss": 0.9937,
+      "step": 4393
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.38611942650071e-06,
+      "loss": 1.131,
+      "step": 4394
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.38535922483063e-06,
+      "loss": 1.1018,
+      "step": 4395
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.384599023160548e-06,
+      "loss": 0.9425,
+      "step": 4396
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.383838821490468e-06,
+      "loss": 1.254,
+      "step": 4397
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.383078619820388e-06,
+      "loss": 1.0556,
+      "step": 4398
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.382318418150308e-06,
+      "loss": 1.0858,
+      "step": 4399
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.381558216480228e-06,
+      "loss": 1.2476,
+      "step": 4400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.380798014810147e-06,
+      "loss": 1.0624,
+      "step": 4401
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.3800378131400665e-06,
+      "loss": 0.9351,
+      "step": 4402
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.379277611469986e-06,
+      "loss": 1.2213,
+      "step": 4403
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.378517409799906e-06,
+      "loss": 1.1335,
+      "step": 4404
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.377757208129826e-06,
+      "loss": 1.1402,
+      "step": 4405
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.376997006459745e-06,
+      "loss": 0.9916,
+      "step": 4406
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.376236804789665e-06,
+      "loss": 1.0809,
+      "step": 4407
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.375476603119585e-06,
+      "loss": 1.0205,
+      "step": 4408
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.374716401449505e-06,
+      "loss": 0.9902,
+      "step": 4409
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.3739561997794245e-06,
+      "loss": 0.8812,
+      "step": 4410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.373195998109344e-06,
+      "loss": 1.045,
+      "step": 4411
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.3724357964392626e-06,
+      "loss": 1.2936,
+      "step": 4412
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.371675594769182e-06,
+      "loss": 1.1942,
+      "step": 4413
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.370915393099102e-06,
+      "loss": 0.9919,
+      "step": 4414
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.370155191429022e-06,
+      "loss": 1.2393,
+      "step": 4415
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.369394989758942e-06,
+      "loss": 0.9614,
+      "step": 4416
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.368634788088861e-06,
+      "loss": 1.2497,
+      "step": 4417
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.367874586418781e-06,
+      "loss": 0.8957,
+      "step": 4418
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.367114384748701e-06,
+      "loss": 1.1749,
+      "step": 4419
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.3663541830786206e-06,
+      "loss": 1.0291,
+      "step": 4420
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.3655939814085404e-06,
+      "loss": 0.9736,
+      "step": 4421
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.3648337797384594e-06,
+      "loss": 1.1178,
+      "step": 4422
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.364073578068379e-06,
+      "loss": 1.0554,
+      "step": 4423
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.363313376398299e-06,
+      "loss": 1.0401,
+      "step": 4424
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.362553174728219e-06,
+      "loss": 0.9738,
+      "step": 4425
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.361792973058139e-06,
+      "loss": 1.0257,
+      "step": 4426
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.361032771388059e-06,
+      "loss": 0.9644,
+      "step": 4427
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.360272569717977e-06,
+      "loss": 1.1169,
+      "step": 4428
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.359512368047897e-06,
+      "loss": 1.0603,
+      "step": 4429
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.358752166377817e-06,
+      "loss": 1.0024,
+      "step": 4430
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.3579919647077364e-06,
+      "loss": 0.9938,
+      "step": 4431
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.357231763037656e-06,
+      "loss": 1.0638,
+      "step": 4432
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.356471561367575e-06,
+      "loss": 1.0207,
+      "step": 4433
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.355711359697495e-06,
+      "loss": 1.2033,
+      "step": 4434
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.354951158027415e-06,
+      "loss": 1.1152,
+      "step": 4435
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.354190956357335e-06,
+      "loss": 1.0057,
+      "step": 4436
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.353430754687255e-06,
+      "loss": 0.9801,
+      "step": 4437
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.352670553017174e-06,
+      "loss": 1.1508,
+      "step": 4438
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.351910351347094e-06,
+      "loss": 1.0048,
+      "step": 4439
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.3511501496770135e-06,
+      "loss": 1.2979,
+      "step": 4440
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.350389948006933e-06,
+      "loss": 1.1098,
+      "step": 4441
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.349629746336853e-06,
+      "loss": 0.9643,
+      "step": 4442
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.348869544666773e-06,
+      "loss": 1.0377,
+      "step": 4443
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.348109342996691e-06,
+      "loss": 1.0308,
+      "step": 4444
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.347349141326611e-06,
+      "loss": 1.0396,
+      "step": 4445
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.346588939656531e-06,
+      "loss": 1.312,
+      "step": 4446
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.345828737986451e-06,
+      "loss": 1.0252,
+      "step": 4447
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.345068536316371e-06,
+      "loss": 1.1136,
+      "step": 4448
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.34430833464629e-06,
+      "loss": 1.1671,
+      "step": 4449
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.3435481329762095e-06,
+      "loss": 1.0787,
+      "step": 4450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.342787931306129e-06,
+      "loss": 1.1209,
+      "step": 4451
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.342027729636049e-06,
+      "loss": 1.1419,
+      "step": 4452
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.341267527965969e-06,
+      "loss": 1.25,
+      "step": 4453
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.340507326295888e-06,
+      "loss": 1.0927,
+      "step": 4454
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.339747124625808e-06,
+      "loss": 0.9686,
+      "step": 4455
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.338986922955728e-06,
+      "loss": 1.1686,
+      "step": 4456
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.338226721285648e-06,
+      "loss": 1.0838,
+      "step": 4457
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.3374665196155675e-06,
+      "loss": 1.0929,
+      "step": 4458
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.336706317945487e-06,
+      "loss": 0.9825,
+      "step": 4459
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.335946116275406e-06,
+      "loss": 0.9934,
+      "step": 4460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.335185914605326e-06,
+      "loss": 1.2145,
+      "step": 4461
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.334425712935246e-06,
+      "loss": 1.0257,
+      "step": 4462
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.333665511265166e-06,
+      "loss": 1.1634,
+      "step": 4463
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.332905309595086e-06,
+      "loss": 1.2357,
+      "step": 4464
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.332145107925004e-06,
+      "loss": 1.1754,
+      "step": 4465
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.331384906254924e-06,
+      "loss": 1.167,
+      "step": 4466
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.330624704584844e-06,
+      "loss": 1.1173,
+      "step": 4467
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.3298645029147635e-06,
+      "loss": 1.0783,
+      "step": 4468
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.329104301244683e-06,
+      "loss": 0.9839,
+      "step": 4469
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.328344099574602e-06,
+      "loss": 1.1384,
+      "step": 4470
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.327583897904522e-06,
+      "loss": 0.9745,
+      "step": 4471
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.326823696234442e-06,
+      "loss": 1.2734,
+      "step": 4472
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.326063494564362e-06,
+      "loss": 1.111,
+      "step": 4473
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.325303292894282e-06,
+      "loss": 1.0695,
+      "step": 4474
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.324543091224202e-06,
+      "loss": 0.9631,
+      "step": 4475
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.323782889554121e-06,
+      "loss": 1.0684,
+      "step": 4476
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.3230226878840405e-06,
+      "loss": 1.1077,
+      "step": 4477
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.32226248621396e-06,
+      "loss": 0.9891,
+      "step": 4478
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.32150228454388e-06,
+      "loss": 0.9373,
+      "step": 4479
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.3207420828738e-06,
+      "loss": 0.9937,
+      "step": 4480
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.319981881203718e-06,
+      "loss": 0.9762,
+      "step": 4481
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.319221679533638e-06,
+      "loss": 1.2824,
+      "step": 4482
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.318461477863558e-06,
+      "loss": 1.0602,
+      "step": 4483
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.317701276193478e-06,
+      "loss": 1.0952,
+      "step": 4484
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.316941074523398e-06,
+      "loss": 1.1946,
+      "step": 4485
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.316180872853317e-06,
+      "loss": 1.0116,
+      "step": 4486
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.3154206711832365e-06,
+      "loss": 1.271,
+      "step": 4487
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.314660469513156e-06,
+      "loss": 1.2197,
+      "step": 4488
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.313900267843076e-06,
+      "loss": 1.0027,
+      "step": 4489
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.313140066172996e-06,
+      "loss": 1.0327,
+      "step": 4490
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.312379864502915e-06,
+      "loss": 0.9755,
+      "step": 4491
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.311619662832835e-06,
+      "loss": 1.139,
+      "step": 4492
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.310859461162755e-06,
+      "loss": 1.0028,
+      "step": 4493
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.310099259492675e-06,
+      "loss": 0.9951,
+      "step": 4494
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.3093390578225945e-06,
+      "loss": 1.1346,
+      "step": 4495
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.308578856152514e-06,
+      "loss": 1.2053,
+      "step": 4496
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.3078186544824325e-06,
+      "loss": 0.9382,
+      "step": 4497
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.307058452812352e-06,
+      "loss": 0.992,
+      "step": 4498
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.306298251142272e-06,
+      "loss": 1.0657,
+      "step": 4499
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.305538049472192e-06,
+      "loss": 1.0784,
+      "step": 4500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.304777847802112e-06,
+      "loss": 1.1472,
+      "step": 4501
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.304017646132031e-06,
+      "loss": 1.0285,
+      "step": 4502
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.303257444461951e-06,
+      "loss": 1.0842,
+      "step": 4503
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.302497242791871e-06,
+      "loss": 0.9784,
+      "step": 4504
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.3017370411217905e-06,
+      "loss": 1.2029,
+      "step": 4505
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.30097683945171e-06,
+      "loss": 0.9324,
+      "step": 4506
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.300216637781629e-06,
+      "loss": 1.2686,
+      "step": 4507
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.299456436111549e-06,
+      "loss": 0.9387,
+      "step": 4508
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.298696234441469e-06,
+      "loss": 1.175,
+      "step": 4509
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.297936032771389e-06,
+      "loss": 1.185,
+      "step": 4510
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.297175831101309e-06,
+      "loss": 0.9831,
+      "step": 4511
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.296415629431229e-06,
+      "loss": 1.1843,
+      "step": 4512
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.295655427761147e-06,
+      "loss": 1.0739,
+      "step": 4513
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.294895226091067e-06,
+      "loss": 1.0755,
+      "step": 4514
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.2941350244209866e-06,
+      "loss": 1.2647,
+      "step": 4515
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.293374822750906e-06,
+      "loss": 1.0646,
+      "step": 4516
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.292614621080826e-06,
+      "loss": 1.0122,
+      "step": 4517
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.291854419410745e-06,
+      "loss": 0.9387,
+      "step": 4518
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.291094217740665e-06,
+      "loss": 1.1648,
+      "step": 4519
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.290334016070585e-06,
+      "loss": 1.0088,
+      "step": 4520
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.289573814400505e-06,
+      "loss": 1.1927,
+      "step": 4521
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.288813612730425e-06,
+      "loss": 1.1385,
+      "step": 4522
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.288053411060344e-06,
+      "loss": 1.1385,
+      "step": 4523
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.2872932093902636e-06,
+      "loss": 1.0908,
+      "step": 4524
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.2865330077201834e-06,
+      "loss": 1.2553,
+      "step": 4525
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.285772806050103e-06,
+      "loss": 1.1402,
+      "step": 4526
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.285012604380023e-06,
+      "loss": 1.1097,
+      "step": 4527
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.284252402709943e-06,
+      "loss": 1.0054,
+      "step": 4528
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.283492201039861e-06,
+      "loss": 1.049,
+      "step": 4529
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.282731999369781e-06,
+      "loss": 1.0619,
+      "step": 4530
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.281971797699701e-06,
+      "loss": 1.0369,
+      "step": 4531
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.281211596029621e-06,
+      "loss": 1.1325,
+      "step": 4532
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.280451394359541e-06,
+      "loss": 1.188,
+      "step": 4533
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.27969119268946e-06,
+      "loss": 1.0733,
+      "step": 4534
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.2789309910193794e-06,
+      "loss": 0.9448,
+      "step": 4535
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.278170789349299e-06,
+      "loss": 1.1616,
+      "step": 4536
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.277410587679219e-06,
+      "loss": 1.1752,
+      "step": 4537
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.276650386009139e-06,
+      "loss": 1.0583,
+      "step": 4538
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.275890184339058e-06,
+      "loss": 1.0836,
+      "step": 4539
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.275129982668978e-06,
+      "loss": 1.1256,
+      "step": 4540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.274369780998898e-06,
+      "loss": 1.1712,
+      "step": 4541
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.273609579328818e-06,
+      "loss": 1.1026,
+      "step": 4542
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.2728493776587374e-06,
+      "loss": 1.2391,
+      "step": 4543
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.272089175988657e-06,
+      "loss": 1.0814,
+      "step": 4544
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.271328974318576e-06,
+      "loss": 1.0048,
+      "step": 4545
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.270568772648496e-06,
+      "loss": 0.9597,
+      "step": 4546
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.269808570978416e-06,
+      "loss": 1.2705,
+      "step": 4547
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.269048369308335e-06,
+      "loss": 1.0636,
+      "step": 4548
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.268288167638255e-06,
+      "loss": 0.8577,
+      "step": 4549
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.267527965968174e-06,
+      "loss": 0.8535,
+      "step": 4550
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.266767764298094e-06,
+      "loss": 0.904,
+      "step": 4551
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.266007562628014e-06,
+      "loss": 1.1359,
+      "step": 4552
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.2652473609579335e-06,
+      "loss": 0.9756,
+      "step": 4553
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.264487159287853e-06,
+      "loss": 0.9463,
+      "step": 4554
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.263726957617772e-06,
+      "loss": 1.2459,
+      "step": 4555
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.262966755947692e-06,
+      "loss": 1.1731,
+      "step": 4556
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.262206554277612e-06,
+      "loss": 0.9149,
+      "step": 4557
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.261446352607532e-06,
+      "loss": 1.3306,
+      "step": 4558
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.260686150937452e-06,
+      "loss": 1.13,
+      "step": 4559
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.259925949267372e-06,
+      "loss": 1.0026,
+      "step": 4560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.259165747597291e-06,
+      "loss": 1.1956,
+      "step": 4561
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.2584055459272105e-06,
+      "loss": 1.0855,
+      "step": 4562
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.25764534425713e-06,
+      "loss": 1.1643,
+      "step": 4563
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.25688514258705e-06,
+      "loss": 1.0261,
+      "step": 4564
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.25612494091697e-06,
+      "loss": 1.0793,
+      "step": 4565
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.255364739246888e-06,
+      "loss": 1.2703,
+      "step": 4566
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.254604537576808e-06,
+      "loss": 0.9928,
+      "step": 4567
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.253844335906728e-06,
+      "loss": 1.1247,
+      "step": 4568
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.253084134236648e-06,
+      "loss": 0.8755,
+      "step": 4569
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.252323932566568e-06,
+      "loss": 1.1498,
+      "step": 4570
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.251563730896487e-06,
+      "loss": 1.1782,
+      "step": 4571
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.2508035292264065e-06,
+      "loss": 0.9769,
+      "step": 4572
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.250043327556326e-06,
+      "loss": 1.0108,
+      "step": 4573
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.249283125886246e-06,
+      "loss": 1.2862,
+      "step": 4574
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.248522924216166e-06,
+      "loss": 1.053,
+      "step": 4575
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.247762722546086e-06,
+      "loss": 1.0169,
+      "step": 4576
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.247002520876005e-06,
+      "loss": 1.0348,
+      "step": 4577
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.246242319205925e-06,
+      "loss": 0.9812,
+      "step": 4578
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.245482117535845e-06,
+      "loss": 0.9458,
+      "step": 4579
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.2447219158657645e-06,
+      "loss": 1.1009,
+      "step": 4580
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.243961714195684e-06,
+      "loss": 1.1204,
+      "step": 4581
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.2432015125256025e-06,
+      "loss": 1.1508,
+      "step": 4582
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.242441310855522e-06,
+      "loss": 1.1026,
+      "step": 4583
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.241681109185442e-06,
+      "loss": 1.079,
+      "step": 4584
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.240920907515362e-06,
+      "loss": 1.14,
+      "step": 4585
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.240160705845282e-06,
+      "loss": 1.0846,
+      "step": 4586
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.239400504175201e-06,
+      "loss": 0.9278,
+      "step": 4587
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.238640302505121e-06,
+      "loss": 1.1842,
+      "step": 4588
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.237880100835041e-06,
+      "loss": 0.8929,
+      "step": 4589
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.2371198991649605e-06,
+      "loss": 1.0515,
+      "step": 4590
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.23635969749488e-06,
+      "loss": 0.8798,
+      "step": 4591
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.2355994958248e-06,
+      "loss": 1.1126,
+      "step": 4592
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.234839294154719e-06,
+      "loss": 1.0114,
+      "step": 4593
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.234079092484639e-06,
+      "loss": 1.0899,
+      "step": 4594
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.233318890814559e-06,
+      "loss": 1.0112,
+      "step": 4595
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.232558689144479e-06,
+      "loss": 1.1437,
+      "step": 4596
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.231798487474399e-06,
+      "loss": 1.1142,
+      "step": 4597
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.231038285804317e-06,
+      "loss": 1.0774,
+      "step": 4598
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.230278084134237e-06,
+      "loss": 1.1555,
+      "step": 4599
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.2295178824641565e-06,
+      "loss": 0.9202,
+      "step": 4600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.228757680794076e-06,
+      "loss": 1.0752,
+      "step": 4601
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.227997479123996e-06,
+      "loss": 0.7781,
+      "step": 4602
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.227237277453915e-06,
+      "loss": 0.9297,
+      "step": 4603
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.226477075783835e-06,
+      "loss": 1.1638,
+      "step": 4604
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.225716874113755e-06,
+      "loss": 1.1856,
+      "step": 4605
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.224956672443675e-06,
+      "loss": 1.2431,
+      "step": 4606
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.224196470773595e-06,
+      "loss": 1.189,
+      "step": 4607
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.2234362691035145e-06,
+      "loss": 0.9194,
+      "step": 4608
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.2226760674334335e-06,
+      "loss": 1.1693,
+      "step": 4609
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.221915865763353e-06,
+      "loss": 0.9434,
+      "step": 4610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.221155664093273e-06,
+      "loss": 1.0356,
+      "step": 4611
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.220395462423193e-06,
+      "loss": 1.0712,
+      "step": 4612
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.219635260753113e-06,
+      "loss": 1.0977,
+      "step": 4613
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.218875059083031e-06,
+      "loss": 1.0361,
+      "step": 4614
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.218114857412951e-06,
+      "loss": 1.119,
+      "step": 4615
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.217354655742871e-06,
+      "loss": 1.2692,
+      "step": 4616
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.216594454072791e-06,
+      "loss": 1.0441,
+      "step": 4617
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.2158342524027106e-06,
+      "loss": 0.9722,
+      "step": 4618
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.2150740507326296e-06,
+      "loss": 0.9589,
+      "step": 4619
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.214313849062549e-06,
+      "loss": 1.0774,
+      "step": 4620
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.213553647392469e-06,
+      "loss": 1.1404,
+      "step": 4621
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.212793445722389e-06,
+      "loss": 1.0388,
+      "step": 4622
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.212033244052309e-06,
+      "loss": 1.0708,
+      "step": 4623
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.211273042382229e-06,
+      "loss": 1.1181,
+      "step": 4624
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.210512840712148e-06,
+      "loss": 1.1062,
+      "step": 4625
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.209752639042068e-06,
+      "loss": 1.0623,
+      "step": 4626
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.2089924373719876e-06,
+      "loss": 1.1893,
+      "step": 4627
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.208232235701907e-06,
+      "loss": 0.9775,
+      "step": 4628
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.207472034031827e-06,
+      "loss": 0.9657,
+      "step": 4629
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.206711832361746e-06,
+      "loss": 1.1371,
+      "step": 4630
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.205951630691666e-06,
+      "loss": 1.0471,
+      "step": 4631
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.205191429021585e-06,
+      "loss": 1.0594,
+      "step": 4632
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.204431227351505e-06,
+      "loss": 1.0649,
+      "step": 4633
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.203671025681425e-06,
+      "loss": 1.1669,
+      "step": 4634
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.202910824011344e-06,
+      "loss": 1.1776,
+      "step": 4635
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.202150622341264e-06,
+      "loss": 1.1124,
+      "step": 4636
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.201390420671184e-06,
+      "loss": 1.0227,
+      "step": 4637
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.2006302190011034e-06,
+      "loss": 1.0678,
+      "step": 4638
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.199870017331023e-06,
+      "loss": 1.2823,
+      "step": 4639
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.199109815660943e-06,
+      "loss": 1.0764,
+      "step": 4640
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.198349613990862e-06,
+      "loss": 1.0443,
+      "step": 4641
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.197589412320782e-06,
+      "loss": 1.2488,
+      "step": 4642
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.196829210650702e-06,
+      "loss": 1.2829,
+      "step": 4643
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.196069008980622e-06,
+      "loss": 1.0121,
+      "step": 4644
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.195308807310542e-06,
+      "loss": 1.1479,
+      "step": 4645
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.194548605640461e-06,
+      "loss": 1.0734,
+      "step": 4646
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.1937884039703804e-06,
+      "loss": 1.2081,
+      "step": 4647
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.1930282023003e-06,
+      "loss": 0.965,
+      "step": 4648
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.19226800063022e-06,
+      "loss": 1.1704,
+      "step": 4649
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.19150779896014e-06,
+      "loss": 1.0012,
+      "step": 4650
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.190747597290058e-06,
+      "loss": 1.0317,
+      "step": 4651
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.189987395619978e-06,
+      "loss": 1.2813,
+      "step": 4652
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.189227193949898e-06,
+      "loss": 1.1631,
+      "step": 4653
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.188466992279818e-06,
+      "loss": 1.164,
+      "step": 4654
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.187706790609738e-06,
+      "loss": 1.0072,
+      "step": 4655
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.1869465889396575e-06,
+      "loss": 0.9815,
+      "step": 4656
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.1861863872695765e-06,
+      "loss": 0.8803,
+      "step": 4657
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.185426185599496e-06,
+      "loss": 1.0062,
+      "step": 4658
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.184665983929416e-06,
+      "loss": 1.0128,
+      "step": 4659
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.183905782259336e-06,
+      "loss": 1.1144,
+      "step": 4660
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.183145580589256e-06,
+      "loss": 1.2203,
+      "step": 4661
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.182385378919175e-06,
+      "loss": 1.033,
+      "step": 4662
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.181625177249095e-06,
+      "loss": 1.1283,
+      "step": 4663
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.180864975579015e-06,
+      "loss": 1.0946,
+      "step": 4664
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.1801047739089345e-06,
+      "loss": 0.9143,
+      "step": 4665
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.179344572238854e-06,
+      "loss": 1.1107,
+      "step": 4666
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.1785843705687725e-06,
+      "loss": 1.1848,
+      "step": 4667
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.177824168898692e-06,
+      "loss": 1.0709,
+      "step": 4668
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.177063967228612e-06,
+      "loss": 1.0302,
+      "step": 4669
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.176303765558532e-06,
+      "loss": 1.1087,
+      "step": 4670
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.175543563888452e-06,
+      "loss": 0.9766,
+      "step": 4671
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.174783362218372e-06,
+      "loss": 1.0409,
+      "step": 4672
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.174023160548291e-06,
+      "loss": 1.023,
+      "step": 4673
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.173262958878211e-06,
+      "loss": 1.103,
+      "step": 4674
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.1725027572081305e-06,
+      "loss": 0.9831,
+      "step": 4675
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.17174255553805e-06,
+      "loss": 1.1206,
+      "step": 4676
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.17098235386797e-06,
+      "loss": 1.1063,
+      "step": 4677
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.170222152197889e-06,
+      "loss": 1.0079,
+      "step": 4678
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.169461950527809e-06,
+      "loss": 1.1535,
+      "step": 4679
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.168701748857729e-06,
+      "loss": 1.0627,
+      "step": 4680
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.167941547187649e-06,
+      "loss": 1.0525,
+      "step": 4681
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.167181345517569e-06,
+      "loss": 1.0715,
+      "step": 4682
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.166421143847487e-06,
+      "loss": 0.8551,
+      "step": 4683
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.165660942177407e-06,
+      "loss": 1.0984,
+      "step": 4684
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.1649007405073265e-06,
+      "loss": 1.0091,
+      "step": 4685
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.164140538837246e-06,
+      "loss": 1.0919,
+      "step": 4686
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.163380337167166e-06,
+      "loss": 0.9919,
+      "step": 4687
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.162620135497086e-06,
+      "loss": 1.1052,
+      "step": 4688
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.161859933827005e-06,
+      "loss": 1.1369,
+      "step": 4689
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.161099732156925e-06,
+      "loss": 1.1796,
+      "step": 4690
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.160339530486845e-06,
+      "loss": 1.2057,
+      "step": 4691
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.159579328816765e-06,
+      "loss": 1.263,
+      "step": 4692
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.1588191271466845e-06,
+      "loss": 1.111,
+      "step": 4693
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.1580589254766035e-06,
+      "loss": 0.8798,
+      "step": 4694
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.157298723806523e-06,
+      "loss": 1.1665,
+      "step": 4695
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.156538522136443e-06,
+      "loss": 1.1122,
+      "step": 4696
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.155778320466363e-06,
+      "loss": 0.9037,
+      "step": 4697
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.155018118796283e-06,
+      "loss": 1.1412,
+      "step": 4698
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.154257917126201e-06,
+      "loss": 1.0569,
+      "step": 4699
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.153497715456121e-06,
+      "loss": 1.0466,
+      "step": 4700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.152737513786041e-06,
+      "loss": 1.0417,
+      "step": 4701
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.151977312115961e-06,
+      "loss": 1.0047,
+      "step": 4702
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.1512171104458805e-06,
+      "loss": 1.0576,
+      "step": 4703
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.1504569087757995e-06,
+      "loss": 0.9459,
+      "step": 4704
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.149696707105719e-06,
+      "loss": 0.9971,
+      "step": 4705
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.148936505435639e-06,
+      "loss": 0.9491,
+      "step": 4706
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.148176303765559e-06,
+      "loss": 1.0664,
+      "step": 4707
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.147416102095479e-06,
+      "loss": 0.8967,
+      "step": 4708
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.146655900425399e-06,
+      "loss": 0.9926,
+      "step": 4709
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.145895698755318e-06,
+      "loss": 1.0039,
+      "step": 4710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.145135497085238e-06,
+      "loss": 1.2579,
+      "step": 4711
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.1443752954151575e-06,
+      "loss": 0.9934,
+      "step": 4712
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.143615093745077e-06,
+      "loss": 1.1007,
+      "step": 4713
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.142854892074997e-06,
+      "loss": 0.8734,
+      "step": 4714
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.142094690404915e-06,
+      "loss": 1.0195,
+      "step": 4715
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.141334488734835e-06,
+      "loss": 0.8597,
+      "step": 4716
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.140574287064755e-06,
+      "loss": 1.0554,
+      "step": 4717
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.139814085394675e-06,
+      "loss": 0.9766,
+      "step": 4718
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.139053883724595e-06,
+      "loss": 1.0224,
+      "step": 4719
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.138293682054514e-06,
+      "loss": 0.8602,
+      "step": 4720
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.137533480384434e-06,
+      "loss": 1.1946,
+      "step": 4721
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.1367732787143536e-06,
+      "loss": 0.9991,
+      "step": 4722
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.136013077044273e-06,
+      "loss": 1.1196,
+      "step": 4723
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.135252875374193e-06,
+      "loss": 1.1109,
+      "step": 4724
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.134492673704113e-06,
+      "loss": 0.8625,
+      "step": 4725
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.133732472034032e-06,
+      "loss": 0.9119,
+      "step": 4726
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.132972270363952e-06,
+      "loss": 1.0753,
+      "step": 4727
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.132212068693872e-06,
+      "loss": 0.8571,
+      "step": 4728
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.131451867023792e-06,
+      "loss": 0.9647,
+      "step": 4729
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.1306916653537116e-06,
+      "loss": 1.0974,
+      "step": 4730
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.1299314636836306e-06,
+      "loss": 1.1607,
+      "step": 4731
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.12917126201355e-06,
+      "loss": 0.9738,
+      "step": 4732
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.12841106034347e-06,
+      "loss": 1.216,
+      "step": 4733
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.12765085867339e-06,
+      "loss": 0.9085,
+      "step": 4734
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.12689065700331e-06,
+      "loss": 0.9348,
+      "step": 4735
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.126130455333228e-06,
+      "loss": 1.0914,
+      "step": 4736
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.125370253663148e-06,
+      "loss": 0.9198,
+      "step": 4737
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.124610051993068e-06,
+      "loss": 0.9591,
+      "step": 4738
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.123849850322988e-06,
+      "loss": 1.0351,
+      "step": 4739
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.1230896486529076e-06,
+      "loss": 1.0666,
+      "step": 4740
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.1223294469828274e-06,
+      "loss": 1.0444,
+      "step": 4741
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.1215692453127464e-06,
+      "loss": 0.9248,
+      "step": 4742
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.120809043642666e-06,
+      "loss": 0.9323,
+      "step": 4743
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.120048841972586e-06,
+      "loss": 0.9066,
+      "step": 4744
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.119288640302506e-06,
+      "loss": 1.1043,
+      "step": 4745
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.118528438632426e-06,
+      "loss": 1.1016,
+      "step": 4746
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.117768236962345e-06,
+      "loss": 1.0515,
+      "step": 4747
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.117008035292265e-06,
+      "loss": 1.1215,
+      "step": 4748
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.116247833622185e-06,
+      "loss": 1.059,
+      "step": 4749
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.1154876319521044e-06,
+      "loss": 0.98,
+      "step": 4750
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.114727430282024e-06,
+      "loss": 0.9882,
+      "step": 4751
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.1139672286119425e-06,
+      "loss": 1.2878,
+      "step": 4752
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.113207026941862e-06,
+      "loss": 1.2855,
+      "step": 4753
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.112446825271782e-06,
+      "loss": 0.8927,
+      "step": 4754
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.111686623601702e-06,
+      "loss": 1.5147,
+      "step": 4755
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.110926421931622e-06,
+      "loss": 1.1965,
+      "step": 4756
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.110166220261542e-06,
+      "loss": 0.9534,
+      "step": 4757
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.109406018591461e-06,
+      "loss": 1.0112,
+      "step": 4758
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.108645816921381e-06,
+      "loss": 1.1187,
+      "step": 4759
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.1078856152513005e-06,
+      "loss": 1.1287,
+      "step": 4760
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.10712541358122e-06,
+      "loss": 0.9983,
+      "step": 4761
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.10636521191114e-06,
+      "loss": 0.6839,
+      "step": 4762
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.105605010241059e-06,
+      "loss": 1.0491,
+      "step": 4763
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.104844808570979e-06,
+      "loss": 0.9929,
+      "step": 4764
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.104084606900899e-06,
+      "loss": 0.9674,
+      "step": 4765
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.103324405230819e-06,
+      "loss": 1.134,
+      "step": 4766
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.102564203560739e-06,
+      "loss": 0.9897,
+      "step": 4767
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.101804001890657e-06,
+      "loss": 1.0407,
+      "step": 4768
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.101043800220577e-06,
+      "loss": 0.9121,
+      "step": 4769
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.1002835985504965e-06,
+      "loss": 0.927,
+      "step": 4770
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.099523396880416e-06,
+      "loss": 1.012,
+      "step": 4771
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.098763195210336e-06,
+      "loss": 1.0095,
+      "step": 4772
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.098002993540256e-06,
+      "loss": 1.0979,
+      "step": 4773
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.097242791870175e-06,
+      "loss": 1.1579,
+      "step": 4774
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.096482590200095e-06,
+      "loss": 0.8953,
+      "step": 4775
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.095722388530015e-06,
+      "loss": 1.0893,
+      "step": 4776
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.094962186859935e-06,
+      "loss": 1.1387,
+      "step": 4777
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.0942019851898545e-06,
+      "loss": 1.0661,
+      "step": 4778
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.0934417835197735e-06,
+      "loss": 1.0452,
+      "step": 4779
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.092681581849693e-06,
+      "loss": 1.0206,
+      "step": 4780
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.091921380179613e-06,
+      "loss": 1.1467,
+      "step": 4781
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.091161178509533e-06,
+      "loss": 1.073,
+      "step": 4782
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.090400976839453e-06,
+      "loss": 0.8911,
+      "step": 4783
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.089640775169371e-06,
+      "loss": 1.1741,
+      "step": 4784
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.088880573499291e-06,
+      "loss": 1.1006,
+      "step": 4785
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.088120371829211e-06,
+      "loss": 1.028,
+      "step": 4786
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.087360170159131e-06,
+      "loss": 0.9973,
+      "step": 4787
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.0865999684890505e-06,
+      "loss": 1.0621,
+      "step": 4788
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.08583976681897e-06,
+      "loss": 0.9314,
+      "step": 4789
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.085079565148889e-06,
+      "loss": 1.0796,
+      "step": 4790
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.084319363478809e-06,
+      "loss": 1.1446,
+      "step": 4791
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.083559161808729e-06,
+      "loss": 1.1554,
+      "step": 4792
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.082798960138649e-06,
+      "loss": 0.9584,
+      "step": 4793
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.082038758468569e-06,
+      "loss": 1.0903,
+      "step": 4794
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.081278556798488e-06,
+      "loss": 0.9574,
+      "step": 4795
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.080518355128408e-06,
+      "loss": 1.002,
+      "step": 4796
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.0797581534583275e-06,
+      "loss": 1.1353,
+      "step": 4797
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.078997951788247e-06,
+      "loss": 1.3354,
+      "step": 4798
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.078237750118167e-06,
+      "loss": 0.8989,
+      "step": 4799
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.077477548448085e-06,
+      "loss": 1.1499,
+      "step": 4800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.076717346778005e-06,
+      "loss": 1.2256,
+      "step": 4801
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.075957145107925e-06,
+      "loss": 1.1862,
+      "step": 4802
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.075196943437845e-06,
+      "loss": 1.0362,
+      "step": 4803
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.074436741767765e-06,
+      "loss": 0.9553,
+      "step": 4804
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.073676540097685e-06,
+      "loss": 1.0608,
+      "step": 4805
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.072916338427604e-06,
+      "loss": 0.9446,
+      "step": 4806
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.0721561367575235e-06,
+      "loss": 1.042,
+      "step": 4807
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.071395935087443e-06,
+      "loss": 1.2331,
+      "step": 4808
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.070635733417363e-06,
+      "loss": 0.9031,
+      "step": 4809
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.069875531747283e-06,
+      "loss": 1.151,
+      "step": 4810
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.069115330077202e-06,
+      "loss": 1.1672,
+      "step": 4811
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.068355128407122e-06,
+      "loss": 1.2529,
+      "step": 4812
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.067594926737042e-06,
+      "loss": 1.1506,
+      "step": 4813
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.066834725066962e-06,
+      "loss": 0.9101,
+      "step": 4814
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.0660745233968815e-06,
+      "loss": 1.0208,
+      "step": 4815
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.0653143217268005e-06,
+      "loss": 0.9102,
+      "step": 4816
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.06455412005672e-06,
+      "loss": 1.1162,
+      "step": 4817
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.06379391838664e-06,
+      "loss": 1.2056,
+      "step": 4818
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.06303371671656e-06,
+      "loss": 1.1056,
+      "step": 4819
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.062273515046479e-06,
+      "loss": 1.2503,
+      "step": 4820
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.061513313376399e-06,
+      "loss": 1.0565,
+      "step": 4821
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.060753111706318e-06,
+      "loss": 1.0153,
+      "step": 4822
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.059992910036238e-06,
+      "loss": 1.0444,
+      "step": 4823
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.059232708366158e-06,
+      "loss": 0.9455,
+      "step": 4824
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.0584725066960775e-06,
+      "loss": 1.0344,
+      "step": 4825
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.057712305025997e-06,
+      "loss": 0.9644,
+      "step": 4826
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.056952103355916e-06,
+      "loss": 1.1756,
+      "step": 4827
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.056191901685836e-06,
+      "loss": 1.1444,
+      "step": 4828
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.055431700015756e-06,
+      "loss": 1.1664,
+      "step": 4829
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.054671498345676e-06,
+      "loss": 1.04,
+      "step": 4830
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.053911296675596e-06,
+      "loss": 1.05,
+      "step": 4831
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.053151095005515e-06,
+      "loss": 1.3545,
+      "step": 4832
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.052390893335435e-06,
+      "loss": 0.9647,
+      "step": 4833
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.0516306916653546e-06,
+      "loss": 1.1466,
+      "step": 4834
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.050870489995274e-06,
+      "loss": 0.7517,
+      "step": 4835
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.050110288325194e-06,
+      "loss": 0.8885,
+      "step": 4836
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.049350086655114e-06,
+      "loss": 1.061,
+      "step": 4837
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.048589884985032e-06,
+      "loss": 1.0637,
+      "step": 4838
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.047829683314952e-06,
+      "loss": 1.0017,
+      "step": 4839
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.047069481644872e-06,
+      "loss": 1.0774,
+      "step": 4840
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.046309279974792e-06,
+      "loss": 1.0365,
+      "step": 4841
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.045549078304712e-06,
+      "loss": 1.0904,
+      "step": 4842
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.044788876634631e-06,
+      "loss": 1.0488,
+      "step": 4843
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.0440286749645506e-06,
+      "loss": 1.2014,
+      "step": 4844
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.0432684732944704e-06,
+      "loss": 1.07,
+      "step": 4845
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.04250827162439e-06,
+      "loss": 1.1249,
+      "step": 4846
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.04174806995431e-06,
+      "loss": 1.2127,
+      "step": 4847
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.040987868284229e-06,
+      "loss": 1.0025,
+      "step": 4848
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.040227666614149e-06,
+      "loss": 1.0709,
+      "step": 4849
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.039467464944069e-06,
+      "loss": 1.2309,
+      "step": 4850
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.038707263273989e-06,
+      "loss": 1.0686,
+      "step": 4851
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.037947061603909e-06,
+      "loss": 0.9185,
+      "step": 4852
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.0371868599338284e-06,
+      "loss": 0.9889,
+      "step": 4853
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.036426658263747e-06,
+      "loss": 1.0811,
+      "step": 4854
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.0356664565936665e-06,
+      "loss": 1.1809,
+      "step": 4855
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.034906254923586e-06,
+      "loss": 1.0227,
+      "step": 4856
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.034146053253506e-06,
+      "loss": 1.0947,
+      "step": 4857
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.033385851583426e-06,
+      "loss": 1.045,
+      "step": 4858
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.032625649913345e-06,
+      "loss": 1.0847,
+      "step": 4859
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.031865448243265e-06,
+      "loss": 1.0637,
+      "step": 4860
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.031105246573185e-06,
+      "loss": 1.0512,
+      "step": 4861
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.030345044903105e-06,
+      "loss": 1.0145,
+      "step": 4862
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.0295848432330245e-06,
+      "loss": 1.1202,
+      "step": 4863
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.0288246415629435e-06,
+      "loss": 1.2208,
+      "step": 4864
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.028064439892863e-06,
+      "loss": 1.0588,
+      "step": 4865
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.027304238222783e-06,
+      "loss": 0.9808,
+      "step": 4866
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.026544036552703e-06,
+      "loss": 1.1269,
+      "step": 4867
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.025783834882623e-06,
+      "loss": 0.9467,
+      "step": 4868
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.025023633212543e-06,
+      "loss": 1.0667,
+      "step": 4869
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.024263431542461e-06,
+      "loss": 1.0002,
+      "step": 4870
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.023503229872381e-06,
+      "loss": 1.0655,
+      "step": 4871
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.022743028202301e-06,
+      "loss": 1.1725,
+      "step": 4872
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.0219828265322205e-06,
+      "loss": 1.0408,
+      "step": 4873
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.02122262486214e-06,
+      "loss": 0.9959,
+      "step": 4874
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.020462423192059e-06,
+      "loss": 1.1453,
+      "step": 4875
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.019702221521979e-06,
+      "loss": 1.2242,
+      "step": 4876
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.018942019851899e-06,
+      "loss": 1.141,
+      "step": 4877
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.018181818181819e-06,
+      "loss": 0.9163,
+      "step": 4878
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.017421616511739e-06,
+      "loss": 0.9621,
+      "step": 4879
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.016661414841658e-06,
+      "loss": 1.0502,
+      "step": 4880
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.015901213171578e-06,
+      "loss": 1.0725,
+      "step": 4881
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.0151410115014975e-06,
+      "loss": 1.0583,
+      "step": 4882
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.014380809831417e-06,
+      "loss": 1.145,
+      "step": 4883
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.013620608161337e-06,
+      "loss": 0.9453,
+      "step": 4884
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.012860406491257e-06,
+      "loss": 1.0216,
+      "step": 4885
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.012100204821175e-06,
+      "loss": 1.0166,
+      "step": 4886
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.011340003151095e-06,
+      "loss": 1.3155,
+      "step": 4887
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.010579801481015e-06,
+      "loss": 1.1327,
+      "step": 4888
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.009819599810935e-06,
+      "loss": 1.276,
+      "step": 4889
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.009059398140855e-06,
+      "loss": 1.0088,
+      "step": 4890
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.008299196470774e-06,
+      "loss": 1.0753,
+      "step": 4891
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.0075389948006935e-06,
+      "loss": 1.0732,
+      "step": 4892
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.006778793130613e-06,
+      "loss": 1.0211,
+      "step": 4893
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.006018591460533e-06,
+      "loss": 1.0549,
+      "step": 4894
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.005258389790453e-06,
+      "loss": 1.2533,
+      "step": 4895
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.004498188120372e-06,
+      "loss": 1.158,
+      "step": 4896
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.003737986450292e-06,
+      "loss": 1.0646,
+      "step": 4897
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.002977784780212e-06,
+      "loss": 1.1101,
+      "step": 4898
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.002217583110132e-06,
+      "loss": 1.1399,
+      "step": 4899
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.0014573814400515e-06,
+      "loss": 1.1292,
+      "step": 4900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.0006971797699705e-06,
+      "loss": 1.1511,
+      "step": 4901
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.99993697809989e-06,
+      "loss": 1.0812,
+      "step": 4902
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.999176776429809e-06,
+      "loss": 1.1422,
+      "step": 4903
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.998416574759729e-06,
+      "loss": 0.9759,
+      "step": 4904
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.997656373089649e-06,
+      "loss": 1.09,
+      "step": 4905
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.996896171419569e-06,
+      "loss": 1.0833,
+      "step": 4906
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.996135969749488e-06,
+      "loss": 1.1204,
+      "step": 4907
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.995375768079408e-06,
+      "loss": 1.2325,
+      "step": 4908
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.994615566409328e-06,
+      "loss": 1.0603,
+      "step": 4909
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.9938553647392475e-06,
+      "loss": 0.9586,
+      "step": 4910
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.993095163069167e-06,
+      "loss": 1.1319,
+      "step": 4911
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.992334961399086e-06,
+      "loss": 1.0328,
+      "step": 4912
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.991574759729006e-06,
+      "loss": 1.0108,
+      "step": 4913
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.990814558058926e-06,
+      "loss": 1.2315,
+      "step": 4914
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.990054356388846e-06,
+      "loss": 1.1342,
+      "step": 4915
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.989294154718766e-06,
+      "loss": 0.9943,
+      "step": 4916
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.988533953048685e-06,
+      "loss": 1.2247,
+      "step": 4917
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.987773751378605e-06,
+      "loss": 0.9552,
+      "step": 4918
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.9870135497085245e-06,
+      "loss": 1.1944,
+      "step": 4919
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.986253348038444e-06,
+      "loss": 1.2082,
+      "step": 4920
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.985493146368364e-06,
+      "loss": 0.9713,
+      "step": 4921
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.984732944698284e-06,
+      "loss": 1.0906,
+      "step": 4922
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.983972743028202e-06,
+      "loss": 1.2014,
+      "step": 4923
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.983212541358122e-06,
+      "loss": 1.2028,
+      "step": 4924
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.982452339688042e-06,
+      "loss": 1.014,
+      "step": 4925
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.981692138017962e-06,
+      "loss": 1.1026,
+      "step": 4926
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.980931936347882e-06,
+      "loss": 1.0069,
+      "step": 4927
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.980171734677801e-06,
+      "loss": 0.9114,
+      "step": 4928
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.9794115330077205e-06,
+      "loss": 0.9528,
+      "step": 4929
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.97865133133764e-06,
+      "loss": 1.0596,
+      "step": 4930
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.97789112966756e-06,
+      "loss": 1.0778,
+      "step": 4931
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.97713092799748e-06,
+      "loss": 1.0702,
+      "step": 4932
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.976370726327399e-06,
+      "loss": 1.0401,
+      "step": 4933
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.975610524657319e-06,
+      "loss": 1.0704,
+      "step": 4934
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.974850322987239e-06,
+      "loss": 1.0753,
+      "step": 4935
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.974090121317159e-06,
+      "loss": 1.0529,
+      "step": 4936
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.9733299196470785e-06,
+      "loss": 1.0585,
+      "step": 4937
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.972569717976998e-06,
+      "loss": 0.8276,
+      "step": 4938
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.9718095163069166e-06,
+      "loss": 1.1574,
+      "step": 4939
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.9710493146368364e-06,
+      "loss": 1.0263,
+      "step": 4940
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.970289112966756e-06,
+      "loss": 1.2029,
+      "step": 4941
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.969528911296676e-06,
+      "loss": 1.1925,
+      "step": 4942
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.968768709626596e-06,
+      "loss": 0.8737,
+      "step": 4943
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.968008507956515e-06,
+      "loss": 1.0446,
+      "step": 4944
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.967248306286435e-06,
+      "loss": 0.9388,
+      "step": 4945
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.966488104616355e-06,
+      "loss": 0.8925,
+      "step": 4946
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.9657279029462746e-06,
+      "loss": 0.9024,
+      "step": 4947
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.9649677012761944e-06,
+      "loss": 0.9568,
+      "step": 4948
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.9642074996061134e-06,
+      "loss": 1.0863,
+      "step": 4949
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.963447297936033e-06,
+      "loss": 1.1289,
+      "step": 4950
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.962687096265953e-06,
+      "loss": 1.1598,
+      "step": 4951
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.961926894595873e-06,
+      "loss": 1.0392,
+      "step": 4952
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.961166692925793e-06,
+      "loss": 1.1313,
+      "step": 4953
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.960406491255713e-06,
+      "loss": 1.1063,
+      "step": 4954
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.959646289585631e-06,
+      "loss": 1.404,
+      "step": 4955
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.958886087915551e-06,
+      "loss": 1.0234,
+      "step": 4956
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.958125886245471e-06,
+      "loss": 1.0167,
+      "step": 4957
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.9573656845753904e-06,
+      "loss": 1.126,
+      "step": 4958
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.95660548290531e-06,
+      "loss": 1.064,
+      "step": 4959
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.955845281235229e-06,
+      "loss": 0.8451,
+      "step": 4960
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.955085079565149e-06,
+      "loss": 1.0354,
+      "step": 4961
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.954324877895069e-06,
+      "loss": 0.9005,
+      "step": 4962
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.953564676224989e-06,
+      "loss": 1.0454,
+      "step": 4963
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.952804474554909e-06,
+      "loss": 0.9579,
+      "step": 4964
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.952044272884828e-06,
+      "loss": 1.1004,
+      "step": 4965
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.951284071214748e-06,
+      "loss": 0.8492,
+      "step": 4966
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.9505238695446675e-06,
+      "loss": 0.8696,
+      "step": 4967
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.949763667874587e-06,
+      "loss": 1.1871,
+      "step": 4968
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.949003466204507e-06,
+      "loss": 1.0503,
+      "step": 4969
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.948243264534427e-06,
+      "loss": 1.0722,
+      "step": 4970
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.947483062864345e-06,
+      "loss": 1.104,
+      "step": 4971
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.946722861194265e-06,
+      "loss": 1.1037,
+      "step": 4972
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.945962659524185e-06,
+      "loss": 0.9905,
+      "step": 4973
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.945202457854105e-06,
+      "loss": 1.0945,
+      "step": 4974
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.944442256184025e-06,
+      "loss": 1.1712,
+      "step": 4975
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.943682054513944e-06,
+      "loss": 1.3744,
+      "step": 4976
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.9429218528438635e-06,
+      "loss": 1.0031,
+      "step": 4977
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.942161651173783e-06,
+      "loss": 0.9349,
+      "step": 4978
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.941401449503703e-06,
+      "loss": 1.2543,
+      "step": 4979
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.940641247833623e-06,
+      "loss": 1.1694,
+      "step": 4980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.939881046163542e-06,
+      "loss": 1.0127,
+      "step": 4981
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.939120844493462e-06,
+      "loss": 1.1694,
+      "step": 4982
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.938360642823382e-06,
+      "loss": 1.1336,
+      "step": 4983
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.937600441153302e-06,
+      "loss": 1.0987,
+      "step": 4984
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.9368402394832215e-06,
+      "loss": 1.032,
+      "step": 4985
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.936080037813141e-06,
+      "loss": 1.1008,
+      "step": 4986
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.9353198361430595e-06,
+      "loss": 1.1309,
+      "step": 4987
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.934559634472979e-06,
+      "loss": 1.1224,
+      "step": 4988
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.933799432802899e-06,
+      "loss": 1.0313,
+      "step": 4989
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.933039231132819e-06,
+      "loss": 0.9768,
+      "step": 4990
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.932279029462739e-06,
+      "loss": 1.0614,
+      "step": 4991
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.931518827792658e-06,
+      "loss": 1.0329,
+      "step": 4992
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.930758626122578e-06,
+      "loss": 1.1508,
+      "step": 4993
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.929998424452498e-06,
+      "loss": 1.1224,
+      "step": 4994
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.9292382227824175e-06,
+      "loss": 1.0811,
+      "step": 4995
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.928478021112337e-06,
+      "loss": 1.0015,
+      "step": 4996
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.927717819442256e-06,
+      "loss": 1.0513,
+      "step": 4997
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.926957617772176e-06,
+      "loss": 1.0384,
+      "step": 4998
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.926197416102096e-06,
+      "loss": 1.0157,
+      "step": 4999
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.925437214432016e-06,
+      "loss": 1.0362,
+      "step": 5000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.924677012761936e-06,
+      "loss": 0.9679,
+      "step": 5001
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.923916811091856e-06,
+      "loss": 0.8972,
+      "step": 5002
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.923156609421775e-06,
+      "loss": 1.0031,
+      "step": 5003
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.9223964077516945e-06,
+      "loss": 1.0339,
+      "step": 5004
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.921636206081614e-06,
+      "loss": 1.2563,
+      "step": 5005
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.920876004411534e-06,
+      "loss": 0.9345,
+      "step": 5006
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.920115802741454e-06,
+      "loss": 1.0168,
+      "step": 5007
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.919355601071372e-06,
+      "loss": 1.2011,
+      "step": 5008
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.918595399401292e-06,
+      "loss": 1.1222,
+      "step": 5009
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.917835197731212e-06,
+      "loss": 1.1112,
+      "step": 5010
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.917074996061132e-06,
+      "loss": 0.9802,
+      "step": 5011
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.916314794391052e-06,
+      "loss": 1.1528,
+      "step": 5012
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.915554592720971e-06,
+      "loss": 0.8541,
+      "step": 5013
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.9147943910508905e-06,
+      "loss": 1.0705,
+      "step": 5014
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.91403418938081e-06,
+      "loss": 1.1321,
+      "step": 5015
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.91327398771073e-06,
+      "loss": 1.0651,
+      "step": 5016
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.91251378604065e-06,
+      "loss": 0.898,
+      "step": 5017
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.91175358437057e-06,
+      "loss": 1.1021,
+      "step": 5018
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.910993382700489e-06,
+      "loss": 1.0682,
+      "step": 5019
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.910233181030409e-06,
+      "loss": 1.1697,
+      "step": 5020
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.909472979360329e-06,
+      "loss": 1.0261,
+      "step": 5021
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.9087127776902485e-06,
+      "loss": 0.9564,
+      "step": 5022
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.907952576020168e-06,
+      "loss": 1.15,
+      "step": 5023
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.9071923743500865e-06,
+      "loss": 1.0453,
+      "step": 5024
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.906432172680006e-06,
+      "loss": 1.027,
+      "step": 5025
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.905671971009926e-06,
+      "loss": 0.9952,
+      "step": 5026
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.904911769339846e-06,
+      "loss": 0.9675,
+      "step": 5027
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.904151567669766e-06,
+      "loss": 1.0299,
+      "step": 5028
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.903391365999685e-06,
+      "loss": 1.2499,
+      "step": 5029
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.902631164329605e-06,
+      "loss": 0.9321,
+      "step": 5030
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.901870962659525e-06,
+      "loss": 0.8779,
+      "step": 5031
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.9011107609894445e-06,
+      "loss": 1.0097,
+      "step": 5032
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.900350559319364e-06,
+      "loss": 1.0979,
+      "step": 5033
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.899590357649284e-06,
+      "loss": 1.1796,
+      "step": 5034
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.898830155979203e-06,
+      "loss": 1.0982,
+      "step": 5035
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.898069954309123e-06,
+      "loss": 1.0861,
+      "step": 5036
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.897309752639043e-06,
+      "loss": 1.0605,
+      "step": 5037
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.896549550968963e-06,
+      "loss": 0.9878,
+      "step": 5038
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.895789349298883e-06,
+      "loss": 1.07,
+      "step": 5039
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.895029147628801e-06,
+      "loss": 1.0594,
+      "step": 5040
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.894268945958721e-06,
+      "loss": 1.1592,
+      "step": 5041
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.8935087442886406e-06,
+      "loss": 0.9903,
+      "step": 5042
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.89274854261856e-06,
+      "loss": 1.003,
+      "step": 5043
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.89198834094848e-06,
+      "loss": 1.0733,
+      "step": 5044
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.891228139278399e-06,
+      "loss": 0.9755,
+      "step": 5045
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.890467937608319e-06,
+      "loss": 1.1304,
+      "step": 5046
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.889707735938239e-06,
+      "loss": 1.2324,
+      "step": 5047
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.888947534268159e-06,
+      "loss": 0.9452,
+      "step": 5048
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.888187332598079e-06,
+      "loss": 0.8276,
+      "step": 5049
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.8874271309279986e-06,
+      "loss": 1.061,
+      "step": 5050
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.8866669292579176e-06,
+      "loss": 1.1451,
+      "step": 5051
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.8859067275878374e-06,
+      "loss": 1.0593,
+      "step": 5052
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.885146525917757e-06,
+      "loss": 1.0152,
+      "step": 5053
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.884386324247677e-06,
+      "loss": 0.9662,
+      "step": 5054
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.883626122577597e-06,
+      "loss": 1.1209,
+      "step": 5055
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.882865920907515e-06,
+      "loss": 0.9666,
+      "step": 5056
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.882105719237435e-06,
+      "loss": 1.1699,
+      "step": 5057
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.881345517567355e-06,
+      "loss": 0.9539,
+      "step": 5058
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.880585315897275e-06,
+      "loss": 1.1043,
+      "step": 5059
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.879825114227195e-06,
+      "loss": 1.1205,
+      "step": 5060
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.879064912557114e-06,
+      "loss": 1.2138,
+      "step": 5061
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.8783047108870334e-06,
+      "loss": 1.0097,
+      "step": 5062
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.877544509216953e-06,
+      "loss": 1.1077,
+      "step": 5063
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.876784307546873e-06,
+      "loss": 1.0364,
+      "step": 5064
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.876024105876793e-06,
+      "loss": 1.047,
+      "step": 5065
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.875263904206713e-06,
+      "loss": 1.2497,
+      "step": 5066
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.874503702536632e-06,
+      "loss": 1.0518,
+      "step": 5067
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.873743500866552e-06,
+      "loss": 1.0684,
+      "step": 5068
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.872983299196472e-06,
+      "loss": 1.1877,
+      "step": 5069
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.8722230975263914e-06,
+      "loss": 1.2479,
+      "step": 5070
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.871462895856311e-06,
+      "loss": 1.1499,
+      "step": 5071
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.8707026941862295e-06,
+      "loss": 1.0898,
+      "step": 5072
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.869942492516149e-06,
+      "loss": 1.036,
+      "step": 5073
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.869182290846069e-06,
+      "loss": 1.146,
+      "step": 5074
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.868422089175989e-06,
+      "loss": 1.1579,
+      "step": 5075
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.867661887505909e-06,
+      "loss": 1.0255,
+      "step": 5076
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.866901685835828e-06,
+      "loss": 0.9489,
+      "step": 5077
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.866141484165748e-06,
+      "loss": 1.0429,
+      "step": 5078
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.865381282495668e-06,
+      "loss": 1.049,
+      "step": 5079
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.8646210808255875e-06,
+      "loss": 1.1064,
+      "step": 5080
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.863860879155507e-06,
+      "loss": 1.0346,
+      "step": 5081
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.863100677485427e-06,
+      "loss": 1.106,
+      "step": 5082
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.862340475815346e-06,
+      "loss": 0.9609,
+      "step": 5083
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.861580274145266e-06,
+      "loss": 0.8835,
+      "step": 5084
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.860820072475186e-06,
+      "loss": 1.0117,
+      "step": 5085
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.860059870805106e-06,
+      "loss": 1.1046,
+      "step": 5086
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.859299669135026e-06,
+      "loss": 0.9232,
+      "step": 5087
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.858539467464945e-06,
+      "loss": 1.114,
+      "step": 5088
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.8577792657948645e-06,
+      "loss": 0.955,
+      "step": 5089
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.857019064124784e-06,
+      "loss": 1.1371,
+      "step": 5090
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.856258862454703e-06,
+      "loss": 1.0199,
+      "step": 5091
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.855498660784623e-06,
+      "loss": 1.0931,
+      "step": 5092
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.854738459114542e-06,
+      "loss": 0.9665,
+      "step": 5093
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.853978257444462e-06,
+      "loss": 0.9325,
+      "step": 5094
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.853218055774382e-06,
+      "loss": 1.0629,
+      "step": 5095
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.852457854104302e-06,
+      "loss": 1.1011,
+      "step": 5096
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.851697652434222e-06,
+      "loss": 1.0749,
+      "step": 5097
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.8509374507641415e-06,
+      "loss": 0.7526,
+      "step": 5098
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.8501772490940605e-06,
+      "loss": 1.4224,
+      "step": 5099
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.84941704742398e-06,
+      "loss": 1.0104,
+      "step": 5100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.8486568457539e-06,
+      "loss": 1.0551,
+      "step": 5101
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.84789664408382e-06,
+      "loss": 1.1131,
+      "step": 5102
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.84713644241374e-06,
+      "loss": 1.1864,
+      "step": 5103
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.846376240743659e-06,
+      "loss": 0.9252,
+      "step": 5104
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.845616039073579e-06,
+      "loss": 1.0144,
+      "step": 5105
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.844855837403499e-06,
+      "loss": 1.1893,
+      "step": 5106
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.8440956357334185e-06,
+      "loss": 1.2163,
+      "step": 5107
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.843335434063338e-06,
+      "loss": 1.1684,
+      "step": 5108
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.8425752323932565e-06,
+      "loss": 0.9917,
+      "step": 5109
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.841815030723176e-06,
+      "loss": 1.1779,
+      "step": 5110
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.841054829053096e-06,
+      "loss": 1.1318,
+      "step": 5111
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.840294627383016e-06,
+      "loss": 1.1789,
+      "step": 5112
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.839534425712936e-06,
+      "loss": 0.9643,
+      "step": 5113
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.838774224042855e-06,
+      "loss": 1.1437,
+      "step": 5114
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.838014022372775e-06,
+      "loss": 1.0117,
+      "step": 5115
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.837253820702695e-06,
+      "loss": 1.1367,
+      "step": 5116
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.8364936190326145e-06,
+      "loss": 1.0281,
+      "step": 5117
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.835733417362534e-06,
+      "loss": 1.0129,
+      "step": 5118
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.834973215692454e-06,
+      "loss": 1.022,
+      "step": 5119
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.834213014022373e-06,
+      "loss": 1.033,
+      "step": 5120
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.833452812352293e-06,
+      "loss": 1.0991,
+      "step": 5121
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.832692610682213e-06,
+      "loss": 0.7539,
+      "step": 5122
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.831932409012133e-06,
+      "loss": 1.1503,
+      "step": 5123
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.831172207342053e-06,
+      "loss": 1.0197,
+      "step": 5124
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.830412005671971e-06,
+      "loss": 1.1014,
+      "step": 5125
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.829651804001891e-06,
+      "loss": 0.9239,
+      "step": 5126
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.8288916023318105e-06,
+      "loss": 1.0456,
+      "step": 5127
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.82813140066173e-06,
+      "loss": 1.1439,
+      "step": 5128
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.82737119899165e-06,
+      "loss": 1.0367,
+      "step": 5129
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.826610997321569e-06,
+      "loss": 1.1394,
+      "step": 5130
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.825850795651489e-06,
+      "loss": 1.1912,
+      "step": 5131
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.825090593981409e-06,
+      "loss": 1.0485,
+      "step": 5132
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.824330392311329e-06,
+      "loss": 1.0797,
+      "step": 5133
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.823570190641249e-06,
+      "loss": 1.2266,
+      "step": 5134
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.8228099889711685e-06,
+      "loss": 1.1224,
+      "step": 5135
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.8220497873010875e-06,
+      "loss": 1.2679,
+      "step": 5136
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.821289585631007e-06,
+      "loss": 1.1754,
+      "step": 5137
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.820529383960927e-06,
+      "loss": 1.0917,
+      "step": 5138
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.819769182290847e-06,
+      "loss": 0.987,
+      "step": 5139
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.819008980620767e-06,
+      "loss": 1.108,
+      "step": 5140
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.818248778950685e-06,
+      "loss": 1.2361,
+      "step": 5141
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.817488577280605e-06,
+      "loss": 1.0396,
+      "step": 5142
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.816728375610525e-06,
+      "loss": 1.0268,
+      "step": 5143
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.815968173940445e-06,
+      "loss": 1.1521,
+      "step": 5144
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.8152079722703646e-06,
+      "loss": 1.144,
+      "step": 5145
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.8144477706002836e-06,
+      "loss": 1.0466,
+      "step": 5146
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.813687568930203e-06,
+      "loss": 1.19,
+      "step": 5147
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.812927367260123e-06,
+      "loss": 0.9371,
+      "step": 5148
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.812167165590043e-06,
+      "loss": 0.9509,
+      "step": 5149
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.811406963919963e-06,
+      "loss": 1.0633,
+      "step": 5150
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.810646762249883e-06,
+      "loss": 1.0416,
+      "step": 5151
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.809886560579802e-06,
+      "loss": 1.0819,
+      "step": 5152
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.809126358909722e-06,
+      "loss": 0.9672,
+      "step": 5153
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.8083661572396416e-06,
+      "loss": 1.079,
+      "step": 5154
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.807605955569561e-06,
+      "loss": 1.2516,
+      "step": 5155
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.806845753899481e-06,
+      "loss": 1.0604,
+      "step": 5156
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.8060855522293994e-06,
+      "loss": 1.1804,
+      "step": 5157
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.805325350559319e-06,
+      "loss": 1.2413,
+      "step": 5158
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.804565148889239e-06,
+      "loss": 1.0642,
+      "step": 5159
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.803804947219159e-06,
+      "loss": 0.9997,
+      "step": 5160
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.803044745549079e-06,
+      "loss": 1.018,
+      "step": 5161
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.802284543878998e-06,
+      "loss": 0.9087,
+      "step": 5162
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.801524342208918e-06,
+      "loss": 1.0183,
+      "step": 5163
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.800764140538838e-06,
+      "loss": 1.3397,
+      "step": 5164
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.8000039388687574e-06,
+      "loss": 0.9179,
+      "step": 5165
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.799243737198677e-06,
+      "loss": 1.093,
+      "step": 5166
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.798483535528597e-06,
+      "loss": 1.0391,
+      "step": 5167
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.797723333858516e-06,
+      "loss": 0.9541,
+      "step": 5168
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.796963132188436e-06,
+      "loss": 0.9398,
+      "step": 5169
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.796202930518356e-06,
+      "loss": 1.0842,
+      "step": 5170
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.795442728848276e-06,
+      "loss": 1.0104,
+      "step": 5171
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.794682527178196e-06,
+      "loss": 0.9457,
+      "step": 5172
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.793922325508115e-06,
+      "loss": 1.0382,
+      "step": 5173
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.793162123838034e-06,
+      "loss": 1.0947,
+      "step": 5174
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.7924019221679535e-06,
+      "loss": 1.0622,
+      "step": 5175
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.791641720497873e-06,
+      "loss": 1.3067,
+      "step": 5176
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.790881518827793e-06,
+      "loss": 1.2487,
+      "step": 5177
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.790121317157712e-06,
+      "loss": 1.254,
+      "step": 5178
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.789361115487632e-06,
+      "loss": 1.1324,
+      "step": 5179
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.788600913817552e-06,
+      "loss": 1.1187,
+      "step": 5180
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.787840712147472e-06,
+      "loss": 1.1974,
+      "step": 5181
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.787080510477392e-06,
+      "loss": 0.8883,
+      "step": 5182
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.7863203088073115e-06,
+      "loss": 1.0954,
+      "step": 5183
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.7855601071372305e-06,
+      "loss": 1.1039,
+      "step": 5184
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.78479990546715e-06,
+      "loss": 0.8514,
+      "step": 5185
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.78403970379707e-06,
+      "loss": 1.1909,
+      "step": 5186
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.78327950212699e-06,
+      "loss": 1.1615,
+      "step": 5187
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.78251930045691e-06,
+      "loss": 0.9814,
+      "step": 5188
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.781759098786829e-06,
+      "loss": 1.0987,
+      "step": 5189
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.780998897116749e-06,
+      "loss": 0.9672,
+      "step": 5190
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.780238695446669e-06,
+      "loss": 1.0473,
+      "step": 5191
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.7794784937765885e-06,
+      "loss": 1.1204,
+      "step": 5192
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.778718292106508e-06,
+      "loss": 0.9521,
+      "step": 5193
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.7779580904364265e-06,
+      "loss": 0.9608,
+      "step": 5194
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.777197888766346e-06,
+      "loss": 0.9999,
+      "step": 5195
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.776437687096266e-06,
+      "loss": 1.0303,
+      "step": 5196
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.775677485426186e-06,
+      "loss": 1.2821,
+      "step": 5197
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.774917283756106e-06,
+      "loss": 0.9887,
+      "step": 5198
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.774157082086026e-06,
+      "loss": 0.989,
+      "step": 5199
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.773396880415945e-06,
+      "loss": 1.104,
+      "step": 5200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.772636678745865e-06,
+      "loss": 0.9607,
+      "step": 5201
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.7718764770757845e-06,
+      "loss": 1.0852,
+      "step": 5202
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.771116275405704e-06,
+      "loss": 1.0183,
+      "step": 5203
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.770356073735624e-06,
+      "loss": 1.0631,
+      "step": 5204
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.769595872065543e-06,
+      "loss": 1.1573,
+      "step": 5205
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.768835670395463e-06,
+      "loss": 0.7345,
+      "step": 5206
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.768075468725383e-06,
+      "loss": 1.2785,
+      "step": 5207
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.767315267055303e-06,
+      "loss": 1.3193,
+      "step": 5208
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.766555065385223e-06,
+      "loss": 1.1738,
+      "step": 5209
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.765794863715141e-06,
+      "loss": 0.9073,
+      "step": 5210
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.765034662045061e-06,
+      "loss": 0.9314,
+      "step": 5211
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7642744603749805e-06,
+      "loss": 1.1081,
+      "step": 5212
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7635142587049e-06,
+      "loss": 0.9556,
+      "step": 5213
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.76275405703482e-06,
+      "loss": 0.9544,
+      "step": 5214
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.76199385536474e-06,
+      "loss": 1.136,
+      "step": 5215
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.761233653694659e-06,
+      "loss": 1.2267,
+      "step": 5216
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.760473452024579e-06,
+      "loss": 1.004,
+      "step": 5217
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.759713250354499e-06,
+      "loss": 0.9776,
+      "step": 5218
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.758953048684419e-06,
+      "loss": 1.1028,
+      "step": 5219
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7581928470143385e-06,
+      "loss": 1.0356,
+      "step": 5220
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7574326453442575e-06,
+      "loss": 1.1661,
+      "step": 5221
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.756672443674177e-06,
+      "loss": 1.0679,
+      "step": 5222
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.755912242004097e-06,
+      "loss": 0.9316,
+      "step": 5223
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.755152040334017e-06,
+      "loss": 1.0081,
+      "step": 5224
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.754391838663937e-06,
+      "loss": 0.9587,
+      "step": 5225
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.753631636993855e-06,
+      "loss": 0.9901,
+      "step": 5226
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.752871435323775e-06,
+      "loss": 1.0539,
+      "step": 5227
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.752111233653695e-06,
+      "loss": 1.0976,
+      "step": 5228
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.751351031983615e-06,
+      "loss": 0.9916,
+      "step": 5229
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7505908303135345e-06,
+      "loss": 0.9971,
+      "step": 5230
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.749830628643454e-06,
+      "loss": 0.8459,
+      "step": 5231
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.749070426973373e-06,
+      "loss": 0.902,
+      "step": 5232
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.748310225303293e-06,
+      "loss": 1.0837,
+      "step": 5233
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.747550023633213e-06,
+      "loss": 0.9837,
+      "step": 5234
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.746789821963133e-06,
+      "loss": 1.0842,
+      "step": 5235
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.746029620293053e-06,
+      "loss": 0.9322,
+      "step": 5236
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.745269418622972e-06,
+      "loss": 0.9026,
+      "step": 5237
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.744509216952892e-06,
+      "loss": 0.8389,
+      "step": 5238
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7437490152828115e-06,
+      "loss": 0.9874,
+      "step": 5239
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.742988813612731e-06,
+      "loss": 1.1778,
+      "step": 5240
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.742228611942651e-06,
+      "loss": 1.0994,
+      "step": 5241
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.741468410272569e-06,
+      "loss": 0.9443,
+      "step": 5242
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.740708208602489e-06,
+      "loss": 1.0307,
+      "step": 5243
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.739948006932409e-06,
+      "loss": 1.3263,
+      "step": 5244
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.739187805262329e-06,
+      "loss": 0.9612,
+      "step": 5245
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.738427603592249e-06,
+      "loss": 0.9278,
+      "step": 5246
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.737667401922169e-06,
+      "loss": 0.9867,
+      "step": 5247
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.736907200252088e-06,
+      "loss": 0.9415,
+      "step": 5248
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7361469985820076e-06,
+      "loss": 1.0724,
+      "step": 5249
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.735386796911927e-06,
+      "loss": 0.945,
+      "step": 5250
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.734626595241847e-06,
+      "loss": 0.9112,
+      "step": 5251
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.733866393571767e-06,
+      "loss": 1.1743,
+      "step": 5252
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.733106191901686e-06,
+      "loss": 1.0967,
+      "step": 5253
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.732345990231606e-06,
+      "loss": 1.0182,
+      "step": 5254
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.731585788561526e-06,
+      "loss": 1.0572,
+      "step": 5255
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.730825586891446e-06,
+      "loss": 1.1202,
+      "step": 5256
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7300653852213656e-06,
+      "loss": 0.9019,
+      "step": 5257
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.729305183551284e-06,
+      "loss": 1.0978,
+      "step": 5258
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7285449818812036e-06,
+      "loss": 1.0472,
+      "step": 5259
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7277847802111234e-06,
+      "loss": 1.2425,
+      "step": 5260
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.727024578541043e-06,
+      "loss": 1.0846,
+      "step": 5261
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.726264376870963e-06,
+      "loss": 1.1436,
+      "step": 5262
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.725504175200883e-06,
+      "loss": 1.0741,
+      "step": 5263
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.724743973530802e-06,
+      "loss": 1.2562,
+      "step": 5264
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.723983771860722e-06,
+      "loss": 1.2783,
+      "step": 5265
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.723223570190642e-06,
+      "loss": 1.1218,
+      "step": 5266
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.7224633685205616e-06,
+      "loss": 1.1312,
+      "step": 5267
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.7217031668504814e-06,
+      "loss": 0.9116,
+      "step": 5268
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.7209429651804004e-06,
+      "loss": 1.0596,
+      "step": 5269
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.72018276351032e-06,
+      "loss": 1.0855,
+      "step": 5270
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.71942256184024e-06,
+      "loss": 1.1078,
+      "step": 5271
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.71866236017016e-06,
+      "loss": 1.2771,
+      "step": 5272
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.71790215850008e-06,
+      "loss": 1.1455,
+      "step": 5273
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.717141956829999e-06,
+      "loss": 1.0304,
+      "step": 5274
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.716381755159919e-06,
+      "loss": 1.0521,
+      "step": 5275
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.715621553489839e-06,
+      "loss": 1.0307,
+      "step": 5276
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.7148613518197584e-06,
+      "loss": 1.0486,
+      "step": 5277
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.714101150149678e-06,
+      "loss": 1.1699,
+      "step": 5278
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.713340948479598e-06,
+      "loss": 0.9484,
+      "step": 5279
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.712580746809516e-06,
+      "loss": 1.4416,
+      "step": 5280
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.711820545139436e-06,
+      "loss": 0.9737,
+      "step": 5281
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.711060343469356e-06,
+      "loss": 0.919,
+      "step": 5282
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.710300141799276e-06,
+      "loss": 1.0668,
+      "step": 5283
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.709539940129196e-06,
+      "loss": 1.1081,
+      "step": 5284
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.708779738459115e-06,
+      "loss": 1.088,
+      "step": 5285
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.708019536789035e-06,
+      "loss": 0.7921,
+      "step": 5286
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.7072593351189545e-06,
+      "loss": 0.9822,
+      "step": 5287
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.706499133448874e-06,
+      "loss": 0.99,
+      "step": 5288
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.705738931778794e-06,
+      "loss": 1.0253,
+      "step": 5289
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.704978730108713e-06,
+      "loss": 1.0792,
+      "step": 5290
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.704218528438633e-06,
+      "loss": 1.2213,
+      "step": 5291
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.703458326768553e-06,
+      "loss": 0.9837,
+      "step": 5292
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.702698125098473e-06,
+      "loss": 0.9949,
+      "step": 5293
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.701937923428393e-06,
+      "loss": 1.2394,
+      "step": 5294
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.7011777217583125e-06,
+      "loss": 0.7816,
+      "step": 5295
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.700417520088231e-06,
+      "loss": 1.0534,
+      "step": 5296
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.6996573184181505e-06,
+      "loss": 1.2381,
+      "step": 5297
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.69889711674807e-06,
+      "loss": 1.1478,
+      "step": 5298
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.69813691507799e-06,
+      "loss": 1.0296,
+      "step": 5299
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.69737671340791e-06,
+      "loss": 1.0022,
+      "step": 5300
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.696616511737829e-06,
+      "loss": 1.322,
+      "step": 5301
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.695856310067749e-06,
+      "loss": 0.9673,
+      "step": 5302
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.695096108397669e-06,
+      "loss": 1.323,
+      "step": 5303
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.694335906727589e-06,
+      "loss": 1.0504,
+      "step": 5304
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.6935757050575085e-06,
+      "loss": 1.1588,
+      "step": 5305
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.6928155033874275e-06,
+      "loss": 0.8511,
+      "step": 5306
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.692055301717347e-06,
+      "loss": 1.228,
+      "step": 5307
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.691295100047267e-06,
+      "loss": 1.0307,
+      "step": 5308
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.690534898377187e-06,
+      "loss": 1.0475,
+      "step": 5309
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.689774696707107e-06,
+      "loss": 1.1511,
+      "step": 5310
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.689014495037025e-06,
+      "loss": 1.0616,
+      "step": 5311
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.688254293366945e-06,
+      "loss": 0.9044,
+      "step": 5312
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.687494091696865e-06,
+      "loss": 0.8873,
+      "step": 5313
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.686733890026785e-06,
+      "loss": 1.2557,
+      "step": 5314
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.6859736883567045e-06,
+      "loss": 1.048,
+      "step": 5315
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.685213486686624e-06,
+      "loss": 0.9083,
+      "step": 5316
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.684453285016543e-06,
+      "loss": 1.2783,
+      "step": 5317
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.683693083346463e-06,
+      "loss": 1.1333,
+      "step": 5318
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.682932881676383e-06,
+      "loss": 1.3227,
+      "step": 5319
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.682172680006303e-06,
+      "loss": 1.0656,
+      "step": 5320
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.681412478336223e-06,
+      "loss": 0.9113,
+      "step": 5321
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.680652276666142e-06,
+      "loss": 1.1387,
+      "step": 5322
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.679892074996062e-06,
+      "loss": 1.0087,
+      "step": 5323
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.6791318733259815e-06,
+      "loss": 0.9986,
+      "step": 5324
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.678371671655901e-06,
+      "loss": 1.2725,
+      "step": 5325
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.677611469985821e-06,
+      "loss": 0.9962,
+      "step": 5326
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.676851268315739e-06,
+      "loss": 0.9183,
+      "step": 5327
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.676091066645659e-06,
+      "loss": 1.1378,
+      "step": 5328
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.675330864975579e-06,
+      "loss": 1.0902,
+      "step": 5329
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.674570663305499e-06,
+      "loss": 0.949,
+      "step": 5330
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.673810461635419e-06,
+      "loss": 1.0795,
+      "step": 5331
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.673050259965339e-06,
+      "loss": 0.9168,
+      "step": 5332
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.672290058295258e-06,
+      "loss": 1.0988,
+      "step": 5333
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.6715298566251775e-06,
+      "loss": 0.8832,
+      "step": 5334
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.670769654955097e-06,
+      "loss": 1.0166,
+      "step": 5335
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.670009453285017e-06,
+      "loss": 0.9111,
+      "step": 5336
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.669249251614937e-06,
+      "loss": 1.0246,
+      "step": 5337
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.668489049944856e-06,
+      "loss": 1.0444,
+      "step": 5338
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.667728848274776e-06,
+      "loss": 1.2201,
+      "step": 5339
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.666968646604696e-06,
+      "loss": 0.9384,
+      "step": 5340
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.666208444934616e-06,
+      "loss": 0.8501,
+      "step": 5341
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.6654482432645355e-06,
+      "loss": 1.2842,
+      "step": 5342
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.664688041594454e-06,
+      "loss": 1.109,
+      "step": 5343
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.6639278399243735e-06,
+      "loss": 1.0066,
+      "step": 5344
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.663167638254293e-06,
+      "loss": 1.3321,
+      "step": 5345
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.662407436584213e-06,
+      "loss": 0.8292,
+      "step": 5346
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.661647234914133e-06,
+      "loss": 1.2978,
+      "step": 5347
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.660887033244053e-06,
+      "loss": 1.0404,
+      "step": 5348
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.660126831573972e-06,
+      "loss": 1.411,
+      "step": 5349
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.659366629903892e-06,
+      "loss": 1.1382,
+      "step": 5350
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.658606428233812e-06,
+      "loss": 0.8322,
+      "step": 5351
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.6578462265637315e-06,
+      "loss": 0.9897,
+      "step": 5352
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.657086024893651e-06,
+      "loss": 1.0593,
+      "step": 5353
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.65632582322357e-06,
+      "loss": 1.1005,
+      "step": 5354
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.65556562155349e-06,
+      "loss": 1.0654,
+      "step": 5355
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.65480541988341e-06,
+      "loss": 1.0336,
+      "step": 5356
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.65404521821333e-06,
+      "loss": 1.0407,
+      "step": 5357
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.65328501654325e-06,
+      "loss": 1.0165,
+      "step": 5358
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.652524814873169e-06,
+      "loss": 1.0312,
+      "step": 5359
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.651764613203089e-06,
+      "loss": 1.1219,
+      "step": 5360
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.6510044115330086e-06,
+      "loss": 0.8412,
+      "step": 5361
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.650244209862928e-06,
+      "loss": 1.0905,
+      "step": 5362
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.649484008192847e-06,
+      "loss": 0.9867,
+      "step": 5363
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.648723806522767e-06,
+      "loss": 0.9421,
+      "step": 5364
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.647963604852686e-06,
+      "loss": 0.9756,
+      "step": 5365
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.647203403182606e-06,
+      "loss": 1.079,
+      "step": 5366
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.646443201512526e-06,
+      "loss": 1.1754,
+      "step": 5367
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.645682999842446e-06,
+      "loss": 1.1842,
+      "step": 5368
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.644922798172366e-06,
+      "loss": 1.1488,
+      "step": 5369
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.644162596502285e-06,
+      "loss": 0.8711,
+      "step": 5370
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.6434023948322046e-06,
+      "loss": 1.217,
+      "step": 5371
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.6426421931621244e-06,
+      "loss": 0.9564,
+      "step": 5372
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.641881991492044e-06,
+      "loss": 0.9435,
+      "step": 5373
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.641121789821964e-06,
+      "loss": 1.1184,
+      "step": 5374
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.640361588151883e-06,
+      "loss": 1.113,
+      "step": 5375
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.639601386481803e-06,
+      "loss": 1.2174,
+      "step": 5376
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.638841184811723e-06,
+      "loss": 0.8542,
+      "step": 5377
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.638080983141643e-06,
+      "loss": 1.1953,
+      "step": 5378
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.637320781471563e-06,
+      "loss": 1.1173,
+      "step": 5379
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.6365605798014824e-06,
+      "loss": 0.9372,
+      "step": 5380
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.635800378131401e-06,
+      "loss": 1.2331,
+      "step": 5381
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.6350401764613205e-06,
+      "loss": 1.109,
+      "step": 5382
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.63427997479124e-06,
+      "loss": 1.0866,
+      "step": 5383
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.63351977312116e-06,
+      "loss": 1.1903,
+      "step": 5384
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.63275957145108e-06,
+      "loss": 0.8253,
+      "step": 5385
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.631999369780999e-06,
+      "loss": 1.0475,
+      "step": 5386
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.631239168110919e-06,
+      "loss": 1.157,
+      "step": 5387
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.630478966440839e-06,
+      "loss": 0.8651,
+      "step": 5388
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.629718764770759e-06,
+      "loss": 1.0667,
+      "step": 5389
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.6289585631006785e-06,
+      "loss": 1.0808,
+      "step": 5390
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.6281983614305975e-06,
+      "loss": 1.32,
+      "step": 5391
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.627438159760517e-06,
+      "loss": 1.0416,
+      "step": 5392
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.626677958090437e-06,
+      "loss": 0.9601,
+      "step": 5393
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.625917756420357e-06,
+      "loss": 0.9415,
+      "step": 5394
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.625157554750277e-06,
+      "loss": 0.9619,
+      "step": 5395
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.624397353080197e-06,
+      "loss": 1.1101,
+      "step": 5396
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.623637151410115e-06,
+      "loss": 1.3398,
+      "step": 5397
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.622876949740035e-06,
+      "loss": 1.2303,
+      "step": 5398
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.622116748069955e-06,
+      "loss": 1.1153,
+      "step": 5399
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.6213565463998745e-06,
+      "loss": 0.935,
+      "step": 5400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.620596344729794e-06,
+      "loss": 0.9114,
+      "step": 5401
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.619836143059713e-06,
+      "loss": 1.0421,
+      "step": 5402
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.619075941389633e-06,
+      "loss": 1.0011,
+      "step": 5403
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.618315739719553e-06,
+      "loss": 0.8937,
+      "step": 5404
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.617555538049473e-06,
+      "loss": 0.9931,
+      "step": 5405
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.616795336379393e-06,
+      "loss": 1.1112,
+      "step": 5406
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.616035134709312e-06,
+      "loss": 0.9453,
+      "step": 5407
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.615274933039232e-06,
+      "loss": 1.3376,
+      "step": 5408
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.6145147313691515e-06,
+      "loss": 1.1084,
+      "step": 5409
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.613754529699071e-06,
+      "loss": 1.0258,
+      "step": 5410
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.612994328028991e-06,
+      "loss": 1.3228,
+      "step": 5411
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.612234126358911e-06,
+      "loss": 1.0881,
+      "step": 5412
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.611473924688829e-06,
+      "loss": 1.3124,
+      "step": 5413
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.610713723018749e-06,
+      "loss": 0.8808,
+      "step": 5414
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.609953521348669e-06,
+      "loss": 1.076,
+      "step": 5415
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.609193319678589e-06,
+      "loss": 0.9809,
+      "step": 5416
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.608433118008509e-06,
+      "loss": 1.1108,
+      "step": 5417
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.607672916338428e-06,
+      "loss": 1.2061,
+      "step": 5418
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.6069127146683475e-06,
+      "loss": 1.2316,
+      "step": 5419
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.606152512998267e-06,
+      "loss": 1.1385,
+      "step": 5420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.605392311328187e-06,
+      "loss": 0.8747,
+      "step": 5421
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.604632109658107e-06,
+      "loss": 1.1377,
+      "step": 5422
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.603871907988026e-06,
+      "loss": 1.0787,
+      "step": 5423
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.603111706317946e-06,
+      "loss": 1.1089,
+      "step": 5424
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.602351504647866e-06,
+      "loss": 0.974,
+      "step": 5425
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.601591302977786e-06,
+      "loss": 0.9646,
+      "step": 5426
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.6008311013077055e-06,
+      "loss": 1.1,
+      "step": 5427
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.600070899637625e-06,
+      "loss": 1.1623,
+      "step": 5428
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.5993106979675435e-06,
+      "loss": 1.1462,
+      "step": 5429
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.598550496297463e-06,
+      "loss": 1.1081,
+      "step": 5430
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.597790294627383e-06,
+      "loss": 1.0839,
+      "step": 5431
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.597030092957303e-06,
+      "loss": 0.9627,
+      "step": 5432
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.596269891287223e-06,
+      "loss": 0.9711,
+      "step": 5433
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.595509689617142e-06,
+      "loss": 0.9165,
+      "step": 5434
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.594749487947062e-06,
+      "loss": 0.9524,
+      "step": 5435
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.593989286276982e-06,
+      "loss": 0.9057,
+      "step": 5436
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.5932290846069015e-06,
+      "loss": 0.9382,
+      "step": 5437
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.592468882936821e-06,
+      "loss": 1.0426,
+      "step": 5438
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.59170868126674e-06,
+      "loss": 1.2961,
+      "step": 5439
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.59094847959666e-06,
+      "loss": 1.1864,
+      "step": 5440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.59018827792658e-06,
+      "loss": 1.0787,
+      "step": 5441
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.5894280762565e-06,
+      "loss": 1.0737,
+      "step": 5442
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.58866787458642e-06,
+      "loss": 1.0158,
+      "step": 5443
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.58790767291634e-06,
+      "loss": 1.0676,
+      "step": 5444
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.587147471246259e-06,
+      "loss": 1.0693,
+      "step": 5445
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.586387269576178e-06,
+      "loss": 1.1143,
+      "step": 5446
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.5856270679060975e-06,
+      "loss": 1.0959,
+      "step": 5447
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.584866866236017e-06,
+      "loss": 1.1888,
+      "step": 5448
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.584106664565937e-06,
+      "loss": 0.9068,
+      "step": 5449
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.583346462895856e-06,
+      "loss": 1.1595,
+      "step": 5450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.582586261225776e-06,
+      "loss": 1.0438,
+      "step": 5451
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.581826059555696e-06,
+      "loss": 1.0361,
+      "step": 5452
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.581065857885616e-06,
+      "loss": 1.1434,
+      "step": 5453
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.580305656215536e-06,
+      "loss": 1.059,
+      "step": 5454
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.579545454545455e-06,
+      "loss": 0.9039,
+      "step": 5455
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.5787852528753745e-06,
+      "loss": 1.2491,
+      "step": 5456
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.578025051205294e-06,
+      "loss": 1.0548,
+      "step": 5457
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.577264849535214e-06,
+      "loss": 0.9966,
+      "step": 5458
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.576504647865134e-06,
+      "loss": 0.8246,
+      "step": 5459
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.575744446195054e-06,
+      "loss": 0.9677,
+      "step": 5460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.574984244524973e-06,
+      "loss": 1.0984,
+      "step": 5461
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.574224042854893e-06,
+      "loss": 1.0966,
+      "step": 5462
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.573463841184813e-06,
+      "loss": 0.9531,
+      "step": 5463
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.5727036395147325e-06,
+      "loss": 1.1494,
+      "step": 5464
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.571943437844652e-06,
+      "loss": 0.9568,
+      "step": 5465
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.5711832361745706e-06,
+      "loss": 1.1841,
+      "step": 5466
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.5704230345044904e-06,
+      "loss": 1.0421,
+      "step": 5467
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.56966283283441e-06,
+      "loss": 1.0071,
+      "step": 5468
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.56890263116433e-06,
+      "loss": 1.0101,
+      "step": 5469
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.56814242949425e-06,
+      "loss": 1.0556,
+      "step": 5470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.567382227824169e-06,
+      "loss": 1.1737,
+      "step": 5471
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.566622026154089e-06,
+      "loss": 0.8816,
+      "step": 5472
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.565861824484009e-06,
+      "loss": 1.1678,
+      "step": 5473
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.5651016228139286e-06,
+      "loss": 1.0647,
+      "step": 5474
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.5643414211438484e-06,
+      "loss": 1.0654,
+      "step": 5475
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.563581219473768e-06,
+      "loss": 1.0352,
+      "step": 5476
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.562821017803687e-06,
+      "loss": 0.8582,
+      "step": 5477
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.562060816133607e-06,
+      "loss": 1.137,
+      "step": 5478
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.561300614463527e-06,
+      "loss": 0.9828,
+      "step": 5479
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.560540412793447e-06,
+      "loss": 0.9275,
+      "step": 5480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.559780211123367e-06,
+      "loss": 1.2225,
+      "step": 5481
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.559020009453285e-06,
+      "loss": 1.0193,
+      "step": 5482
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.558259807783205e-06,
+      "loss": 1.0851,
+      "step": 5483
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.557499606113125e-06,
+      "loss": 1.125,
+      "step": 5484
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.5567394044430444e-06,
+      "loss": 0.9315,
+      "step": 5485
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.555979202772964e-06,
+      "loss": 1.046,
+      "step": 5486
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.555219001102883e-06,
+      "loss": 0.8623,
+      "step": 5487
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.554458799432803e-06,
+      "loss": 1.2358,
+      "step": 5488
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.553698597762723e-06,
+      "loss": 0.8929,
+      "step": 5489
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.552938396092643e-06,
+      "loss": 1.1661,
+      "step": 5490
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.552178194422563e-06,
+      "loss": 0.9951,
+      "step": 5491
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.551417992752483e-06,
+      "loss": 1.0837,
+      "step": 5492
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.550657791082402e-06,
+      "loss": 1.0888,
+      "step": 5493
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.5498975894123215e-06,
+      "loss": 1.1871,
+      "step": 5494
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.549137387742241e-06,
+      "loss": 0.9718,
+      "step": 5495
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.548377186072161e-06,
+      "loss": 1.1271,
+      "step": 5496
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.547616984402081e-06,
+      "loss": 1.0939,
+      "step": 5497
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.546856782731999e-06,
+      "loss": 0.8979,
+      "step": 5498
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.546096581061919e-06,
+      "loss": 1.1108,
+      "step": 5499
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.545336379391839e-06,
+      "loss": 1.0985,
+      "step": 5500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.544576177721759e-06,
+      "loss": 1.0286,
+      "step": 5501
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.543815976051679e-06,
+      "loss": 1.0665,
+      "step": 5502
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.543055774381598e-06,
+      "loss": 1.2019,
+      "step": 5503
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.5422955727115175e-06,
+      "loss": 0.9428,
+      "step": 5504
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.541535371041437e-06,
+      "loss": 1.1064,
+      "step": 5505
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.540775169371357e-06,
+      "loss": 0.93,
+      "step": 5506
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.540014967701277e-06,
+      "loss": 1.1117,
+      "step": 5507
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.539254766031197e-06,
+      "loss": 0.9564,
+      "step": 5508
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.538494564361116e-06,
+      "loss": 0.9249,
+      "step": 5509
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.537734362691036e-06,
+      "loss": 1.1016,
+      "step": 5510
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.536974161020956e-06,
+      "loss": 1.0209,
+      "step": 5511
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.5362139593508755e-06,
+      "loss": 1.0747,
+      "step": 5512
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.535453757680795e-06,
+      "loss": 0.9774,
+      "step": 5513
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.5346935560107135e-06,
+      "loss": 1.2066,
+      "step": 5514
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.533933354340633e-06,
+      "loss": 1.0685,
+      "step": 5515
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.533173152670553e-06,
+      "loss": 1.0101,
+      "step": 5516
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.532412951000473e-06,
+      "loss": 0.8987,
+      "step": 5517
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.531652749330393e-06,
+      "loss": 1.1937,
+      "step": 5518
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.530892547660312e-06,
+      "loss": 1.1319,
+      "step": 5519
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.530132345990232e-06,
+      "loss": 1.0494,
+      "step": 5520
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.529372144320152e-06,
+      "loss": 0.9253,
+      "step": 5521
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.5286119426500715e-06,
+      "loss": 1.0908,
+      "step": 5522
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.527851740979991e-06,
+      "loss": 1.2423,
+      "step": 5523
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.52709153930991e-06,
+      "loss": 1.0411,
+      "step": 5524
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.52633133763983e-06,
+      "loss": 1.161,
+      "step": 5525
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.52557113596975e-06,
+      "loss": 0.9948,
+      "step": 5526
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.52481093429967e-06,
+      "loss": 1.0791,
+      "step": 5527
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.52405073262959e-06,
+      "loss": 1.1543,
+      "step": 5528
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.52329053095951e-06,
+      "loss": 1.1426,
+      "step": 5529
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.522530329289428e-06,
+      "loss": 0.9091,
+      "step": 5530
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.521770127619348e-06,
+      "loss": 1.1393,
+      "step": 5531
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.5210099259492675e-06,
+      "loss": 0.9458,
+      "step": 5532
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.520249724279187e-06,
+      "loss": 1.008,
+      "step": 5533
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.519489522609107e-06,
+      "loss": 1.0657,
+      "step": 5534
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.518729320939026e-06,
+      "loss": 1.212,
+      "step": 5535
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.517969119268946e-06,
+      "loss": 0.9773,
+      "step": 5536
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.517208917598866e-06,
+      "loss": 0.8776,
+      "step": 5537
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.516448715928786e-06,
+      "loss": 1.0962,
+      "step": 5538
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.515688514258706e-06,
+      "loss": 1.1381,
+      "step": 5539
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.514928312588625e-06,
+      "loss": 1.1055,
+      "step": 5540
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.5141681109185445e-06,
+      "loss": 1.0358,
+      "step": 5541
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.513407909248464e-06,
+      "loss": 1.1726,
+      "step": 5542
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.512647707578384e-06,
+      "loss": 0.9705,
+      "step": 5543
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.511887505908304e-06,
+      "loss": 0.9103,
+      "step": 5544
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.511127304238224e-06,
+      "loss": 1.015,
+      "step": 5545
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.510367102568143e-06,
+      "loss": 0.9633,
+      "step": 5546
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.509606900898063e-06,
+      "loss": 1.1595,
+      "step": 5547
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.508846699227983e-06,
+      "loss": 1.0233,
+      "step": 5548
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.5080864975579025e-06,
+      "loss": 1.1802,
+      "step": 5549
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.507326295887822e-06,
+      "loss": 0.9647,
+      "step": 5550
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.5065660942177405e-06,
+      "loss": 1.119,
+      "step": 5551
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.50580589254766e-06,
+      "loss": 1.0136,
+      "step": 5552
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.50504569087758e-06,
+      "loss": 1.0609,
+      "step": 5553
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.5042854892075e-06,
+      "loss": 1.0538,
+      "step": 5554
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.50352528753742e-06,
+      "loss": 1.1652,
+      "step": 5555
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.502765085867339e-06,
+      "loss": 0.8727,
+      "step": 5556
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.502004884197259e-06,
+      "loss": 1.0779,
+      "step": 5557
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.501244682527179e-06,
+      "loss": 1.0796,
+      "step": 5558
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.5004844808570985e-06,
+      "loss": 1.0378,
+      "step": 5559
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.499724279187018e-06,
+      "loss": 1.1172,
+      "step": 5560
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.498964077516938e-06,
+      "loss": 0.8423,
+      "step": 5561
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.498203875846857e-06,
+      "loss": 1.0549,
+      "step": 5562
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.497443674176777e-06,
+      "loss": 0.9417,
+      "step": 5563
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.496683472506697e-06,
+      "loss": 1.0616,
+      "step": 5564
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.495923270836617e-06,
+      "loss": 1.0712,
+      "step": 5565
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.495163069166537e-06,
+      "loss": 1.1173,
+      "step": 5566
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.494402867496455e-06,
+      "loss": 0.7657,
+      "step": 5567
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.493642665826375e-06,
+      "loss": 1.0155,
+      "step": 5568
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.4928824641562946e-06,
+      "loss": 1.0525,
+      "step": 5569
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.492122262486214e-06,
+      "loss": 1.1625,
+      "step": 5570
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.491362060816134e-06,
+      "loss": 0.9715,
+      "step": 5571
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.490601859146053e-06,
+      "loss": 1.0278,
+      "step": 5572
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.489841657475973e-06,
+      "loss": 0.9448,
+      "step": 5573
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.489081455805893e-06,
+      "loss": 1.117,
+      "step": 5574
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.488321254135813e-06,
+      "loss": 1.1426,
+      "step": 5575
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.487561052465733e-06,
+      "loss": 1.0557,
+      "step": 5576
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.4868008507956526e-06,
+      "loss": 1.0497,
+      "step": 5577
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.4860406491255716e-06,
+      "loss": 1.0662,
+      "step": 5578
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.4852804474554914e-06,
+      "loss": 0.9012,
+      "step": 5579
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.484520245785411e-06,
+      "loss": 1.0032,
+      "step": 5580
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.483760044115331e-06,
+      "loss": 0.9904,
+      "step": 5581
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.482999842445251e-06,
+      "loss": 0.9728,
+      "step": 5582
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.482239640775169e-06,
+      "loss": 0.8384,
+      "step": 5583
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.481479439105089e-06,
+      "loss": 1.0227,
+      "step": 5584
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.480719237435009e-06,
+      "loss": 1.1117,
+      "step": 5585
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.479959035764929e-06,
+      "loss": 0.9752,
+      "step": 5586
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.479198834094849e-06,
+      "loss": 0.968,
+      "step": 5587
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.478438632424768e-06,
+      "loss": 1.1473,
+      "step": 5588
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.4776784307546874e-06,
+      "loss": 1.2455,
+      "step": 5589
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.476918229084607e-06,
+      "loss": 1.2482,
+      "step": 5590
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.476158027414527e-06,
+      "loss": 1.0644,
+      "step": 5591
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.475397825744447e-06,
+      "loss": 1.2213,
+      "step": 5592
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.474637624074367e-06,
+      "loss": 1.0672,
+      "step": 5593
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.473877422404286e-06,
+      "loss": 1.0893,
+      "step": 5594
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.473117220734206e-06,
+      "loss": 1.235,
+      "step": 5595
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.472357019064126e-06,
+      "loss": 1.121,
+      "step": 5596
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.4715968173940454e-06,
+      "loss": 1.0337,
+      "step": 5597
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.470836615723965e-06,
+      "loss": 0.9923,
+      "step": 5598
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.4700764140538835e-06,
+      "loss": 0.9959,
+      "step": 5599
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.469316212383803e-06,
+      "loss": 1.0775,
+      "step": 5600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.468556010713723e-06,
+      "loss": 1.2224,
+      "step": 5601
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.467795809043643e-06,
+      "loss": 1.1008,
+      "step": 5602
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.467035607373563e-06,
+      "loss": 1.1004,
+      "step": 5603
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.466275405703482e-06,
+      "loss": 1.216,
+      "step": 5604
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.465515204033402e-06,
+      "loss": 1.2133,
+      "step": 5605
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.464755002363322e-06,
+      "loss": 1.1341,
+      "step": 5606
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.4639948006932415e-06,
+      "loss": 1.0453,
+      "step": 5607
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.463234599023161e-06,
+      "loss": 0.9285,
+      "step": 5608
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.462474397353081e-06,
+      "loss": 1.1779,
+      "step": 5609
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.461714195683e-06,
+      "loss": 1.0922,
+      "step": 5610
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.46095399401292e-06,
+      "loss": 1.1694,
+      "step": 5611
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.46019379234284e-06,
+      "loss": 0.9508,
+      "step": 5612
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.45943359067276e-06,
+      "loss": 1.1685,
+      "step": 5613
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.45867338900268e-06,
+      "loss": 1.0387,
+      "step": 5614
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.457913187332598e-06,
+      "loss": 1.1865,
+      "step": 5615
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.457152985662518e-06,
+      "loss": 1.3128,
+      "step": 5616
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.4563927839924375e-06,
+      "loss": 1.0525,
+      "step": 5617
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.455632582322357e-06,
+      "loss": 1.1769,
+      "step": 5618
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.454872380652277e-06,
+      "loss": 0.9376,
+      "step": 5619
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.454112178982196e-06,
+      "loss": 1.0671,
+      "step": 5620
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.453351977312116e-06,
+      "loss": 1.1202,
+      "step": 5621
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.452591775642036e-06,
+      "loss": 0.9884,
+      "step": 5622
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.451831573971956e-06,
+      "loss": 1.0339,
+      "step": 5623
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.451071372301876e-06,
+      "loss": 0.8113,
+      "step": 5624
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.4503111706317955e-06,
+      "loss": 1.1529,
+      "step": 5625
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.4495509689617145e-06,
+      "loss": 0.9946,
+      "step": 5626
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.448790767291634e-06,
+      "loss": 1.0157,
+      "step": 5627
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.448030565621554e-06,
+      "loss": 1.2465,
+      "step": 5628
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.447270363951474e-06,
+      "loss": 1.1704,
+      "step": 5629
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.446510162281394e-06,
+      "loss": 0.9415,
+      "step": 5630
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.445749960611313e-06,
+      "loss": 1.1091,
+      "step": 5631
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.444989758941233e-06,
+      "loss": 0.9249,
+      "step": 5632
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.444229557271153e-06,
+      "loss": 1.0263,
+      "step": 5633
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.443469355601072e-06,
+      "loss": 0.9881,
+      "step": 5634
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.4427091539309915e-06,
+      "loss": 1.0967,
+      "step": 5635
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.4419489522609105e-06,
+      "loss": 1.161,
+      "step": 5636
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.44118875059083e-06,
+      "loss": 1.105,
+      "step": 5637
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.44042854892075e-06,
+      "loss": 0.9586,
+      "step": 5638
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.43966834725067e-06,
+      "loss": 0.8567,
+      "step": 5639
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.43890814558059e-06,
+      "loss": 0.9614,
+      "step": 5640
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.43814794391051e-06,
+      "loss": 0.9342,
+      "step": 5641
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.437387742240429e-06,
+      "loss": 1.0128,
+      "step": 5642
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.436627540570349e-06,
+      "loss": 0.8877,
+      "step": 5643
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.4358673389002685e-06,
+      "loss": 0.8723,
+      "step": 5644
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.435107137230188e-06,
+      "loss": 0.9916,
+      "step": 5645
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.434346935560108e-06,
+      "loss": 1.0293,
+      "step": 5646
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.433586733890027e-06,
+      "loss": 0.8356,
+      "step": 5647
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.432826532219947e-06,
+      "loss": 0.865,
+      "step": 5648
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.432066330549867e-06,
+      "loss": 0.7847,
+      "step": 5649
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.431306128879787e-06,
+      "loss": 0.7762,
+      "step": 5650
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.430545927209707e-06,
+      "loss": 0.7616,
+      "step": 5651
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.429785725539625e-06,
+      "loss": 0.8277,
+      "step": 5652
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.429025523869545e-06,
+      "loss": 0.7824,
+      "step": 5653
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.4282653221994645e-06,
+      "loss": 0.9354,
+      "step": 5654
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.427505120529384e-06,
+      "loss": 0.8363,
+      "step": 5655
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.426744918859304e-06,
+      "loss": 0.8554,
+      "step": 5656
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.425984717189224e-06,
+      "loss": 0.7687,
+      "step": 5657
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.425224515519143e-06,
+      "loss": 0.8245,
+      "step": 5658
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.424464313849063e-06,
+      "loss": 0.9632,
+      "step": 5659
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.423704112178983e-06,
+      "loss": 0.8909,
+      "step": 5660
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.422943910508903e-06,
+      "loss": 1.0306,
+      "step": 5661
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.4221837088388225e-06,
+      "loss": 0.9341,
+      "step": 5662
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.4214235071687415e-06,
+      "loss": 0.8779,
+      "step": 5663
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.420663305498661e-06,
+      "loss": 0.8768,
+      "step": 5664
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.419903103828581e-06,
+      "loss": 0.8,
+      "step": 5665
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.419142902158501e-06,
+      "loss": 0.8907,
+      "step": 5666
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.418382700488421e-06,
+      "loss": 0.7561,
+      "step": 5667
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.417622498818339e-06,
+      "loss": 0.8261,
+      "step": 5668
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.416862297148259e-06,
+      "loss": 0.8295,
+      "step": 5669
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.416102095478179e-06,
+      "loss": 0.7364,
+      "step": 5670
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.415341893808099e-06,
+      "loss": 0.8689,
+      "step": 5671
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.4145816921380186e-06,
+      "loss": 0.8303,
+      "step": 5672
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.413821490467938e-06,
+      "loss": 0.7504,
+      "step": 5673
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.413061288797857e-06,
+      "loss": 0.811,
+      "step": 5674
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.412301087127777e-06,
+      "loss": 0.9062,
+      "step": 5675
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.411540885457697e-06,
+      "loss": 0.8525,
+      "step": 5676
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.410780683787617e-06,
+      "loss": 0.7794,
+      "step": 5677
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.410020482117537e-06,
+      "loss": 0.8341,
+      "step": 5678
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.409260280447456e-06,
+      "loss": 0.8954,
+      "step": 5679
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.408500078777376e-06,
+      "loss": 0.7123,
+      "step": 5680
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.4077398771072956e-06,
+      "loss": 0.7578,
+      "step": 5681
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.406979675437215e-06,
+      "loss": 1.1013,
+      "step": 5682
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.406219473767135e-06,
+      "loss": 0.6936,
+      "step": 5683
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.4054592720970534e-06,
+      "loss": 0.7773,
+      "step": 5684
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.404699070426973e-06,
+      "loss": 0.7452,
+      "step": 5685
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.403938868756893e-06,
+      "loss": 1.0092,
+      "step": 5686
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.403178667086813e-06,
+      "loss": 0.775,
+      "step": 5687
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.402418465416733e-06,
+      "loss": 1.0272,
+      "step": 5688
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.401658263746653e-06,
+      "loss": 0.9822,
+      "step": 5689
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.400898062076572e-06,
+      "loss": 0.835,
+      "step": 5690
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.400137860406492e-06,
+      "loss": 0.8494,
+      "step": 5691
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.3993776587364114e-06,
+      "loss": 0.7576,
+      "step": 5692
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.398617457066331e-06,
+      "loss": 0.7702,
+      "step": 5693
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.397857255396251e-06,
+      "loss": 0.9524,
+      "step": 5694
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.39709705372617e-06,
+      "loss": 1.0257,
+      "step": 5695
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.39633685205609e-06,
+      "loss": 0.7605,
+      "step": 5696
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.39557665038601e-06,
+      "loss": 0.8066,
+      "step": 5697
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.39481644871593e-06,
+      "loss": 0.8568,
+      "step": 5698
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.39405624704585e-06,
+      "loss": 0.786,
+      "step": 5699
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.393296045375768e-06,
+      "loss": 0.8767,
+      "step": 5700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.392535843705688e-06,
+      "loss": 0.7762,
+      "step": 5701
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.3917756420356075e-06,
+      "loss": 0.6722,
+      "step": 5702
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.391015440365527e-06,
+      "loss": 0.828,
+      "step": 5703
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.390255238695447e-06,
+      "loss": 0.6348,
+      "step": 5704
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.389495037025367e-06,
+      "loss": 0.9474,
+      "step": 5705
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.388734835355286e-06,
+      "loss": 0.7576,
+      "step": 5706
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.387974633685206e-06,
+      "loss": 0.671,
+      "step": 5707
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.387214432015126e-06,
+      "loss": 0.9736,
+      "step": 5708
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.386454230345046e-06,
+      "loss": 0.8287,
+      "step": 5709
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.3856940286749655e-06,
+      "loss": 0.7948,
+      "step": 5710
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.3849338270048845e-06,
+      "loss": 0.8657,
+      "step": 5711
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.384173625334804e-06,
+      "loss": 0.8508,
+      "step": 5712
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.383413423664724e-06,
+      "loss": 0.8927,
+      "step": 5713
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.382653221994644e-06,
+      "loss": 0.7684,
+      "step": 5714
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.381893020324564e-06,
+      "loss": 0.9975,
+      "step": 5715
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.381132818654483e-06,
+      "loss": 0.7626,
+      "step": 5716
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.380372616984402e-06,
+      "loss": 0.7011,
+      "step": 5717
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.379612415314322e-06,
+      "loss": 0.917,
+      "step": 5718
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.378852213644242e-06,
+      "loss": 0.8251,
+      "step": 5719
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.3780920119741615e-06,
+      "loss": 0.9021,
+      "step": 5720
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.3773318103040805e-06,
+      "loss": 0.9014,
+      "step": 5721
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.376571608634e-06,
+      "loss": 0.7944,
+      "step": 5722
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.37581140696392e-06,
+      "loss": 0.761,
+      "step": 5723
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.37505120529384e-06,
+      "loss": 0.8171,
+      "step": 5724
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.37429100362376e-06,
+      "loss": 0.7952,
+      "step": 5725
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.37353080195368e-06,
+      "loss": 0.8729,
+      "step": 5726
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.372770600283599e-06,
+      "loss": 0.8227,
+      "step": 5727
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.372010398613519e-06,
+      "loss": 0.7091,
+      "step": 5728
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.3712501969434385e-06,
+      "loss": 0.7938,
+      "step": 5729
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.370489995273358e-06,
+      "loss": 0.9526,
+      "step": 5730
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.369729793603278e-06,
+      "loss": 0.8558,
+      "step": 5731
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.368969591933197e-06,
+      "loss": 0.9174,
+      "step": 5732
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.368209390263117e-06,
+      "loss": 0.9274,
+      "step": 5733
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.367449188593037e-06,
+      "loss": 0.7741,
+      "step": 5734
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.366688986922957e-06,
+      "loss": 0.8112,
+      "step": 5735
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.365928785252877e-06,
+      "loss": 0.79,
+      "step": 5736
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.365168583582795e-06,
+      "loss": 0.9502,
+      "step": 5737
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.364408381912715e-06,
+      "loss": 0.9022,
+      "step": 5738
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.3636481802426345e-06,
+      "loss": 0.8581,
+      "step": 5739
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.362887978572554e-06,
+      "loss": 0.9557,
+      "step": 5740
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.362127776902474e-06,
+      "loss": 0.952,
+      "step": 5741
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.361367575232394e-06,
+      "loss": 0.779,
+      "step": 5742
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.360607373562313e-06,
+      "loss": 0.8152,
+      "step": 5743
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.359847171892233e-06,
+      "loss": 0.689,
+      "step": 5744
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.359086970222153e-06,
+      "loss": 0.7481,
+      "step": 5745
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.358326768552073e-06,
+      "loss": 0.8548,
+      "step": 5746
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.3575665668819925e-06,
+      "loss": 1.0789,
+      "step": 5747
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.3568063652119115e-06,
+      "loss": 0.8602,
+      "step": 5748
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.356046163541831e-06,
+      "loss": 0.6424,
+      "step": 5749
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.355285961871751e-06,
+      "loss": 0.9069,
+      "step": 5750
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.354525760201671e-06,
+      "loss": 0.8114,
+      "step": 5751
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.353765558531591e-06,
+      "loss": 0.6947,
+      "step": 5752
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.353005356861509e-06,
+      "loss": 0.8801,
+      "step": 5753
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.352245155191429e-06,
+      "loss": 0.7228,
+      "step": 5754
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.351484953521349e-06,
+      "loss": 0.9085,
+      "step": 5755
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.350724751851269e-06,
+      "loss": 1.0149,
+      "step": 5756
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.3499645501811885e-06,
+      "loss": 0.8811,
+      "step": 5757
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.349204348511108e-06,
+      "loss": 0.7257,
+      "step": 5758
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.348444146841027e-06,
+      "loss": 0.7251,
+      "step": 5759
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.347683945170947e-06,
+      "loss": 0.8268,
+      "step": 5760
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.346923743500867e-06,
+      "loss": 0.694,
+      "step": 5761
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.346163541830787e-06,
+      "loss": 0.9078,
+      "step": 5762
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.345403340160707e-06,
+      "loss": 0.9244,
+      "step": 5763
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.344643138490626e-06,
+      "loss": 0.8337,
+      "step": 5764
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.343882936820546e-06,
+      "loss": 0.9182,
+      "step": 5765
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.3431227351504655e-06,
+      "loss": 0.9682,
+      "step": 5766
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.342362533480385e-06,
+      "loss": 0.9438,
+      "step": 5767
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.341602331810305e-06,
+      "loss": 0.9345,
+      "step": 5768
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.340842130140223e-06,
+      "loss": 0.7859,
+      "step": 5769
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.340081928470143e-06,
+      "loss": 0.8478,
+      "step": 5770
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.339321726800063e-06,
+      "loss": 0.7803,
+      "step": 5771
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.338561525129983e-06,
+      "loss": 0.7191,
+      "step": 5772
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.337801323459903e-06,
+      "loss": 0.9391,
+      "step": 5773
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.337041121789823e-06,
+      "loss": 0.8531,
+      "step": 5774
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.336280920119742e-06,
+      "loss": 0.8156,
+      "step": 5775
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.3355207184496616e-06,
+      "loss": 0.758,
+      "step": 5776
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.334760516779581e-06,
+      "loss": 1.0358,
+      "step": 5777
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.334000315109501e-06,
+      "loss": 0.7486,
+      "step": 5778
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.333240113439421e-06,
+      "loss": 0.8841,
+      "step": 5779
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.33247991176934e-06,
+      "loss": 0.8611,
+      "step": 5780
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.33171971009926e-06,
+      "loss": 1.0207,
+      "step": 5781
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.33095950842918e-06,
+      "loss": 0.7438,
+      "step": 5782
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.3301993067591e-06,
+      "loss": 0.8139,
+      "step": 5783
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.3294391050890196e-06,
+      "loss": 0.911,
+      "step": 5784
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.328678903418938e-06,
+      "loss": 0.8496,
+      "step": 5785
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.3279187017488576e-06,
+      "loss": 1.1146,
+      "step": 5786
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.3271585000787774e-06,
+      "loss": 0.7537,
+      "step": 5787
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.326398298408697e-06,
+      "loss": 0.9254,
+      "step": 5788
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.325638096738617e-06,
+      "loss": 1.0058,
+      "step": 5789
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.324877895068537e-06,
+      "loss": 0.8266,
+      "step": 5790
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.324117693398456e-06,
+      "loss": 1.1052,
+      "step": 5791
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.323357491728376e-06,
+      "loss": 0.8539,
+      "step": 5792
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.322597290058296e-06,
+      "loss": 0.7797,
+      "step": 5793
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.3218370883882156e-06,
+      "loss": 0.784,
+      "step": 5794
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.3210768867181354e-06,
+      "loss": 0.9024,
+      "step": 5795
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.3203166850480544e-06,
+      "loss": 0.8271,
+      "step": 5796
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.319556483377974e-06,
+      "loss": 1.0787,
+      "step": 5797
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.318796281707894e-06,
+      "loss": 0.9947,
+      "step": 5798
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.318036080037814e-06,
+      "loss": 0.8078,
+      "step": 5799
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.317275878367734e-06,
+      "loss": 0.9169,
+      "step": 5800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.316515676697652e-06,
+      "loss": 0.7125,
+      "step": 5801
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.315755475027572e-06,
+      "loss": 0.8033,
+      "step": 5802
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.314995273357492e-06,
+      "loss": 0.8905,
+      "step": 5803
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.314235071687412e-06,
+      "loss": 0.8699,
+      "step": 5804
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.3134748700173314e-06,
+      "loss": 0.936,
+      "step": 5805
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.312714668347251e-06,
+      "loss": 0.8397,
+      "step": 5806
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.31195446667717e-06,
+      "loss": 0.8094,
+      "step": 5807
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.31119426500709e-06,
+      "loss": 0.7207,
+      "step": 5808
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.31043406333701e-06,
+      "loss": 0.9034,
+      "step": 5809
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.30967386166693e-06,
+      "loss": 0.9061,
+      "step": 5810
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.30891365999685e-06,
+      "loss": 0.8172,
+      "step": 5811
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.308153458326769e-06,
+      "loss": 0.7389,
+      "step": 5812
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.307393256656689e-06,
+      "loss": 0.738,
+      "step": 5813
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.3066330549866085e-06,
+      "loss": 0.7721,
+      "step": 5814
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.305872853316528e-06,
+      "loss": 0.9465,
+      "step": 5815
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.305112651646448e-06,
+      "loss": 0.8461,
+      "step": 5816
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.304352449976367e-06,
+      "loss": 0.8586,
+      "step": 5817
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.303592248306287e-06,
+      "loss": 0.7668,
+      "step": 5818
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.302832046636207e-06,
+      "loss": 1.0058,
+      "step": 5819
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.302071844966127e-06,
+      "loss": 0.9301,
+      "step": 5820
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.301311643296047e-06,
+      "loss": 0.8501,
+      "step": 5821
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.300551441625966e-06,
+      "loss": 0.8473,
+      "step": 5822
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.299791239955885e-06,
+      "loss": 0.7179,
+      "step": 5823
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.2990310382858045e-06,
+      "loss": 0.8113,
+      "step": 5824
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.298270836615724e-06,
+      "loss": 0.7587,
+      "step": 5825
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.297510634945644e-06,
+      "loss": 0.9099,
+      "step": 5826
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.296750433275564e-06,
+      "loss": 0.8856,
+      "step": 5827
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.295990231605483e-06,
+      "loss": 0.8776,
+      "step": 5828
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.295230029935403e-06,
+      "loss": 0.9387,
+      "step": 5829
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.294469828265323e-06,
+      "loss": 0.623,
+      "step": 5830
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.293709626595243e-06,
+      "loss": 0.8632,
+      "step": 5831
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.2929494249251625e-06,
+      "loss": 0.9622,
+      "step": 5832
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.2921892232550815e-06,
+      "loss": 0.8198,
+      "step": 5833
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.291429021585001e-06,
+      "loss": 0.817,
+      "step": 5834
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.290668819914921e-06,
+      "loss": 0.716,
+      "step": 5835
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.289908618244841e-06,
+      "loss": 0.9527,
+      "step": 5836
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.289148416574761e-06,
+      "loss": 0.7877,
+      "step": 5837
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.288388214904681e-06,
+      "loss": 0.924,
+      "step": 5838
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.287628013234599e-06,
+      "loss": 1.0498,
+      "step": 5839
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.286867811564519e-06,
+      "loss": 0.9384,
+      "step": 5840
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.286107609894439e-06,
+      "loss": 0.9117,
+      "step": 5841
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.2853474082243585e-06,
+      "loss": 0.8473,
+      "step": 5842
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.284587206554278e-06,
+      "loss": 0.8965,
+      "step": 5843
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.283827004884197e-06,
+      "loss": 0.7617,
+      "step": 5844
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.283066803214117e-06,
+      "loss": 1.0781,
+      "step": 5845
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.282306601544037e-06,
+      "loss": 0.916,
+      "step": 5846
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.281546399873957e-06,
+      "loss": 0.7073,
+      "step": 5847
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.280786198203877e-06,
+      "loss": 0.9728,
+      "step": 5848
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.280025996533796e-06,
+      "loss": 0.7841,
+      "step": 5849
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.279265794863716e-06,
+      "loss": 0.9804,
+      "step": 5850
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.2785055931936355e-06,
+      "loss": 0.7629,
+      "step": 5851
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.277745391523555e-06,
+      "loss": 0.8069,
+      "step": 5852
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.276985189853475e-06,
+      "loss": 1.0019,
+      "step": 5853
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.276224988183395e-06,
+      "loss": 0.8719,
+      "step": 5854
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.275464786513313e-06,
+      "loss": 1.1121,
+      "step": 5855
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.274704584843233e-06,
+      "loss": 0.7495,
+      "step": 5856
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.273944383173153e-06,
+      "loss": 0.9133,
+      "step": 5857
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.273184181503073e-06,
+      "loss": 0.6893,
+      "step": 5858
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.272423979832993e-06,
+      "loss": 0.8767,
+      "step": 5859
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.271663778162912e-06,
+      "loss": 0.9504,
+      "step": 5860
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.2709035764928315e-06,
+      "loss": 0.8584,
+      "step": 5861
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.270143374822751e-06,
+      "loss": 0.8228,
+      "step": 5862
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.269383173152671e-06,
+      "loss": 0.7734,
+      "step": 5863
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.268622971482591e-06,
+      "loss": 0.7167,
+      "step": 5864
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.26786276981251e-06,
+      "loss": 0.8928,
+      "step": 5865
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.26710256814243e-06,
+      "loss": 0.7639,
+      "step": 5866
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.26634236647235e-06,
+      "loss": 0.8615,
+      "step": 5867
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.26558216480227e-06,
+      "loss": 0.7948,
+      "step": 5868
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.2648219631321895e-06,
+      "loss": 0.7503,
+      "step": 5869
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.264061761462109e-06,
+      "loss": 0.8691,
+      "step": 5870
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.2633015597920275e-06,
+      "loss": 0.859,
+      "step": 5871
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.262541358121947e-06,
+      "loss": 0.9072,
+      "step": 5872
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.261781156451867e-06,
+      "loss": 0.9408,
+      "step": 5873
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.261020954781787e-06,
+      "loss": 0.6576,
+      "step": 5874
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.260260753111707e-06,
+      "loss": 0.8987,
+      "step": 5875
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.259500551441626e-06,
+      "loss": 0.9406,
+      "step": 5876
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.258740349771546e-06,
+      "loss": 1.02,
+      "step": 5877
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.257980148101466e-06,
+      "loss": 0.8305,
+      "step": 5878
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.2572199464313855e-06,
+      "loss": 0.9148,
+      "step": 5879
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.256459744761305e-06,
+      "loss": 0.7219,
+      "step": 5880
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.255699543091224e-06,
+      "loss": 0.9085,
+      "step": 5881
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.254939341421144e-06,
+      "loss": 0.8189,
+      "step": 5882
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.254179139751064e-06,
+      "loss": 0.695,
+      "step": 5883
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.253418938080984e-06,
+      "loss": 0.8854,
+      "step": 5884
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.252658736410904e-06,
+      "loss": 0.9414,
+      "step": 5885
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.251898534740824e-06,
+      "loss": 0.9396,
+      "step": 5886
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.251138333070742e-06,
+      "loss": 0.9529,
+      "step": 5887
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.250378131400662e-06,
+      "loss": 0.8379,
+      "step": 5888
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2496179297305816e-06,
+      "loss": 0.7005,
+      "step": 5889
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.248857728060501e-06,
+      "loss": 0.9154,
+      "step": 5890
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.248097526390421e-06,
+      "loss": 0.8062,
+      "step": 5891
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.24733732472034e-06,
+      "loss": 0.9711,
+      "step": 5892
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.24657712305026e-06,
+      "loss": 0.8646,
+      "step": 5893
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.24581692138018e-06,
+      "loss": 0.9108,
+      "step": 5894
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2450567197101e-06,
+      "loss": 0.7461,
+      "step": 5895
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.24429651804002e-06,
+      "loss": 0.8016,
+      "step": 5896
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.243536316369939e-06,
+      "loss": 0.9639,
+      "step": 5897
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2427761146998586e-06,
+      "loss": 0.8769,
+      "step": 5898
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2420159130297784e-06,
+      "loss": 0.8037,
+      "step": 5899
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.241255711359698e-06,
+      "loss": 0.817,
+      "step": 5900
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.240495509689618e-06,
+      "loss": 0.9757,
+      "step": 5901
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.239735308019538e-06,
+      "loss": 0.8688,
+      "step": 5902
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.238975106349457e-06,
+      "loss": 0.815,
+      "step": 5903
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.238214904679377e-06,
+      "loss": 0.7423,
+      "step": 5904
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.237454703009296e-06,
+      "loss": 0.8332,
+      "step": 5905
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.236694501339216e-06,
+      "loss": 0.8242,
+      "step": 5906
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.235934299669136e-06,
+      "loss": 0.8982,
+      "step": 5907
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.235174097999055e-06,
+      "loss": 0.7839,
+      "step": 5908
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2344138963289745e-06,
+      "loss": 0.9548,
+      "step": 5909
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.233653694658894e-06,
+      "loss": 0.7725,
+      "step": 5910
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.232893492988814e-06,
+      "loss": 0.8707,
+      "step": 5911
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.232133291318734e-06,
+      "loss": 0.7082,
+      "step": 5912
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.231373089648653e-06,
+      "loss": 0.9567,
+      "step": 5913
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.230612887978573e-06,
+      "loss": 0.7731,
+      "step": 5914
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.229852686308493e-06,
+      "loss": 0.7836,
+      "step": 5915
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.229092484638413e-06,
+      "loss": 0.6677,
+      "step": 5916
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.2283322829683325e-06,
+      "loss": 0.8005,
+      "step": 5917
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.227572081298252e-06,
+      "loss": 0.9262,
+      "step": 5918
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.226811879628171e-06,
+      "loss": 0.8817,
+      "step": 5919
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.226051677958091e-06,
+      "loss": 0.9605,
+      "step": 5920
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.225291476288011e-06,
+      "loss": 0.887,
+      "step": 5921
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.224531274617931e-06,
+      "loss": 0.8238,
+      "step": 5922
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.223771072947851e-06,
+      "loss": 1.0214,
+      "step": 5923
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.223010871277769e-06,
+      "loss": 1.0238,
+      "step": 5924
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.222250669607689e-06,
+      "loss": 0.9413,
+      "step": 5925
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.221490467937609e-06,
+      "loss": 0.9131,
+      "step": 5926
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.2207302662675285e-06,
+      "loss": 0.871,
+      "step": 5927
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.219970064597448e-06,
+      "loss": 0.6158,
+      "step": 5928
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.219209862927367e-06,
+      "loss": 0.8006,
+      "step": 5929
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.218449661257287e-06,
+      "loss": 0.7272,
+      "step": 5930
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.217689459587207e-06,
+      "loss": 0.8559,
+      "step": 5931
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.216929257917127e-06,
+      "loss": 0.8405,
+      "step": 5932
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.216169056247047e-06,
+      "loss": 0.7117,
+      "step": 5933
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.215408854576966e-06,
+      "loss": 0.8788,
+      "step": 5934
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.214648652906886e-06,
+      "loss": 0.839,
+      "step": 5935
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.2138884512368055e-06,
+      "loss": 0.7942,
+      "step": 5936
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.213128249566725e-06,
+      "loss": 0.8879,
+      "step": 5937
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.212368047896645e-06,
+      "loss": 0.9042,
+      "step": 5938
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.211607846226565e-06,
+      "loss": 0.7943,
+      "step": 5939
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.210847644556483e-06,
+      "loss": 0.7287,
+      "step": 5940
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.210087442886403e-06,
+      "loss": 0.9156,
+      "step": 5941
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.209327241216323e-06,
+      "loss": 0.9361,
+      "step": 5942
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.208567039546243e-06,
+      "loss": 0.906,
+      "step": 5943
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.207806837876163e-06,
+      "loss": 0.785,
+      "step": 5944
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.207046636206082e-06,
+      "loss": 0.8428,
+      "step": 5945
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.2062864345360015e-06,
+      "loss": 0.7521,
+      "step": 5946
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.205526232865921e-06,
+      "loss": 1.0865,
+      "step": 5947
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.204766031195841e-06,
+      "loss": 0.9042,
+      "step": 5948
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.204005829525761e-06,
+      "loss": 0.8067,
+      "step": 5949
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.20324562785568e-06,
+      "loss": 0.703,
+      "step": 5950
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.2024854261856e-06,
+      "loss": 1.0394,
+      "step": 5951
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.20172522451552e-06,
+      "loss": 0.9216,
+      "step": 5952
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.20096502284544e-06,
+      "loss": 0.9096,
+      "step": 5953
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.2002048211753595e-06,
+      "loss": 0.8621,
+      "step": 5954
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.199444619505279e-06,
+      "loss": 0.811,
+      "step": 5955
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.1986844178351975e-06,
+      "loss": 0.7228,
+      "step": 5956
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.197924216165117e-06,
+      "loss": 0.8961,
+      "step": 5957
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.197164014495037e-06,
+      "loss": 0.7404,
+      "step": 5958
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.196403812824957e-06,
+      "loss": 0.9354,
+      "step": 5959
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.195643611154877e-06,
+      "loss": 0.8271,
+      "step": 5960
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.194883409484796e-06,
+      "loss": 0.7904,
+      "step": 5961
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.194123207814716e-06,
+      "loss": 0.7848,
+      "step": 5962
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.193363006144636e-06,
+      "loss": 0.9227,
+      "step": 5963
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.1926028044745555e-06,
+      "loss": 0.9341,
+      "step": 5964
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.191842602804475e-06,
+      "loss": 0.9387,
+      "step": 5965
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.191082401134394e-06,
+      "loss": 0.889,
+      "step": 5966
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.190322199464314e-06,
+      "loss": 0.7734,
+      "step": 5967
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.189561997794234e-06,
+      "loss": 0.9541,
+      "step": 5968
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.188801796124154e-06,
+      "loss": 1.0233,
+      "step": 5969
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.188041594454074e-06,
+      "loss": 0.8085,
+      "step": 5970
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.187281392783994e-06,
+      "loss": 0.7725,
+      "step": 5971
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.186521191113912e-06,
+      "loss": 0.6208,
+      "step": 5972
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.185760989443832e-06,
+      "loss": 0.8851,
+      "step": 5973
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.1850007877737515e-06,
+      "loss": 0.7188,
+      "step": 5974
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.184240586103671e-06,
+      "loss": 0.7493,
+      "step": 5975
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.183480384433591e-06,
+      "loss": 0.8636,
+      "step": 5976
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.18272018276351e-06,
+      "loss": 1.0323,
+      "step": 5977
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.18195998109343e-06,
+      "loss": 0.8043,
+      "step": 5978
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.18119977942335e-06,
+      "loss": 1.0144,
+      "step": 5979
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.18043957775327e-06,
+      "loss": 1.0541,
+      "step": 5980
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.17967937608319e-06,
+      "loss": 0.8262,
+      "step": 5981
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.178919174413109e-06,
+      "loss": 0.9798,
+      "step": 5982
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.1781589727430285e-06,
+      "loss": 1.0328,
+      "step": 5983
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.177398771072948e-06,
+      "loss": 0.7778,
+      "step": 5984
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.176638569402868e-06,
+      "loss": 0.9027,
+      "step": 5985
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.175878367732788e-06,
+      "loss": 0.8288,
+      "step": 5986
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.175118166062708e-06,
+      "loss": 0.8567,
+      "step": 5987
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.174357964392626e-06,
+      "loss": 0.9043,
+      "step": 5988
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.173597762722546e-06,
+      "loss": 1.0038,
+      "step": 5989
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.172837561052466e-06,
+      "loss": 0.9271,
+      "step": 5990
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.172077359382386e-06,
+      "loss": 0.7866,
+      "step": 5991
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.1713171577123056e-06,
+      "loss": 0.6999,
+      "step": 5992
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.1705569560422246e-06,
+      "loss": 0.8419,
+      "step": 5993
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.1697967543721444e-06,
+      "loss": 1.17,
+      "step": 5994
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.169036552702064e-06,
+      "loss": 0.8226,
+      "step": 5995
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.168276351031984e-06,
+      "loss": 0.7521,
+      "step": 5996
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.167516149361904e-06,
+      "loss": 0.7106,
+      "step": 5997
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.166755947691823e-06,
+      "loss": 0.8101,
+      "step": 5998
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.165995746021743e-06,
+      "loss": 0.6088,
+      "step": 5999
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.165235544351663e-06,
+      "loss": 0.8305,
+      "step": 6000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.1644753426815826e-06,
+      "loss": 0.8648,
+      "step": 6001
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.1637151410115024e-06,
+      "loss": 0.8879,
+      "step": 6002
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.162954939341422e-06,
+      "loss": 0.8842,
+      "step": 6003
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.162194737671341e-06,
+      "loss": 0.8448,
+      "step": 6004
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.161434536001261e-06,
+      "loss": 0.9241,
+      "step": 6005
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.160674334331181e-06,
+      "loss": 0.7932,
+      "step": 6006
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.159914132661101e-06,
+      "loss": 0.6255,
+      "step": 6007
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.159153930991021e-06,
+      "loss": 0.7904,
+      "step": 6008
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.158393729320939e-06,
+      "loss": 0.9649,
+      "step": 6009
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.157633527650859e-06,
+      "loss": 0.6896,
+      "step": 6010
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.156873325980779e-06,
+      "loss": 0.9896,
+      "step": 6011
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.1561131243106984e-06,
+      "loss": 0.7535,
+      "step": 6012
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.155352922640618e-06,
+      "loss": 0.7635,
+      "step": 6013
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.154592720970537e-06,
+      "loss": 0.6425,
+      "step": 6014
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.153832519300457e-06,
+      "loss": 1.1044,
+      "step": 6015
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.153072317630377e-06,
+      "loss": 0.956,
+      "step": 6016
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.152312115960297e-06,
+      "loss": 1.0868,
+      "step": 6017
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.151551914290217e-06,
+      "loss": 0.9131,
+      "step": 6018
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.150791712620137e-06,
+      "loss": 0.746,
+      "step": 6019
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.150031510950056e-06,
+      "loss": 0.8225,
+      "step": 6020
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.1492713092799755e-06,
+      "loss": 0.8212,
+      "step": 6021
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.148511107609895e-06,
+      "loss": 0.8624,
+      "step": 6022
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.147750905939815e-06,
+      "loss": 0.8418,
+      "step": 6023
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.146990704269735e-06,
+      "loss": 0.9071,
+      "step": 6024
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.146230502599653e-06,
+      "loss": 1.041,
+      "step": 6025
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.145470300929573e-06,
+      "loss": 1.0208,
+      "step": 6026
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.144710099259493e-06,
+      "loss": 0.9198,
+      "step": 6027
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.143949897589413e-06,
+      "loss": 0.7793,
+      "step": 6028
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.143189695919333e-06,
+      "loss": 0.847,
+      "step": 6029
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.142429494249252e-06,
+      "loss": 0.8424,
+      "step": 6030
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.1416692925791715e-06,
+      "loss": 0.8416,
+      "step": 6031
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.140909090909091e-06,
+      "loss": 0.8052,
+      "step": 6032
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.140148889239011e-06,
+      "loss": 0.7987,
+      "step": 6033
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.139388687568931e-06,
+      "loss": 0.7493,
+      "step": 6034
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.138628485898851e-06,
+      "loss": 0.7983,
+      "step": 6035
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.13786828422877e-06,
+      "loss": 0.7117,
+      "step": 6036
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.13710808255869e-06,
+      "loss": 0.724,
+      "step": 6037
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.13634788088861e-06,
+      "loss": 0.6956,
+      "step": 6038
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.1355876792185295e-06,
+      "loss": 1.0085,
+      "step": 6039
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.134827477548449e-06,
+      "loss": 0.9454,
+      "step": 6040
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.1340672758783675e-06,
+      "loss": 0.7706,
+      "step": 6041
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.133307074208287e-06,
+      "loss": 0.8592,
+      "step": 6042
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.132546872538207e-06,
+      "loss": 1.0032,
+      "step": 6043
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.131786670868127e-06,
+      "loss": 0.8888,
+      "step": 6044
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.131026469198047e-06,
+      "loss": 0.8693,
+      "step": 6045
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.130266267527966e-06,
+      "loss": 0.851,
+      "step": 6046
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.129506065857886e-06,
+      "loss": 0.9103,
+      "step": 6047
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.128745864187806e-06,
+      "loss": 0.9077,
+      "step": 6048
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.1279856625177255e-06,
+      "loss": 0.8889,
+      "step": 6049
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.127225460847645e-06,
+      "loss": 0.8259,
+      "step": 6050
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.126465259177565e-06,
+      "loss": 0.7978,
+      "step": 6051
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.125705057507484e-06,
+      "loss": 0.7267,
+      "step": 6052
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.124944855837404e-06,
+      "loss": 0.6348,
+      "step": 6053
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.124184654167324e-06,
+      "loss": 0.7802,
+      "step": 6054
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.123424452497244e-06,
+      "loss": 0.9109,
+      "step": 6055
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.122664250827164e-06,
+      "loss": 1.0408,
+      "step": 6056
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.121904049157082e-06,
+      "loss": 0.9503,
+      "step": 6057
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.121143847487002e-06,
+      "loss": 0.9774,
+      "step": 6058
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.1203836458169215e-06,
+      "loss": 0.768,
+      "step": 6059
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.119623444146841e-06,
+      "loss": 0.7846,
+      "step": 6060
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.118863242476761e-06,
+      "loss": 1.1016,
+      "step": 6061
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.11810304080668e-06,
+      "loss": 0.7863,
+      "step": 6062
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.1173428391366e-06,
+      "loss": 0.7644,
+      "step": 6063
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.11658263746652e-06,
+      "loss": 0.6536,
+      "step": 6064
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.11582243579644e-06,
+      "loss": 0.8921,
+      "step": 6065
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.11506223412636e-06,
+      "loss": 0.9754,
+      "step": 6066
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.1143020324562795e-06,
+      "loss": 0.7858,
+      "step": 6067
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.1135418307861985e-06,
+      "loss": 0.9589,
+      "step": 6068
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.112781629116118e-06,
+      "loss": 1.0862,
+      "step": 6069
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.112021427446038e-06,
+      "loss": 0.8781,
+      "step": 6070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.111261225775958e-06,
+      "loss": 0.8684,
+      "step": 6071
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.110501024105878e-06,
+      "loss": 0.7251,
+      "step": 6072
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.109740822435796e-06,
+      "loss": 0.8685,
+      "step": 6073
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.108980620765716e-06,
+      "loss": 0.9106,
+      "step": 6074
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.108220419095636e-06,
+      "loss": 0.7239,
+      "step": 6075
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.107460217425556e-06,
+      "loss": 0.8797,
+      "step": 6076
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.1067000157554755e-06,
+      "loss": 0.696,
+      "step": 6077
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.1059398140853945e-06,
+      "loss": 0.8554,
+      "step": 6078
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.105179612415314e-06,
+      "loss": 0.7723,
+      "step": 6079
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.104419410745234e-06,
+      "loss": 0.8817,
+      "step": 6080
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.103659209075154e-06,
+      "loss": 0.7134,
+      "step": 6081
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.102899007405074e-06,
+      "loss": 0.9493,
+      "step": 6082
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.102138805734994e-06,
+      "loss": 0.7137,
+      "step": 6083
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.101378604064913e-06,
+      "loss": 0.8815,
+      "step": 6084
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.100618402394833e-06,
+      "loss": 0.6752,
+      "step": 6085
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.0998582007247525e-06,
+      "loss": 0.9848,
+      "step": 6086
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.099097999054672e-06,
+      "loss": 0.8872,
+      "step": 6087
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.098337797384592e-06,
+      "loss": 0.7214,
+      "step": 6088
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.097577595714511e-06,
+      "loss": 0.8197,
+      "step": 6089
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.096817394044431e-06,
+      "loss": 0.8493,
+      "step": 6090
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.096057192374351e-06,
+      "loss": 0.8458,
+      "step": 6091
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.095296990704271e-06,
+      "loss": 0.8216,
+      "step": 6092
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.094536789034191e-06,
+      "loss": 0.7998,
+      "step": 6093
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.093776587364109e-06,
+      "loss": 0.8274,
+      "step": 6094
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.093016385694029e-06,
+      "loss": 0.9757,
+      "step": 6095
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.0922561840239486e-06,
+      "loss": 0.6872,
+      "step": 6096
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.091495982353868e-06,
+      "loss": 0.7077,
+      "step": 6097
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.090735780683788e-06,
+      "loss": 0.9764,
+      "step": 6098
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.089975579013708e-06,
+      "loss": 1.0354,
+      "step": 6099
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.089215377343627e-06,
+      "loss": 0.7952,
+      "step": 6100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.088455175673547e-06,
+      "loss": 0.9529,
+      "step": 6101
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.087694974003467e-06,
+      "loss": 0.7055,
+      "step": 6102
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.086934772333387e-06,
+      "loss": 1.003,
+      "step": 6103
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.0861745706633066e-06,
+      "loss": 0.7282,
+      "step": 6104
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.0854143689932256e-06,
+      "loss": 0.8017,
+      "step": 6105
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.0846541673231454e-06,
+      "loss": 1.0092,
+      "step": 6106
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.083893965653065e-06,
+      "loss": 0.8844,
+      "step": 6107
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.083133763982985e-06,
+      "loss": 0.9242,
+      "step": 6108
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.082373562312905e-06,
+      "loss": 0.771,
+      "step": 6109
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.081613360642823e-06,
+      "loss": 1.0832,
+      "step": 6110
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.080853158972743e-06,
+      "loss": 0.7216,
+      "step": 6111
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.080092957302663e-06,
+      "loss": 0.8644,
+      "step": 6112
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.079332755632583e-06,
+      "loss": 0.8402,
+      "step": 6113
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.078572553962503e-06,
+      "loss": 0.9047,
+      "step": 6114
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.0778123522924224e-06,
+      "loss": 0.7407,
+      "step": 6115
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.0770521506223414e-06,
+      "loss": 0.8727,
+      "step": 6116
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.076291948952261e-06,
+      "loss": 0.659,
+      "step": 6117
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.075531747282181e-06,
+      "loss": 0.7823,
+      "step": 6118
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.074771545612101e-06,
+      "loss": 0.8095,
+      "step": 6119
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.074011343942021e-06,
+      "loss": 0.8799,
+      "step": 6120
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.07325114227194e-06,
+      "loss": 0.9283,
+      "step": 6121
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.07249094060186e-06,
+      "loss": 0.6949,
+      "step": 6122
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.07173073893178e-06,
+      "loss": 0.7893,
+      "step": 6123
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.0709705372616994e-06,
+      "loss": 0.8628,
+      "step": 6124
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.070210335591619e-06,
+      "loss": 0.7217,
+      "step": 6125
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.0694501339215375e-06,
+      "loss": 0.9033,
+      "step": 6126
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.068689932251457e-06,
+      "loss": 0.7982,
+      "step": 6127
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.067929730581377e-06,
+      "loss": 0.8579,
+      "step": 6128
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.067169528911297e-06,
+      "loss": 0.6469,
+      "step": 6129
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.066409327241217e-06,
+      "loss": 0.7272,
+      "step": 6130
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.065649125571136e-06,
+      "loss": 0.968,
+      "step": 6131
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.064888923901056e-06,
+      "loss": 0.868,
+      "step": 6132
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.064128722230976e-06,
+      "loss": 0.8381,
+      "step": 6133
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.0633685205608955e-06,
+      "loss": 1.029,
+      "step": 6134
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.062608318890815e-06,
+      "loss": 0.67,
+      "step": 6135
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.061848117220735e-06,
+      "loss": 0.7847,
+      "step": 6136
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.061087915550654e-06,
+      "loss": 0.8168,
+      "step": 6137
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.060327713880574e-06,
+      "loss": 0.7468,
+      "step": 6138
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.059567512210494e-06,
+      "loss": 0.9171,
+      "step": 6139
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.058807310540414e-06,
+      "loss": 0.661,
+      "step": 6140
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.058047108870334e-06,
+      "loss": 0.9069,
+      "step": 6141
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.057286907200252e-06,
+      "loss": 0.9034,
+      "step": 6142
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.056526705530172e-06,
+      "loss": 0.7637,
+      "step": 6143
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.0557665038600915e-06,
+      "loss": 0.7515,
+      "step": 6144
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.055006302190011e-06,
+      "loss": 0.7239,
+      "step": 6145
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.054246100519931e-06,
+      "loss": 0.9053,
+      "step": 6146
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.05348589884985e-06,
+      "loss": 0.8254,
+      "step": 6147
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.05272569717977e-06,
+      "loss": 0.795,
+      "step": 6148
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.05196549550969e-06,
+      "loss": 0.9995,
+      "step": 6149
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.05120529383961e-06,
+      "loss": 0.7959,
+      "step": 6150
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.05044509216953e-06,
+      "loss": 0.8013,
+      "step": 6151
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.0496848904994495e-06,
+      "loss": 0.9503,
+      "step": 6152
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.0489246888293685e-06,
+      "loss": 0.8893,
+      "step": 6153
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.048164487159288e-06,
+      "loss": 0.8562,
+      "step": 6154
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.047404285489208e-06,
+      "loss": 0.9228,
+      "step": 6155
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.046644083819128e-06,
+      "loss": 0.816,
+      "step": 6156
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.045883882149048e-06,
+      "loss": 0.9098,
+      "step": 6157
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.045123680478966e-06,
+      "loss": 0.8546,
+      "step": 6158
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.044363478808886e-06,
+      "loss": 0.7676,
+      "step": 6159
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.043603277138806e-06,
+      "loss": 0.8278,
+      "step": 6160
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.042843075468726e-06,
+      "loss": 0.9324,
+      "step": 6161
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.0420828737986455e-06,
+      "loss": 1.0816,
+      "step": 6162
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.0413226721285645e-06,
+      "loss": 0.8343,
+      "step": 6163
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.040562470458484e-06,
+      "loss": 0.8395,
+      "step": 6164
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.039802268788404e-06,
+      "loss": 0.6959,
+      "step": 6165
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.039042067118324e-06,
+      "loss": 0.8412,
+      "step": 6166
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.038281865448244e-06,
+      "loss": 0.8126,
+      "step": 6167
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.037521663778164e-06,
+      "loss": 0.7935,
+      "step": 6168
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.036761462108083e-06,
+      "loss": 0.9288,
+      "step": 6169
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.036001260438003e-06,
+      "loss": 0.77,
+      "step": 6170
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.0352410587679225e-06,
+      "loss": 0.8773,
+      "step": 6171
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.034480857097842e-06,
+      "loss": 0.9416,
+      "step": 6172
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.033720655427762e-06,
+      "loss": 0.9268,
+      "step": 6173
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.032960453757681e-06,
+      "loss": 0.779,
+      "step": 6174
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.032200252087601e-06,
+      "loss": 0.8198,
+      "step": 6175
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.031440050417521e-06,
+      "loss": 0.8194,
+      "step": 6176
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.03067984874744e-06,
+      "loss": 0.967,
+      "step": 6177
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.02991964707736e-06,
+      "loss": 0.789,
+      "step": 6178
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.029159445407279e-06,
+      "loss": 0.9204,
+      "step": 6179
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.028399243737199e-06,
+      "loss": 1.0253,
+      "step": 6180
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.0276390420671185e-06,
+      "loss": 0.7613,
+      "step": 6181
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.026878840397038e-06,
+      "loss": 1.029,
+      "step": 6182
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.026118638726958e-06,
+      "loss": 0.782,
+      "step": 6183
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.025358437056878e-06,
+      "loss": 0.9666,
+      "step": 6184
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.024598235386797e-06,
+      "loss": 0.8395,
+      "step": 6185
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.023838033716717e-06,
+      "loss": 0.7038,
+      "step": 6186
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.023077832046637e-06,
+      "loss": 0.8156,
+      "step": 6187
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.022317630376557e-06,
+      "loss": 0.9039,
+      "step": 6188
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.0215574287064765e-06,
+      "loss": 0.9225,
+      "step": 6189
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.0207972270363955e-06,
+      "loss": 0.9047,
+      "step": 6190
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.020037025366315e-06,
+      "loss": 0.8857,
+      "step": 6191
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.019276823696235e-06,
+      "loss": 0.9538,
+      "step": 6192
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.018516622026155e-06,
+      "loss": 0.6825,
+      "step": 6193
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.017756420356075e-06,
+      "loss": 0.7053,
+      "step": 6194
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.016996218685993e-06,
+      "loss": 0.7825,
+      "step": 6195
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.016236017015913e-06,
+      "loss": 0.8283,
+      "step": 6196
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.015475815345833e-06,
+      "loss": 0.7419,
+      "step": 6197
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.014715613675753e-06,
+      "loss": 0.8424,
+      "step": 6198
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.0139554120056726e-06,
+      "loss": 1.048,
+      "step": 6199
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.013195210335592e-06,
+      "loss": 0.8695,
+      "step": 6200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.012435008665511e-06,
+      "loss": 0.9744,
+      "step": 6201
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.011674806995431e-06,
+      "loss": 0.7598,
+      "step": 6202
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.010914605325351e-06,
+      "loss": 0.8972,
+      "step": 6203
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.010154403655271e-06,
+      "loss": 0.9195,
+      "step": 6204
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.009394201985191e-06,
+      "loss": 0.9055,
+      "step": 6205
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.00863400031511e-06,
+      "loss": 0.6847,
+      "step": 6206
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.00787379864503e-06,
+      "loss": 0.9608,
+      "step": 6207
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.0071135969749496e-06,
+      "loss": 0.9102,
+      "step": 6208
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.006353395304869e-06,
+      "loss": 0.9193,
+      "step": 6209
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.005593193634789e-06,
+      "loss": 0.766,
+      "step": 6210
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.0048329919647074e-06,
+      "loss": 1.0312,
+      "step": 6211
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.004072790294627e-06,
+      "loss": 0.9533,
+      "step": 6212
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.003312588624547e-06,
+      "loss": 0.9236,
+      "step": 6213
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.002552386954467e-06,
+      "loss": 0.939,
+      "step": 6214
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.001792185284387e-06,
+      "loss": 0.8927,
+      "step": 6215
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.001031983614307e-06,
+      "loss": 1.0038,
+      "step": 6216
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.000271781944226e-06,
+      "loss": 0.9319,
+      "step": 6217
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.999511580274146e-06,
+      "loss": 0.7951,
+      "step": 6218
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.9987513786040654e-06,
+      "loss": 0.9206,
+      "step": 6219
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.997991176933985e-06,
+      "loss": 1.0093,
+      "step": 6220
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.997230975263905e-06,
+      "loss": 1.0061,
+      "step": 6221
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.996470773593824e-06,
+      "loss": 1.0686,
+      "step": 6222
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.995710571923744e-06,
+      "loss": 0.8312,
+      "step": 6223
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.994950370253664e-06,
+      "loss": 0.7262,
+      "step": 6224
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.994190168583584e-06,
+      "loss": 0.9937,
+      "step": 6225
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.993429966913504e-06,
+      "loss": 0.9042,
+      "step": 6226
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.992669765243422e-06,
+      "loss": 0.8381,
+      "step": 6227
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.991909563573342e-06,
+      "loss": 0.7833,
+      "step": 6228
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.9911493619032615e-06,
+      "loss": 0.8167,
+      "step": 6229
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.990389160233181e-06,
+      "loss": 0.8646,
+      "step": 6230
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.989628958563101e-06,
+      "loss": 0.8432,
+      "step": 6231
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.988868756893021e-06,
+      "loss": 0.7316,
+      "step": 6232
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.98810855522294e-06,
+      "loss": 0.8452,
+      "step": 6233
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.98734835355286e-06,
+      "loss": 0.9721,
+      "step": 6234
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.98658815188278e-06,
+      "loss": 0.7167,
+      "step": 6235
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.9858279502127e-06,
+      "loss": 0.8938,
+      "step": 6236
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.9850677485426195e-06,
+      "loss": 0.7602,
+      "step": 6237
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.9843075468725385e-06,
+      "loss": 0.9508,
+      "step": 6238
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.983547345202458e-06,
+      "loss": 0.8122,
+      "step": 6239
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.982787143532378e-06,
+      "loss": 0.76,
+      "step": 6240
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.982026941862298e-06,
+      "loss": 0.7725,
+      "step": 6241
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.981266740192218e-06,
+      "loss": 1.0942,
+      "step": 6242
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.980506538522136e-06,
+      "loss": 0.7732,
+      "step": 6243
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.979746336852056e-06,
+      "loss": 0.6881,
+      "step": 6244
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.978986135181976e-06,
+      "loss": 0.8811,
+      "step": 6245
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.978225933511896e-06,
+      "loss": 0.7841,
+      "step": 6246
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.9774657318418155e-06,
+      "loss": 1.062,
+      "step": 6247
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.976705530171735e-06,
+      "loss": 0.7758,
+      "step": 6248
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.975945328501654e-06,
+      "loss": 0.9718,
+      "step": 6249
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.975185126831574e-06,
+      "loss": 0.8293,
+      "step": 6250
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.974424925161494e-06,
+      "loss": 0.7657,
+      "step": 6251
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.973664723491414e-06,
+      "loss": 0.7323,
+      "step": 6252
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.972904521821334e-06,
+      "loss": 1.0057,
+      "step": 6253
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.972144320151253e-06,
+      "loss": 0.9777,
+      "step": 6254
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.971384118481173e-06,
+      "loss": 0.7736,
+      "step": 6255
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.9706239168110925e-06,
+      "loss": 0.7863,
+      "step": 6256
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.969863715141012e-06,
+      "loss": 0.8347,
+      "step": 6257
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.969103513470932e-06,
+      "loss": 0.8141,
+      "step": 6258
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.968343311800851e-06,
+      "loss": 0.7196,
+      "step": 6259
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.96758311013077e-06,
+      "loss": 1.0126,
+      "step": 6260
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.96682290846069e-06,
+      "loss": 1.0067,
+      "step": 6261
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.96606270679061e-06,
+      "loss": 0.9136,
+      "step": 6262
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.96530250512053e-06,
+      "loss": 0.8287,
+      "step": 6263
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.96454230345045e-06,
+      "loss": 0.8092,
+      "step": 6264
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.963782101780369e-06,
+      "loss": 0.9293,
+      "step": 6265
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.9630219001102885e-06,
+      "loss": 0.8502,
+      "step": 6266
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.962261698440208e-06,
+      "loss": 1.0057,
+      "step": 6267
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.961501496770128e-06,
+      "loss": 0.916,
+      "step": 6268
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.960741295100048e-06,
+      "loss": 0.6699,
+      "step": 6269
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.959981093429967e-06,
+      "loss": 0.7662,
+      "step": 6270
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.959220891759887e-06,
+      "loss": 0.9253,
+      "step": 6271
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.958460690089807e-06,
+      "loss": 0.9377,
+      "step": 6272
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.957700488419727e-06,
+      "loss": 0.7747,
+      "step": 6273
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.9569402867496465e-06,
+      "loss": 0.8479,
+      "step": 6274
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.9561800850795655e-06,
+      "loss": 0.8026,
+      "step": 6275
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.955419883409485e-06,
+      "loss": 0.8121,
+      "step": 6276
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.954659681739405e-06,
+      "loss": 0.9611,
+      "step": 6277
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.953899480069325e-06,
+      "loss": 0.9176,
+      "step": 6278
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.953139278399245e-06,
+      "loss": 0.6953,
+      "step": 6279
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.952379076729165e-06,
+      "loss": 0.9283,
+      "step": 6280
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.951618875059083e-06,
+      "loss": 0.9996,
+      "step": 6281
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.950858673389003e-06,
+      "loss": 0.8465,
+      "step": 6282
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.950098471718923e-06,
+      "loss": 0.7914,
+      "step": 6283
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.9493382700488425e-06,
+      "loss": 0.7197,
+      "step": 6284
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.948578068378762e-06,
+      "loss": 0.9064,
+      "step": 6285
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.947817866708681e-06,
+      "loss": 0.8279,
+      "step": 6286
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.947057665038601e-06,
+      "loss": 0.6892,
+      "step": 6287
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.946297463368521e-06,
+      "loss": 0.7841,
+      "step": 6288
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.945537261698441e-06,
+      "loss": 0.8438,
+      "step": 6289
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.944777060028361e-06,
+      "loss": 0.7704,
+      "step": 6290
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.94401685835828e-06,
+      "loss": 0.8261,
+      "step": 6291
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.9432566566882e-06,
+      "loss": 0.9137,
+      "step": 6292
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.9424964550181195e-06,
+      "loss": 0.8714,
+      "step": 6293
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.941736253348039e-06,
+      "loss": 0.7884,
+      "step": 6294
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.940976051677959e-06,
+      "loss": 0.6909,
+      "step": 6295
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.940215850007879e-06,
+      "loss": 0.8749,
+      "step": 6296
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.939455648337797e-06,
+      "loss": 0.9179,
+      "step": 6297
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.938695446667717e-06,
+      "loss": 0.8209,
+      "step": 6298
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.937935244997637e-06,
+      "loss": 0.8664,
+      "step": 6299
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.937175043327557e-06,
+      "loss": 0.9783,
+      "step": 6300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.936414841657477e-06,
+      "loss": 0.8533,
+      "step": 6301
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.935654639987396e-06,
+      "loss": 0.7632,
+      "step": 6302
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.9348944383173156e-06,
+      "loss": 0.769,
+      "step": 6303
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.934134236647235e-06,
+      "loss": 0.9479,
+      "step": 6304
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.933374034977155e-06,
+      "loss": 0.9362,
+      "step": 6305
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.932613833307075e-06,
+      "loss": 0.6693,
+      "step": 6306
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.931853631636994e-06,
+      "loss": 0.8632,
+      "step": 6307
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.931093429966914e-06,
+      "loss": 0.6474,
+      "step": 6308
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.930333228296834e-06,
+      "loss": 0.8922,
+      "step": 6309
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.929573026626754e-06,
+      "loss": 0.8915,
+      "step": 6310
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.9288128249566736e-06,
+      "loss": 0.8667,
+      "step": 6311
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.928052623286593e-06,
+      "loss": 0.8283,
+      "step": 6312
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.9272924216165116e-06,
+      "loss": 0.8707,
+      "step": 6313
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.9265322199464314e-06,
+      "loss": 1.0345,
+      "step": 6314
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.925772018276351e-06,
+      "loss": 0.7673,
+      "step": 6315
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.925011816606271e-06,
+      "loss": 0.9219,
+      "step": 6316
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.924251614936191e-06,
+      "loss": 0.8532,
+      "step": 6317
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.92349141326611e-06,
+      "loss": 0.7687,
+      "step": 6318
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.92273121159603e-06,
+      "loss": 0.7909,
+      "step": 6319
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.92197100992595e-06,
+      "loss": 0.7581,
+      "step": 6320
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.9212108082558696e-06,
+      "loss": 0.8958,
+      "step": 6321
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.9204506065857894e-06,
+      "loss": 0.9077,
+      "step": 6322
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.9196904049157084e-06,
+      "loss": 0.6914,
+      "step": 6323
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.918930203245628e-06,
+      "loss": 0.7903,
+      "step": 6324
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.918170001575548e-06,
+      "loss": 0.6268,
+      "step": 6325
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.917409799905468e-06,
+      "loss": 0.7107,
+      "step": 6326
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.916649598235388e-06,
+      "loss": 0.7964,
+      "step": 6327
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.915889396565308e-06,
+      "loss": 0.9749,
+      "step": 6328
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.915129194895226e-06,
+      "loss": 0.681,
+      "step": 6329
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.914368993225146e-06,
+      "loss": 0.7521,
+      "step": 6330
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.913608791555066e-06,
+      "loss": 0.9697,
+      "step": 6331
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.9128485898849855e-06,
+      "loss": 0.9591,
+      "step": 6332
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.912088388214905e-06,
+      "loss": 0.8027,
+      "step": 6333
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.911328186544824e-06,
+      "loss": 0.9027,
+      "step": 6334
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.910567984874744e-06,
+      "loss": 0.8556,
+      "step": 6335
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.909807783204664e-06,
+      "loss": 0.8943,
+      "step": 6336
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.909047581534584e-06,
+      "loss": 0.874,
+      "step": 6337
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.908287379864504e-06,
+      "loss": 0.8545,
+      "step": 6338
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.907527178194423e-06,
+      "loss": 0.9265,
+      "step": 6339
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.906766976524343e-06,
+      "loss": 0.793,
+      "step": 6340
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.9060067748542625e-06,
+      "loss": 0.7632,
+      "step": 6341
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.905246573184182e-06,
+      "loss": 0.8883,
+      "step": 6342
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.904486371514102e-06,
+      "loss": 0.9406,
+      "step": 6343
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.90372616984402e-06,
+      "loss": 1.0039,
+      "step": 6344
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.90296596817394e-06,
+      "loss": 0.9136,
+      "step": 6345
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.90220576650386e-06,
+      "loss": 0.7501,
+      "step": 6346
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.90144556483378e-06,
+      "loss": 0.7787,
+      "step": 6347
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.9006853631637e-06,
+      "loss": 0.8631,
+      "step": 6348
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.89992516149362e-06,
+      "loss": 0.8962,
+      "step": 6349
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.899164959823539e-06,
+      "loss": 0.7901,
+      "step": 6350
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.8984047581534585e-06,
+      "loss": 0.8341,
+      "step": 6351
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.897644556483378e-06,
+      "loss": 0.9642,
+      "step": 6352
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.896884354813298e-06,
+      "loss": 0.9364,
+      "step": 6353
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.896124153143218e-06,
+      "loss": 0.7809,
+      "step": 6354
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.895363951473137e-06,
+      "loss": 0.6682,
+      "step": 6355
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.894603749803057e-06,
+      "loss": 0.6611,
+      "step": 6356
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.893843548132977e-06,
+      "loss": 0.8612,
+      "step": 6357
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.893083346462897e-06,
+      "loss": 0.6606,
+      "step": 6358
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.8923231447928165e-06,
+      "loss": 0.8151,
+      "step": 6359
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.8915629431227355e-06,
+      "loss": 0.9677,
+      "step": 6360
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.890802741452655e-06,
+      "loss": 0.8294,
+      "step": 6361
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.890042539782575e-06,
+      "loss": 0.7103,
+      "step": 6362
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.889282338112495e-06,
+      "loss": 0.9841,
+      "step": 6363
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.888522136442415e-06,
+      "loss": 0.8023,
+      "step": 6364
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.887761934772334e-06,
+      "loss": 0.8933,
+      "step": 6365
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.887001733102253e-06,
+      "loss": 0.8909,
+      "step": 6366
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.886241531432173e-06,
+      "loss": 0.8104,
+      "step": 6367
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.885481329762093e-06,
+      "loss": 0.7493,
+      "step": 6368
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.8847211280920125e-06,
+      "loss": 0.862,
+      "step": 6369
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.883960926421932e-06,
+      "loss": 0.7956,
+      "step": 6370
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.883200724751851e-06,
+      "loss": 0.8196,
+      "step": 6371
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.882440523081771e-06,
+      "loss": 0.7579,
+      "step": 6372
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.881680321411691e-06,
+      "loss": 0.9405,
+      "step": 6373
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.880920119741611e-06,
+      "loss": 1.0795,
+      "step": 6374
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.880159918071531e-06,
+      "loss": 0.7798,
+      "step": 6375
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.87939971640145e-06,
+      "loss": 0.8744,
+      "step": 6376
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.87863951473137e-06,
+      "loss": 0.8342,
+      "step": 6377
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.8778793130612895e-06,
+      "loss": 0.8532,
+      "step": 6378
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.877119111391209e-06,
+      "loss": 0.9715,
+      "step": 6379
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.876358909721129e-06,
+      "loss": 0.8379,
+      "step": 6380
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.875598708051049e-06,
+      "loss": 0.8239,
+      "step": 6381
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.874838506380967e-06,
+      "loss": 1.0343,
+      "step": 6382
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.874078304710887e-06,
+      "loss": 0.8568,
+      "step": 6383
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.873318103040807e-06,
+      "loss": 0.802,
+      "step": 6384
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.872557901370727e-06,
+      "loss": 0.9355,
+      "step": 6385
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.871797699700647e-06,
+      "loss": 0.9623,
+      "step": 6386
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.871037498030566e-06,
+      "loss": 0.7896,
+      "step": 6387
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.8702772963604855e-06,
+      "loss": 0.7626,
+      "step": 6388
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.869517094690405e-06,
+      "loss": 0.8807,
+      "step": 6389
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.868756893020325e-06,
+      "loss": 0.8499,
+      "step": 6390
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.867996691350245e-06,
+      "loss": 0.8245,
+      "step": 6391
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.867236489680164e-06,
+      "loss": 0.8507,
+      "step": 6392
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.866476288010084e-06,
+      "loss": 0.8901,
+      "step": 6393
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.865716086340004e-06,
+      "loss": 0.5401,
+      "step": 6394
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.864955884669924e-06,
+      "loss": 0.8053,
+      "step": 6395
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.8641956829998435e-06,
+      "loss": 0.8199,
+      "step": 6396
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.863435481329763e-06,
+      "loss": 0.6739,
+      "step": 6397
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.8626752796596815e-06,
+      "loss": 0.7738,
+      "step": 6398
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.861915077989601e-06,
+      "loss": 0.6557,
+      "step": 6399
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.861154876319521e-06,
+      "loss": 0.7011,
+      "step": 6400
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.860394674649441e-06,
+      "loss": 0.872,
+      "step": 6401
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.859634472979361e-06,
+      "loss": 0.7912,
+      "step": 6402
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.85887427130928e-06,
+      "loss": 0.776,
+      "step": 6403
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.8581140696392e-06,
+      "loss": 0.7506,
+      "step": 6404
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.85735386796912e-06,
+      "loss": 0.8109,
+      "step": 6405
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.8565936662990395e-06,
+      "loss": 0.8957,
+      "step": 6406
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.855833464628959e-06,
+      "loss": 0.8832,
+      "step": 6407
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.855073262958878e-06,
+      "loss": 0.7609,
+      "step": 6408
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.854313061288798e-06,
+      "loss": 0.9922,
+      "step": 6409
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.853552859618718e-06,
+      "loss": 0.8938,
+      "step": 6410
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.852792657948638e-06,
+      "loss": 0.7338,
+      "step": 6411
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.852032456278558e-06,
+      "loss": 0.7453,
+      "step": 6412
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.851272254608478e-06,
+      "loss": 1.0094,
+      "step": 6413
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.850512052938396e-06,
+      "loss": 0.8383,
+      "step": 6414
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.849751851268316e-06,
+      "loss": 0.8738,
+      "step": 6415
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.8489916495982356e-06,
+      "loss": 0.7317,
+      "step": 6416
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.848231447928155e-06,
+      "loss": 0.9386,
+      "step": 6417
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.847471246258075e-06,
+      "loss": 0.7275,
+      "step": 6418
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.846711044587994e-06,
+      "loss": 1.0451,
+      "step": 6419
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.845950842917914e-06,
+      "loss": 0.7954,
+      "step": 6420
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.845190641247834e-06,
+      "loss": 0.8018,
+      "step": 6421
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.844430439577754e-06,
+      "loss": 0.9478,
+      "step": 6422
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.843670237907674e-06,
+      "loss": 0.913,
+      "step": 6423
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.842910036237593e-06,
+      "loss": 0.6908,
+      "step": 6424
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.842149834567513e-06,
+      "loss": 0.8319,
+      "step": 6425
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.8413896328974324e-06,
+      "loss": 0.7525,
+      "step": 6426
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.840629431227352e-06,
+      "loss": 0.7595,
+      "step": 6427
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.839869229557272e-06,
+      "loss": 0.9007,
+      "step": 6428
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.839109027887192e-06,
+      "loss": 0.8069,
+      "step": 6429
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.83834882621711e-06,
+      "loss": 0.7802,
+      "step": 6430
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.83758862454703e-06,
+      "loss": 0.8792,
+      "step": 6431
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.83682842287695e-06,
+      "loss": 0.718,
+      "step": 6432
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.83606822120687e-06,
+      "loss": 0.8336,
+      "step": 6433
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.83530801953679e-06,
+      "loss": 0.9281,
+      "step": 6434
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.834547817866709e-06,
+      "loss": 0.9968,
+      "step": 6435
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.8337876161966285e-06,
+      "loss": 0.79,
+      "step": 6436
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.833027414526548e-06,
+      "loss": 0.9357,
+      "step": 6437
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.832267212856468e-06,
+      "loss": 0.9122,
+      "step": 6438
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.831507011186388e-06,
+      "loss": 1.0026,
+      "step": 6439
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.830746809516307e-06,
+      "loss": 0.8804,
+      "step": 6440
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.829986607846227e-06,
+      "loss": 1.1264,
+      "step": 6441
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.829226406176147e-06,
+      "loss": 0.8415,
+      "step": 6442
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.828466204506067e-06,
+      "loss": 0.7274,
+      "step": 6443
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.8277060028359865e-06,
+      "loss": 0.8961,
+      "step": 6444
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.826945801165906e-06,
+      "loss": 0.8334,
+      "step": 6445
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.826185599495825e-06,
+      "loss": 0.9379,
+      "step": 6446
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.825425397825745e-06,
+      "loss": 1.0199,
+      "step": 6447
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.824665196155664e-06,
+      "loss": 0.7894,
+      "step": 6448
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.823904994485584e-06,
+      "loss": 1.1034,
+      "step": 6449
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.823144792815504e-06,
+      "loss": 0.8759,
+      "step": 6450
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.822384591145423e-06,
+      "loss": 0.8492,
+      "step": 6451
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.821624389475343e-06,
+      "loss": 0.8613,
+      "step": 6452
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.820864187805263e-06,
+      "loss": 0.695,
+      "step": 6453
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.8201039861351825e-06,
+      "loss": 0.8234,
+      "step": 6454
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.819343784465102e-06,
+      "loss": 0.9432,
+      "step": 6455
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.818583582795021e-06,
+      "loss": 0.759,
+      "step": 6456
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.817823381124941e-06,
+      "loss": 0.8425,
+      "step": 6457
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.817063179454861e-06,
+      "loss": 0.7922,
+      "step": 6458
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.816302977784781e-06,
+      "loss": 0.752,
+      "step": 6459
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.815542776114701e-06,
+      "loss": 0.8398,
+      "step": 6460
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.814782574444621e-06,
+      "loss": 0.8018,
+      "step": 6461
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.81402237277454e-06,
+      "loss": 0.7381,
+      "step": 6462
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.8132621711044595e-06,
+      "loss": 0.9223,
+      "step": 6463
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.812501969434379e-06,
+      "loss": 0.9929,
+      "step": 6464
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.811741767764299e-06,
+      "loss": 0.9548,
+      "step": 6465
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.810981566094219e-06,
+      "loss": 0.8216,
+      "step": 6466
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.810221364424137e-06,
+      "loss": 0.6837,
+      "step": 6467
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.809461162754057e-06,
+      "loss": 0.9027,
+      "step": 6468
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.808700961083977e-06,
+      "loss": 0.7866,
+      "step": 6469
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.807940759413897e-06,
+      "loss": 0.861,
+      "step": 6470
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.807180557743817e-06,
+      "loss": 1.014,
+      "step": 6471
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.806420356073736e-06,
+      "loss": 0.8898,
+      "step": 6472
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.8056601544036555e-06,
+      "loss": 0.6519,
+      "step": 6473
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.804899952733575e-06,
+      "loss": 0.8764,
+      "step": 6474
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.804139751063495e-06,
+      "loss": 0.8136,
+      "step": 6475
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.803379549393415e-06,
+      "loss": 0.7965,
+      "step": 6476
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.802619347723335e-06,
+      "loss": 0.9271,
+      "step": 6477
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.801859146053254e-06,
+      "loss": 0.8033,
+      "step": 6478
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.801098944383174e-06,
+      "loss": 1.048,
+      "step": 6479
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.800338742713094e-06,
+      "loss": 0.7449,
+      "step": 6480
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.7995785410430135e-06,
+      "loss": 0.8732,
+      "step": 6481
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.798818339372933e-06,
+      "loss": 0.7924,
+      "step": 6482
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.7980581377028515e-06,
+      "loss": 0.751,
+      "step": 6483
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.797297936032771e-06,
+      "loss": 0.9035,
+      "step": 6484
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.796537734362691e-06,
+      "loss": 0.7871,
+      "step": 6485
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.795777532692611e-06,
+      "loss": 0.7435,
+      "step": 6486
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.795017331022531e-06,
+      "loss": 0.9413,
+      "step": 6487
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.79425712935245e-06,
+      "loss": 0.891,
+      "step": 6488
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.79349692768237e-06,
+      "loss": 0.8685,
+      "step": 6489
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.79273672601229e-06,
+      "loss": 0.7191,
+      "step": 6490
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.7919765243422095e-06,
+      "loss": 0.8641,
+      "step": 6491
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.791216322672129e-06,
+      "loss": 0.7468,
+      "step": 6492
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.790456121002049e-06,
+      "loss": 0.8383,
+      "step": 6493
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.789695919331968e-06,
+      "loss": 1.0968,
+      "step": 6494
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.788935717661888e-06,
+      "loss": 0.7556,
+      "step": 6495
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.788175515991808e-06,
+      "loss": 0.7685,
+      "step": 6496
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.787415314321728e-06,
+      "loss": 0.9201,
+      "step": 6497
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.786655112651648e-06,
+      "loss": 0.9466,
+      "step": 6498
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.785894910981566e-06,
+      "loss": 0.933,
+      "step": 6499
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.785134709311486e-06,
+      "loss": 0.8861,
+      "step": 6500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.7843745076414055e-06,
+      "loss": 0.8293,
+      "step": 6501
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.783614305971325e-06,
+      "loss": 0.7807,
+      "step": 6502
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.782854104301245e-06,
+      "loss": 0.7454,
+      "step": 6503
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.782093902631164e-06,
+      "loss": 0.7797,
+      "step": 6504
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.781333700961084e-06,
+      "loss": 0.8587,
+      "step": 6505
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.780573499291004e-06,
+      "loss": 0.7542,
+      "step": 6506
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.779813297620924e-06,
+      "loss": 0.8796,
+      "step": 6507
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.779053095950844e-06,
+      "loss": 0.8656,
+      "step": 6508
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.7782928942807635e-06,
+      "loss": 0.8461,
+      "step": 6509
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.7775326926106825e-06,
+      "loss": 0.8488,
+      "step": 6510
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.776772490940602e-06,
+      "loss": 0.7905,
+      "step": 6511
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.776012289270522e-06,
+      "loss": 0.8349,
+      "step": 6512
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.775252087600442e-06,
+      "loss": 0.9822,
+      "step": 6513
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.774491885930362e-06,
+      "loss": 0.9026,
+      "step": 6514
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.77373168426028e-06,
+      "loss": 0.8553,
+      "step": 6515
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.7729714825902e-06,
+      "loss": 0.7102,
+      "step": 6516
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.77221128092012e-06,
+      "loss": 1.0157,
+      "step": 6517
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.77145107925004e-06,
+      "loss": 0.7922,
+      "step": 6518
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.7706908775799596e-06,
+      "loss": 0.7304,
+      "step": 6519
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.7699306759098786e-06,
+      "loss": 0.7821,
+      "step": 6520
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.7691704742397984e-06,
+      "loss": 0.7915,
+      "step": 6521
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.768410272569718e-06,
+      "loss": 0.8913,
+      "step": 6522
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.767650070899638e-06,
+      "loss": 0.8459,
+      "step": 6523
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.766889869229558e-06,
+      "loss": 0.7444,
+      "step": 6524
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.766129667559478e-06,
+      "loss": 0.9129,
+      "step": 6525
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.765369465889397e-06,
+      "loss": 0.8417,
+      "step": 6526
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.764609264219317e-06,
+      "loss": 0.7607,
+      "step": 6527
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.7638490625492366e-06,
+      "loss": 0.8041,
+      "step": 6528
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.7630888608791564e-06,
+      "loss": 0.7981,
+      "step": 6529
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.762328659209076e-06,
+      "loss": 0.9209,
+      "step": 6530
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.7615684575389944e-06,
+      "loss": 0.8402,
+      "step": 6531
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.760808255868914e-06,
+      "loss": 0.9206,
+      "step": 6532
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.760048054198834e-06,
+      "loss": 0.9241,
+      "step": 6533
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.759287852528754e-06,
+      "loss": 0.8405,
+      "step": 6534
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.758527650858674e-06,
+      "loss": 0.6829,
+      "step": 6535
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.757767449188593e-06,
+      "loss": 0.7111,
+      "step": 6536
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.757007247518513e-06,
+      "loss": 0.8201,
+      "step": 6537
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.756247045848433e-06,
+      "loss": 0.8277,
+      "step": 6538
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.7554868441783524e-06,
+      "loss": 1.0658,
+      "step": 6539
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.754726642508272e-06,
+      "loss": 0.7448,
+      "step": 6540
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.753966440838191e-06,
+      "loss": 1.026,
+      "step": 6541
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.753206239168111e-06,
+      "loss": 0.8716,
+      "step": 6542
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.752446037498031e-06,
+      "loss": 0.9767,
+      "step": 6543
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.751685835827951e-06,
+      "loss": 0.8621,
+      "step": 6544
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.750925634157871e-06,
+      "loss": 0.7391,
+      "step": 6545
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.750165432487791e-06,
+      "loss": 0.8301,
+      "step": 6546
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.74940523081771e-06,
+      "loss": 0.9363,
+      "step": 6547
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.7486450291476295e-06,
+      "loss": 0.7495,
+      "step": 6548
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.747884827477549e-06,
+      "loss": 0.9352,
+      "step": 6549
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.747124625807469e-06,
+      "loss": 1.0796,
+      "step": 6550
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.746364424137389e-06,
+      "loss": 0.8887,
+      "step": 6551
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.745604222467307e-06,
+      "loss": 0.8253,
+      "step": 6552
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.744844020797227e-06,
+      "loss": 0.9092,
+      "step": 6553
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.744083819127147e-06,
+      "loss": 0.6277,
+      "step": 6554
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.743323617457067e-06,
+      "loss": 0.8316,
+      "step": 6555
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.742563415786987e-06,
+      "loss": 0.8535,
+      "step": 6556
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.741803214116906e-06,
+      "loss": 0.8998,
+      "step": 6557
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.7410430124468255e-06,
+      "loss": 0.7067,
+      "step": 6558
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.740282810776745e-06,
+      "loss": 0.7353,
+      "step": 6559
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.739522609106665e-06,
+      "loss": 0.8047,
+      "step": 6560
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.738762407436585e-06,
+      "loss": 0.7472,
+      "step": 6561
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.738002205766505e-06,
+      "loss": 0.7729,
+      "step": 6562
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.737242004096424e-06,
+      "loss": 1.0433,
+      "step": 6563
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.736481802426344e-06,
+      "loss": 0.8819,
+      "step": 6564
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.735721600756264e-06,
+      "loss": 0.7535,
+      "step": 6565
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7349613990861835e-06,
+      "loss": 0.8001,
+      "step": 6566
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.734201197416103e-06,
+      "loss": 0.7608,
+      "step": 6567
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7334409957460215e-06,
+      "loss": 0.8472,
+      "step": 6568
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.732680794075941e-06,
+      "loss": 0.8489,
+      "step": 6569
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.731920592405861e-06,
+      "loss": 0.7327,
+      "step": 6570
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.731160390735781e-06,
+      "loss": 0.7989,
+      "step": 6571
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.730400189065701e-06,
+      "loss": 0.8325,
+      "step": 6572
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.72963998739562e-06,
+      "loss": 0.8747,
+      "step": 6573
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.72887978572554e-06,
+      "loss": 0.6027,
+      "step": 6574
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.72811958405546e-06,
+      "loss": 0.8884,
+      "step": 6575
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7273593823853795e-06,
+      "loss": 0.9074,
+      "step": 6576
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.726599180715299e-06,
+      "loss": 0.7258,
+      "step": 6577
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.725838979045219e-06,
+      "loss": 0.9546,
+      "step": 6578
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.725078777375138e-06,
+      "loss": 0.7922,
+      "step": 6579
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.724318575705058e-06,
+      "loss": 0.8147,
+      "step": 6580
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.723558374034978e-06,
+      "loss": 0.6772,
+      "step": 6581
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.722798172364898e-06,
+      "loss": 0.8231,
+      "step": 6582
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.722037970694818e-06,
+      "loss": 0.7732,
+      "step": 6583
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.721277769024736e-06,
+      "loss": 0.8379,
+      "step": 6584
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.720517567354656e-06,
+      "loss": 0.7951,
+      "step": 6585
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7197573656845755e-06,
+      "loss": 0.7157,
+      "step": 6586
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.718997164014495e-06,
+      "loss": 0.7773,
+      "step": 6587
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.718236962344415e-06,
+      "loss": 0.9906,
+      "step": 6588
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.717476760674334e-06,
+      "loss": 0.9609,
+      "step": 6589
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.716716559004254e-06,
+      "loss": 0.7608,
+      "step": 6590
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.715956357334174e-06,
+      "loss": 0.6665,
+      "step": 6591
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.715196155664094e-06,
+      "loss": 0.8482,
+      "step": 6592
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.714435953994014e-06,
+      "loss": 0.8364,
+      "step": 6593
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7136757523239335e-06,
+      "loss": 0.904,
+      "step": 6594
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7129155506538525e-06,
+      "loss": 0.835,
+      "step": 6595
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.712155348983772e-06,
+      "loss": 0.857,
+      "step": 6596
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.711395147313692e-06,
+      "loss": 0.9255,
+      "step": 6597
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.710634945643612e-06,
+      "loss": 0.9009,
+      "step": 6598
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.709874743973532e-06,
+      "loss": 0.8727,
+      "step": 6599
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.70911454230345e-06,
+      "loss": 0.6525,
+      "step": 6600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.70835434063337e-06,
+      "loss": 0.7692,
+      "step": 6601
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.70759413896329e-06,
+      "loss": 0.861,
+      "step": 6602
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.70683393729321e-06,
+      "loss": 0.6698,
+      "step": 6603
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7060737356231295e-06,
+      "loss": 0.9089,
+      "step": 6604
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7053135339530485e-06,
+      "loss": 1.1128,
+      "step": 6605
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.704553332282968e-06,
+      "loss": 0.9145,
+      "step": 6606
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.703793130612888e-06,
+      "loss": 0.8233,
+      "step": 6607
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.703032928942808e-06,
+      "loss": 0.8654,
+      "step": 6608
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.702272727272728e-06,
+      "loss": 0.7633,
+      "step": 6609
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.701512525602648e-06,
+      "loss": 0.7682,
+      "step": 6610
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.700752323932567e-06,
+      "loss": 1.0382,
+      "step": 6611
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.699992122262487e-06,
+      "loss": 0.8298,
+      "step": 6612
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.6992319205924065e-06,
+      "loss": 0.9485,
+      "step": 6613
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.698471718922326e-06,
+      "loss": 1.0823,
+      "step": 6614
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.697711517252246e-06,
+      "loss": 0.7649,
+      "step": 6615
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.696951315582164e-06,
+      "loss": 0.7202,
+      "step": 6616
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.696191113912084e-06,
+      "loss": 0.8822,
+      "step": 6617
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.695430912242004e-06,
+      "loss": 0.7673,
+      "step": 6618
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.694670710571924e-06,
+      "loss": 0.777,
+      "step": 6619
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.693910508901844e-06,
+      "loss": 0.6414,
+      "step": 6620
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.693150307231763e-06,
+      "loss": 0.8403,
+      "step": 6621
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.692390105561683e-06,
+      "loss": 0.7319,
+      "step": 6622
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.6916299038916026e-06,
+      "loss": 0.7226,
+      "step": 6623
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.690869702221522e-06,
+      "loss": 0.8009,
+      "step": 6624
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.690109500551442e-06,
+      "loss": 0.9345,
+      "step": 6625
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.689349298881362e-06,
+      "loss": 1.1241,
+      "step": 6626
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.688589097211281e-06,
+      "loss": 0.8116,
+      "step": 6627
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.687828895541201e-06,
+      "loss": 0.7568,
+      "step": 6628
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.687068693871121e-06,
+      "loss": 0.7228,
+      "step": 6629
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.686308492201041e-06,
+      "loss": 0.6991,
+      "step": 6630
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.6855482905309606e-06,
+      "loss": 0.8375,
+      "step": 6631
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.6847880888608796e-06,
+      "loss": 0.7365,
+      "step": 6632
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.6840278871907994e-06,
+      "loss": 0.8967,
+      "step": 6633
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.683267685520719e-06,
+      "loss": 0.8983,
+      "step": 6634
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.682507483850639e-06,
+      "loss": 0.8538,
+      "step": 6635
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.681747282180558e-06,
+      "loss": 0.9304,
+      "step": 6636
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.680987080510477e-06,
+      "loss": 0.8116,
+      "step": 6637
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.680226878840397e-06,
+      "loss": 0.8526,
+      "step": 6638
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.679466677170317e-06,
+      "loss": 0.8514,
+      "step": 6639
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.678706475500237e-06,
+      "loss": 0.7261,
+      "step": 6640
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.677946273830157e-06,
+      "loss": 0.9829,
+      "step": 6641
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.6771860721600764e-06,
+      "loss": 0.8283,
+      "step": 6642
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.6764258704899954e-06,
+      "loss": 0.871,
+      "step": 6643
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.675665668819915e-06,
+      "loss": 0.8798,
+      "step": 6644
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.674905467149835e-06,
+      "loss": 0.8334,
+      "step": 6645
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.674145265479755e-06,
+      "loss": 0.6969,
+      "step": 6646
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.673385063809675e-06,
+      "loss": 0.6466,
+      "step": 6647
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.672624862139594e-06,
+      "loss": 0.9232,
+      "step": 6648
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.671864660469514e-06,
+      "loss": 0.8461,
+      "step": 6649
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.671104458799434e-06,
+      "loss": 0.8152,
+      "step": 6650
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.6703442571293534e-06,
+      "loss": 0.8192,
+      "step": 6651
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.669584055459273e-06,
+      "loss": 0.9046,
+      "step": 6652
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.6688238537891915e-06,
+      "loss": 1.1255,
+      "step": 6653
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.668063652119111e-06,
+      "loss": 0.6535,
+      "step": 6654
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.667303450449031e-06,
+      "loss": 0.8821,
+      "step": 6655
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.666543248778951e-06,
+      "loss": 0.9078,
+      "step": 6656
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.665783047108871e-06,
+      "loss": 0.9339,
+      "step": 6657
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.665022845438791e-06,
+      "loss": 0.8433,
+      "step": 6658
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.66426264376871e-06,
+      "loss": 0.8543,
+      "step": 6659
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.66350244209863e-06,
+      "loss": 0.988,
+      "step": 6660
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.6627422404285495e-06,
+      "loss": 0.7637,
+      "step": 6661
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.661982038758469e-06,
+      "loss": 0.8336,
+      "step": 6662
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.661221837088389e-06,
+      "loss": 1.2635,
+      "step": 6663
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.660461635418308e-06,
+      "loss": 0.9382,
+      "step": 6664
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.659701433748228e-06,
+      "loss": 0.9161,
+      "step": 6665
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.658941232078148e-06,
+      "loss": 0.8709,
+      "step": 6666
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.658181030408068e-06,
+      "loss": 0.8152,
+      "step": 6667
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.657420828737988e-06,
+      "loss": 0.7191,
+      "step": 6668
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.656660627067906e-06,
+      "loss": 0.8742,
+      "step": 6669
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.655900425397826e-06,
+      "loss": 0.8541,
+      "step": 6670
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.6551402237277455e-06,
+      "loss": 0.9604,
+      "step": 6671
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.654380022057665e-06,
+      "loss": 0.7947,
+      "step": 6672
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.653619820387585e-06,
+      "loss": 0.8008,
+      "step": 6673
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.652859618717505e-06,
+      "loss": 0.7914,
+      "step": 6674
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.652099417047424e-06,
+      "loss": 0.7163,
+      "step": 6675
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.651339215377344e-06,
+      "loss": 0.984,
+      "step": 6676
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.650579013707264e-06,
+      "loss": 0.9425,
+      "step": 6677
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.649818812037184e-06,
+      "loss": 0.8776,
+      "step": 6678
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.6490586103671035e-06,
+      "loss": 0.9252,
+      "step": 6679
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.6482984086970225e-06,
+      "loss": 0.7485,
+      "step": 6680
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.647538207026942e-06,
+      "loss": 0.9233,
+      "step": 6681
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.646778005356862e-06,
+      "loss": 0.9151,
+      "step": 6682
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.646017803686782e-06,
+      "loss": 0.8664,
+      "step": 6683
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.645257602016702e-06,
+      "loss": 0.9875,
+      "step": 6684
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.64449740034662e-06,
+      "loss": 0.8508,
+      "step": 6685
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.64373719867654e-06,
+      "loss": 0.868,
+      "step": 6686
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.64297699700646e-06,
+      "loss": 0.8464,
+      "step": 6687
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.64221679533638e-06,
+      "loss": 0.8438,
+      "step": 6688
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.6414565936662995e-06,
+      "loss": 0.8346,
+      "step": 6689
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.640696391996219e-06,
+      "loss": 0.9288,
+      "step": 6690
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.639936190326138e-06,
+      "loss": 0.9014,
+      "step": 6691
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.639175988656058e-06,
+      "loss": 0.9446,
+      "step": 6692
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.638415786985978e-06,
+      "loss": 0.5694,
+      "step": 6693
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.637655585315898e-06,
+      "loss": 0.8122,
+      "step": 6694
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.636895383645818e-06,
+      "loss": 0.8093,
+      "step": 6695
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.636135181975737e-06,
+      "loss": 0.9181,
+      "step": 6696
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.635374980305657e-06,
+      "loss": 0.7592,
+      "step": 6697
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.6346147786355765e-06,
+      "loss": 1.1311,
+      "step": 6698
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.633854576965496e-06,
+      "loss": 0.8604,
+      "step": 6699
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.633094375295416e-06,
+      "loss": 0.8785,
+      "step": 6700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.632334173625334e-06,
+      "loss": 0.8281,
+      "step": 6701
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.631573971955254e-06,
+      "loss": 0.8316,
+      "step": 6702
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.630813770285174e-06,
+      "loss": 0.7576,
+      "step": 6703
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.630053568615094e-06,
+      "loss": 0.87,
+      "step": 6704
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.629293366945014e-06,
+      "loss": 0.9033,
+      "step": 6705
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.628533165274934e-06,
+      "loss": 0.7836,
+      "step": 6706
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.627772963604853e-06,
+      "loss": 0.6729,
+      "step": 6707
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.6270127619347725e-06,
+      "loss": 0.7521,
+      "step": 6708
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.626252560264692e-06,
+      "loss": 0.8408,
+      "step": 6709
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.625492358594612e-06,
+      "loss": 0.7818,
+      "step": 6710
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.624732156924532e-06,
+      "loss": 1.0356,
+      "step": 6711
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.623971955254451e-06,
+      "loss": 0.6954,
+      "step": 6712
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.623211753584371e-06,
+      "loss": 0.9708,
+      "step": 6713
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.622451551914291e-06,
+      "loss": 0.9228,
+      "step": 6714
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.621691350244211e-06,
+      "loss": 0.8653,
+      "step": 6715
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.6209311485741305e-06,
+      "loss": 0.7703,
+      "step": 6716
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.6201709469040495e-06,
+      "loss": 0.8798,
+      "step": 6717
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.619410745233969e-06,
+      "loss": 0.8253,
+      "step": 6718
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.618650543563889e-06,
+      "loss": 0.6481,
+      "step": 6719
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.617890341893808e-06,
+      "loss": 0.7778,
+      "step": 6720
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.617130140223728e-06,
+      "loss": 0.936,
+      "step": 6721
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.616369938553648e-06,
+      "loss": 0.9915,
+      "step": 6722
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.615609736883567e-06,
+      "loss": 0.7821,
+      "step": 6723
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.614849535213487e-06,
+      "loss": 0.9212,
+      "step": 6724
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.614089333543407e-06,
+      "loss": 0.8913,
+      "step": 6725
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.6133291318733266e-06,
+      "loss": 0.8387,
+      "step": 6726
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.612568930203246e-06,
+      "loss": 0.8999,
+      "step": 6727
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.611808728533165e-06,
+      "loss": 0.8123,
+      "step": 6728
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.611048526863085e-06,
+      "loss": 0.9396,
+      "step": 6729
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.610288325193005e-06,
+      "loss": 0.8736,
+      "step": 6730
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.609528123522925e-06,
+      "loss": 0.7627,
+      "step": 6731
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.608767921852845e-06,
+      "loss": 0.877,
+      "step": 6732
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.608007720182764e-06,
+      "loss": 0.9027,
+      "step": 6733
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.607247518512684e-06,
+      "loss": 0.867,
+      "step": 6734
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.6064873168426036e-06,
+      "loss": 0.5964,
+      "step": 6735
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.605727115172523e-06,
+      "loss": 0.7832,
+      "step": 6736
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.604966913502443e-06,
+      "loss": 0.8113,
+      "step": 6737
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.604206711832363e-06,
+      "loss": 0.7422,
+      "step": 6738
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.603446510162281e-06,
+      "loss": 0.947,
+      "step": 6739
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.602686308492201e-06,
+      "loss": 0.7249,
+      "step": 6740
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.601926106822121e-06,
+      "loss": 0.7109,
+      "step": 6741
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.601165905152041e-06,
+      "loss": 0.8463,
+      "step": 6742
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.600405703481961e-06,
+      "loss": 0.9966,
+      "step": 6743
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.59964550181188e-06,
+      "loss": 0.7066,
+      "step": 6744
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.5988853001418e-06,
+      "loss": 0.723,
+      "step": 6745
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.5981250984717194e-06,
+      "loss": 0.8399,
+      "step": 6746
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.597364896801639e-06,
+      "loss": 0.732,
+      "step": 6747
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.596604695131559e-06,
+      "loss": 0.8455,
+      "step": 6748
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.595844493461478e-06,
+      "loss": 0.9487,
+      "step": 6749
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.595084291791398e-06,
+      "loss": 0.7394,
+      "step": 6750
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.594324090121318e-06,
+      "loss": 0.8943,
+      "step": 6751
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.593563888451238e-06,
+      "loss": 1.1152,
+      "step": 6752
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.592803686781158e-06,
+      "loss": 0.9061,
+      "step": 6753
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.592043485111076e-06,
+      "loss": 0.9278,
+      "step": 6754
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.591283283440996e-06,
+      "loss": 0.9666,
+      "step": 6755
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.5905230817709155e-06,
+      "loss": 0.7185,
+      "step": 6756
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.589762880100835e-06,
+      "loss": 0.7837,
+      "step": 6757
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.589002678430755e-06,
+      "loss": 1.0493,
+      "step": 6758
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.588242476760675e-06,
+      "loss": 0.966,
+      "step": 6759
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.587482275090594e-06,
+      "loss": 0.7125,
+      "step": 6760
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.586722073420514e-06,
+      "loss": 0.9978,
+      "step": 6761
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.585961871750434e-06,
+      "loss": 0.8659,
+      "step": 6762
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.585201670080354e-06,
+      "loss": 0.9629,
+      "step": 6763
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.5844414684102735e-06,
+      "loss": 0.8522,
+      "step": 6764
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.5836812667401925e-06,
+      "loss": 0.9664,
+      "step": 6765
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.582921065070112e-06,
+      "loss": 0.8056,
+      "step": 6766
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.582160863400032e-06,
+      "loss": 0.7566,
+      "step": 6767
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.581400661729952e-06,
+      "loss": 0.9094,
+      "step": 6768
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.580640460059872e-06,
+      "loss": 1.029,
+      "step": 6769
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.57988025838979e-06,
+      "loss": 0.8473,
+      "step": 6770
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.57912005671971e-06,
+      "loss": 0.798,
+      "step": 6771
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.57835985504963e-06,
+      "loss": 0.8002,
+      "step": 6772
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.57759965337955e-06,
+      "loss": 0.8172,
+      "step": 6773
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.5768394517094695e-06,
+      "loss": 0.9768,
+      "step": 6774
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.576079250039389e-06,
+      "loss": 0.9311,
+      "step": 6775
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.575319048369308e-06,
+      "loss": 0.7877,
+      "step": 6776
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.574558846699228e-06,
+      "loss": 0.9831,
+      "step": 6777
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.573798645029148e-06,
+      "loss": 0.805,
+      "step": 6778
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.573038443359068e-06,
+      "loss": 0.8602,
+      "step": 6779
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.572278241688988e-06,
+      "loss": 1.1149,
+      "step": 6780
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.571518040018907e-06,
+      "loss": 0.7825,
+      "step": 6781
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.570757838348827e-06,
+      "loss": 1.0448,
+      "step": 6782
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.5699976366787465e-06,
+      "loss": 0.9648,
+      "step": 6783
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.569237435008666e-06,
+      "loss": 0.8743,
+      "step": 6784
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.568477233338586e-06,
+      "loss": 0.6778,
+      "step": 6785
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.567717031668504e-06,
+      "loss": 0.7758,
+      "step": 6786
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.566956829998424e-06,
+      "loss": 0.7855,
+      "step": 6787
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.566196628328344e-06,
+      "loss": 0.8255,
+      "step": 6788
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.565436426658264e-06,
+      "loss": 0.6938,
+      "step": 6789
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.564676224988184e-06,
+      "loss": 0.9758,
+      "step": 6790
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.563916023318104e-06,
+      "loss": 0.7924,
+      "step": 6791
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.563155821648023e-06,
+      "loss": 0.8765,
+      "step": 6792
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.5623956199779425e-06,
+      "loss": 0.8992,
+      "step": 6793
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.561635418307862e-06,
+      "loss": 0.8729,
+      "step": 6794
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.560875216637782e-06,
+      "loss": 0.9257,
+      "step": 6795
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.560115014967702e-06,
+      "loss": 0.8144,
+      "step": 6796
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.559354813297621e-06,
+      "loss": 0.7059,
+      "step": 6797
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.558594611627541e-06,
+      "loss": 0.8765,
+      "step": 6798
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.557834409957461e-06,
+      "loss": 0.6788,
+      "step": 6799
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.557074208287381e-06,
+      "loss": 0.9731,
+      "step": 6800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.5563140066173005e-06,
+      "loss": 1.0038,
+      "step": 6801
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.5555538049472195e-06,
+      "loss": 0.6843,
+      "step": 6802
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.5547936032771385e-06,
+      "loss": 0.9467,
+      "step": 6803
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.554033401607058e-06,
+      "loss": 0.7904,
+      "step": 6804
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.553273199936978e-06,
+      "loss": 0.9859,
+      "step": 6805
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.552512998266898e-06,
+      "loss": 0.8897,
+      "step": 6806
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.551752796596818e-06,
+      "loss": 0.8548,
+      "step": 6807
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.550992594926737e-06,
+      "loss": 0.7078,
+      "step": 6808
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.550232393256657e-06,
+      "loss": 0.8715,
+      "step": 6809
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.549472191586577e-06,
+      "loss": 0.7596,
+      "step": 6810
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.5487119899164965e-06,
+      "loss": 0.7008,
+      "step": 6811
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.547951788246416e-06,
+      "loss": 0.8,
+      "step": 6812
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.547191586576335e-06,
+      "loss": 0.876,
+      "step": 6813
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.546431384906255e-06,
+      "loss": 0.9533,
+      "step": 6814
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.545671183236175e-06,
+      "loss": 0.6857,
+      "step": 6815
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.544910981566095e-06,
+      "loss": 0.7907,
+      "step": 6816
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.544150779896015e-06,
+      "loss": 0.812,
+      "step": 6817
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.543390578225934e-06,
+      "loss": 1.0737,
+      "step": 6818
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.542630376555854e-06,
+      "loss": 0.8183,
+      "step": 6819
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.5418701748857735e-06,
+      "loss": 1.0004,
+      "step": 6820
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.541109973215693e-06,
+      "loss": 0.7752,
+      "step": 6821
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.540349771545613e-06,
+      "loss": 0.8444,
+      "step": 6822
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.539589569875533e-06,
+      "loss": 0.81,
+      "step": 6823
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.538829368205451e-06,
+      "loss": 0.9633,
+      "step": 6824
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.538069166535371e-06,
+      "loss": 0.9847,
+      "step": 6825
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.537308964865291e-06,
+      "loss": 0.7578,
+      "step": 6826
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.536548763195211e-06,
+      "loss": 0.6706,
+      "step": 6827
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.535788561525131e-06,
+      "loss": 0.8453,
+      "step": 6828
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.53502835985505e-06,
+      "loss": 0.868,
+      "step": 6829
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.5342681581849696e-06,
+      "loss": 0.6612,
+      "step": 6830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.533507956514889e-06,
+      "loss": 0.8638,
+      "step": 6831
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.532747754844809e-06,
+      "loss": 1.0664,
+      "step": 6832
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.531987553174729e-06,
+      "loss": 0.7869,
+      "step": 6833
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.531227351504648e-06,
+      "loss": 0.8657,
+      "step": 6834
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.530467149834568e-06,
+      "loss": 0.9565,
+      "step": 6835
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.529706948164488e-06,
+      "loss": 0.6687,
+      "step": 6836
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.528946746494408e-06,
+      "loss": 0.7277,
+      "step": 6837
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.5281865448243276e-06,
+      "loss": 0.8954,
+      "step": 6838
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.527426343154247e-06,
+      "loss": 0.8767,
+      "step": 6839
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.5266661414841656e-06,
+      "loss": 0.8563,
+      "step": 6840
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.5259059398140854e-06,
+      "loss": 0.9181,
+      "step": 6841
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.525145738144005e-06,
+      "loss": 0.8303,
+      "step": 6842
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.524385536473925e-06,
+      "loss": 0.8262,
+      "step": 6843
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.523625334803845e-06,
+      "loss": 0.6977,
+      "step": 6844
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.522865133133764e-06,
+      "loss": 0.8389,
+      "step": 6845
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.522104931463684e-06,
+      "loss": 0.9015,
+      "step": 6846
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.521344729793604e-06,
+      "loss": 0.6892,
+      "step": 6847
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.5205845281235236e-06,
+      "loss": 0.9093,
+      "step": 6848
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.5198243264534434e-06,
+      "loss": 0.7671,
+      "step": 6849
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.5190641247833624e-06,
+      "loss": 0.8796,
+      "step": 6850
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.518303923113282e-06,
+      "loss": 0.785,
+      "step": 6851
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.517543721443202e-06,
+      "loss": 0.7424,
+      "step": 6852
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.516783519773122e-06,
+      "loss": 0.7611,
+      "step": 6853
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.516023318103042e-06,
+      "loss": 1.0868,
+      "step": 6854
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.515263116432962e-06,
+      "loss": 0.8159,
+      "step": 6855
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.51450291476288e-06,
+      "loss": 0.9676,
+      "step": 6856
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.5137427130928e-06,
+      "loss": 0.8138,
+      "step": 6857
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.51298251142272e-06,
+      "loss": 0.8589,
+      "step": 6858
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.5122223097526395e-06,
+      "loss": 0.9878,
+      "step": 6859
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.511462108082559e-06,
+      "loss": 0.6569,
+      "step": 6860
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.510701906412478e-06,
+      "loss": 0.7232,
+      "step": 6861
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.509941704742398e-06,
+      "loss": 0.7773,
+      "step": 6862
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.509181503072318e-06,
+      "loss": 0.7126,
+      "step": 6863
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.508421301402238e-06,
+      "loss": 0.7802,
+      "step": 6864
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.507661099732158e-06,
+      "loss": 0.8133,
+      "step": 6865
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.506900898062077e-06,
+      "loss": 0.9139,
+      "step": 6866
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.506140696391997e-06,
+      "loss": 0.8166,
+      "step": 6867
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.5053804947219165e-06,
+      "loss": 0.7973,
+      "step": 6868
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.504620293051836e-06,
+      "loss": 0.8732,
+      "step": 6869
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.503860091381756e-06,
+      "loss": 0.9702,
+      "step": 6870
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.503099889711676e-06,
+      "loss": 0.8459,
+      "step": 6871
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.502339688041594e-06,
+      "loss": 0.9094,
+      "step": 6872
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.501579486371514e-06,
+      "loss": 0.8034,
+      "step": 6873
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.500819284701434e-06,
+      "loss": 0.8651,
+      "step": 6874
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.500059083031354e-06,
+      "loss": 0.6869,
+      "step": 6875
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.499298881361274e-06,
+      "loss": 0.9155,
+      "step": 6876
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.498538679691193e-06,
+      "loss": 0.9849,
+      "step": 6877
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.4977784780211125e-06,
+      "loss": 0.6223,
+      "step": 6878
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.497018276351032e-06,
+      "loss": 0.6664,
+      "step": 6879
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.496258074680952e-06,
+      "loss": 1.046,
+      "step": 6880
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.495497873010872e-06,
+      "loss": 0.8956,
+      "step": 6881
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.494737671340791e-06,
+      "loss": 0.8889,
+      "step": 6882
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.493977469670711e-06,
+      "loss": 0.8116,
+      "step": 6883
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.493217268000631e-06,
+      "loss": 0.8956,
+      "step": 6884
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.492457066330551e-06,
+      "loss": 0.7473,
+      "step": 6885
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.4916968646604705e-06,
+      "loss": 0.8648,
+      "step": 6886
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.49093666299039e-06,
+      "loss": 0.8101,
+      "step": 6887
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.4901764613203085e-06,
+      "loss": 0.8152,
+      "step": 6888
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.489416259650228e-06,
+      "loss": 0.7996,
+      "step": 6889
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.488656057980148e-06,
+      "loss": 0.8758,
+      "step": 6890
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.487895856310068e-06,
+      "loss": 0.796,
+      "step": 6891
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.487135654639988e-06,
+      "loss": 0.8608,
+      "step": 6892
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.486375452969907e-06,
+      "loss": 0.8877,
+      "step": 6893
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.485615251299827e-06,
+      "loss": 0.6463,
+      "step": 6894
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.484855049629747e-06,
+      "loss": 0.9322,
+      "step": 6895
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.4840948479596665e-06,
+      "loss": 0.9068,
+      "step": 6896
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.483334646289586e-06,
+      "loss": 0.9733,
+      "step": 6897
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.482574444619505e-06,
+      "loss": 0.804,
+      "step": 6898
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.481814242949425e-06,
+      "loss": 0.9387,
+      "step": 6899
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.481054041279345e-06,
+      "loss": 0.7303,
+      "step": 6900
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.480293839609265e-06,
+      "loss": 0.8751,
+      "step": 6901
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.479533637939185e-06,
+      "loss": 0.9614,
+      "step": 6902
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.478773436269105e-06,
+      "loss": 0.8917,
+      "step": 6903
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.478013234599024e-06,
+      "loss": 0.6956,
+      "step": 6904
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.4772530329289435e-06,
+      "loss": 0.8356,
+      "step": 6905
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.476492831258863e-06,
+      "loss": 0.8492,
+      "step": 6906
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.475732629588783e-06,
+      "loss": 1.0291,
+      "step": 6907
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.474972427918702e-06,
+      "loss": 0.8678,
+      "step": 6908
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.474212226248621e-06,
+      "loss": 0.8185,
+      "step": 6909
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.473452024578541e-06,
+      "loss": 0.9639,
+      "step": 6910
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.472691822908461e-06,
+      "loss": 0.8699,
+      "step": 6911
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.471931621238381e-06,
+      "loss": 0.8487,
+      "step": 6912
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.471171419568301e-06,
+      "loss": 0.6612,
+      "step": 6913
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.47041121789822e-06,
+      "loss": 0.8812,
+      "step": 6914
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.4696510162281395e-06,
+      "loss": 0.7535,
+      "step": 6915
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.468890814558059e-06,
+      "loss": 0.827,
+      "step": 6916
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.468130612887979e-06,
+      "loss": 0.8506,
+      "step": 6917
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.467370411217899e-06,
+      "loss": 0.741,
+      "step": 6918
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.466610209547819e-06,
+      "loss": 0.7622,
+      "step": 6919
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.465850007877738e-06,
+      "loss": 0.887,
+      "step": 6920
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.465089806207658e-06,
+      "loss": 0.7153,
+      "step": 6921
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.464329604537578e-06,
+      "loss": 0.723,
+      "step": 6922
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.4635694028674975e-06,
+      "loss": 0.7474,
+      "step": 6923
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.462809201197417e-06,
+      "loss": 0.7547,
+      "step": 6924
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.4620489995273355e-06,
+      "loss": 1.029,
+      "step": 6925
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.461288797857255e-06,
+      "loss": 0.8401,
+      "step": 6926
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.460528596187175e-06,
+      "loss": 0.9656,
+      "step": 6927
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.459768394517095e-06,
+      "loss": 0.8112,
+      "step": 6928
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.459008192847015e-06,
+      "loss": 0.7279,
+      "step": 6929
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.458247991176934e-06,
+      "loss": 0.8853,
+      "step": 6930
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.457487789506854e-06,
+      "loss": 0.9624,
+      "step": 6931
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.456727587836774e-06,
+      "loss": 0.9158,
+      "step": 6932
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.4559673861666935e-06,
+      "loss": 0.839,
+      "step": 6933
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.455207184496613e-06,
+      "loss": 0.9586,
+      "step": 6934
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.454446982826533e-06,
+      "loss": 0.904,
+      "step": 6935
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.453686781156452e-06,
+      "loss": 0.8914,
+      "step": 6936
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.452926579486372e-06,
+      "loss": 0.9606,
+      "step": 6937
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.452166377816292e-06,
+      "loss": 0.9339,
+      "step": 6938
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.451406176146212e-06,
+      "loss": 1.0643,
+      "step": 6939
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.450645974476132e-06,
+      "loss": 0.8428,
+      "step": 6940
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.44988577280605e-06,
+      "loss": 0.8571,
+      "step": 6941
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.44912557113597e-06,
+      "loss": 0.8725,
+      "step": 6942
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.4483653694658896e-06,
+      "loss": 0.836,
+      "step": 6943
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.447605167795809e-06,
+      "loss": 0.9356,
+      "step": 6944
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.446844966125729e-06,
+      "loss": 0.772,
+      "step": 6945
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.446084764455648e-06,
+      "loss": 0.7968,
+      "step": 6946
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.445324562785568e-06,
+      "loss": 0.7454,
+      "step": 6947
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.444564361115488e-06,
+      "loss": 1.0105,
+      "step": 6948
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.443804159445408e-06,
+      "loss": 0.866,
+      "step": 6949
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.443043957775328e-06,
+      "loss": 0.8085,
+      "step": 6950
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.442283756105247e-06,
+      "loss": 1.0405,
+      "step": 6951
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.441523554435167e-06,
+      "loss": 0.8428,
+      "step": 6952
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.4407633527650864e-06,
+      "loss": 0.8223,
+      "step": 6953
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.440003151095006e-06,
+      "loss": 0.9619,
+      "step": 6954
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.439242949424926e-06,
+      "loss": 0.9152,
+      "step": 6955
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.438482747754846e-06,
+      "loss": 0.9363,
+      "step": 6956
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.437722546084764e-06,
+      "loss": 0.8383,
+      "step": 6957
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.436962344414684e-06,
+      "loss": 0.827,
+      "step": 6958
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.436202142744604e-06,
+      "loss": 0.7837,
+      "step": 6959
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.435441941074524e-06,
+      "loss": 0.7531,
+      "step": 6960
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.434681739404444e-06,
+      "loss": 0.7946,
+      "step": 6961
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.433921537734363e-06,
+      "loss": 0.7511,
+      "step": 6962
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.4331613360642825e-06,
+      "loss": 0.7921,
+      "step": 6963
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.432401134394202e-06,
+      "loss": 0.6802,
+      "step": 6964
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.431640932724122e-06,
+      "loss": 0.976,
+      "step": 6965
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.430880731054042e-06,
+      "loss": 0.8062,
+      "step": 6966
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.430120529383961e-06,
+      "loss": 0.8622,
+      "step": 6967
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.429360327713881e-06,
+      "loss": 0.9237,
+      "step": 6968
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.428600126043801e-06,
+      "loss": 0.8217,
+      "step": 6969
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.427839924373721e-06,
+      "loss": 0.8521,
+      "step": 6970
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.4270797227036405e-06,
+      "loss": 0.9068,
+      "step": 6971
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.42631952103356e-06,
+      "loss": 0.7642,
+      "step": 6972
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.4255593193634785e-06,
+      "loss": 0.771,
+      "step": 6973
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.424799117693398e-06,
+      "loss": 0.8286,
+      "step": 6974
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.424038916023318e-06,
+      "loss": 1.1071,
+      "step": 6975
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.423278714353238e-06,
+      "loss": 0.7969,
+      "step": 6976
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.422518512683158e-06,
+      "loss": 0.82,
+      "step": 6977
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.421758311013077e-06,
+      "loss": 0.8502,
+      "step": 6978
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.420998109342997e-06,
+      "loss": 0.8947,
+      "step": 6979
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.420237907672917e-06,
+      "loss": 0.8842,
+      "step": 6980
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.4194777060028365e-06,
+      "loss": 0.8397,
+      "step": 6981
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.418717504332756e-06,
+      "loss": 0.7016,
+      "step": 6982
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.417957302662675e-06,
+      "loss": 0.7688,
+      "step": 6983
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.417197100992595e-06,
+      "loss": 1.0437,
+      "step": 6984
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.416436899322515e-06,
+      "loss": 0.8702,
+      "step": 6985
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.415676697652435e-06,
+      "loss": 0.7709,
+      "step": 6986
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.414916495982355e-06,
+      "loss": 0.8137,
+      "step": 6987
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.414156294312275e-06,
+      "loss": 0.9387,
+      "step": 6988
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.413396092642194e-06,
+      "loss": 1.0113,
+      "step": 6989
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.4126358909721135e-06,
+      "loss": 0.8962,
+      "step": 6990
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.4118756893020325e-06,
+      "loss": 0.8927,
+      "step": 6991
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.411115487631952e-06,
+      "loss": 1.1037,
+      "step": 6992
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.410355285961872e-06,
+      "loss": 0.8376,
+      "step": 6993
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.409595084291791e-06,
+      "loss": 1.0009,
+      "step": 6994
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.408834882621711e-06,
+      "loss": 0.778,
+      "step": 6995
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.408074680951631e-06,
+      "loss": 0.943,
+      "step": 6996
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.407314479281551e-06,
+      "loss": 0.9277,
+      "step": 6997
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.406554277611471e-06,
+      "loss": 0.8852,
+      "step": 6998
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.40579407594139e-06,
+      "loss": 0.8928,
+      "step": 6999
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.4050338742713095e-06,
+      "loss": 0.8216,
+      "step": 7000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.404273672601229e-06,
+      "loss": 0.9329,
+      "step": 7001
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.403513470931149e-06,
+      "loss": 0.7477,
+      "step": 7002
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.402753269261069e-06,
+      "loss": 0.6453,
+      "step": 7003
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.401993067590989e-06,
+      "loss": 0.6738,
+      "step": 7004
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.401232865920908e-06,
+      "loss": 0.8008,
+      "step": 7005
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.400472664250828e-06,
+      "loss": 0.8261,
+      "step": 7006
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.399712462580748e-06,
+      "loss": 0.8505,
+      "step": 7007
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.3989522609106675e-06,
+      "loss": 0.8766,
+      "step": 7008
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.398192059240587e-06,
+      "loss": 0.7504,
+      "step": 7009
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.3974318575705055e-06,
+      "loss": 0.8172,
+      "step": 7010
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.396671655900425e-06,
+      "loss": 0.9755,
+      "step": 7011
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.395911454230345e-06,
+      "loss": 0.7579,
+      "step": 7012
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.395151252560265e-06,
+      "loss": 0.7761,
+      "step": 7013
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.394391050890185e-06,
+      "loss": 0.8632,
+      "step": 7014
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.393630849220104e-06,
+      "loss": 0.8803,
+      "step": 7015
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.392870647550024e-06,
+      "loss": 0.7676,
+      "step": 7016
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.392110445879944e-06,
+      "loss": 0.9251,
+      "step": 7017
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.3913502442098635e-06,
+      "loss": 1.0018,
+      "step": 7018
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.390590042539783e-06,
+      "loss": 0.8337,
+      "step": 7019
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.389829840869703e-06,
+      "loss": 0.8443,
+      "step": 7020
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.389069639199622e-06,
+      "loss": 0.6912,
+      "step": 7021
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.388309437529542e-06,
+      "loss": 0.8123,
+      "step": 7022
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.387549235859462e-06,
+      "loss": 0.9631,
+      "step": 7023
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.386789034189382e-06,
+      "loss": 0.8883,
+      "step": 7024
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.386028832519302e-06,
+      "loss": 0.8695,
+      "step": 7025
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.38526863084922e-06,
+      "loss": 0.9133,
+      "step": 7026
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.38450842917914e-06,
+      "loss": 0.9453,
+      "step": 7027
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.3837482275090595e-06,
+      "loss": 0.8829,
+      "step": 7028
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.382988025838979e-06,
+      "loss": 0.8627,
+      "step": 7029
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.382227824168899e-06,
+      "loss": 0.7652,
+      "step": 7030
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.381467622498818e-06,
+      "loss": 0.8639,
+      "step": 7031
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.380707420828738e-06,
+      "loss": 0.6887,
+      "step": 7032
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.379947219158658e-06,
+      "loss": 0.7034,
+      "step": 7033
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.379187017488578e-06,
+      "loss": 0.8011,
+      "step": 7034
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.378426815818498e-06,
+      "loss": 0.9373,
+      "step": 7035
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.3776666141484175e-06,
+      "loss": 0.9099,
+      "step": 7036
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.3769064124783365e-06,
+      "loss": 0.902,
+      "step": 7037
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.376146210808256e-06,
+      "loss": 0.8194,
+      "step": 7038
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.375386009138176e-06,
+      "loss": 0.8412,
+      "step": 7039
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.374625807468096e-06,
+      "loss": 0.8593,
+      "step": 7040
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.373865605798016e-06,
+      "loss": 0.8642,
+      "step": 7041
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.373105404127934e-06,
+      "loss": 0.8643,
+      "step": 7042
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.372345202457854e-06,
+      "loss": 0.8651,
+      "step": 7043
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.371585000787774e-06,
+      "loss": 0.8583,
+      "step": 7044
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.370824799117694e-06,
+      "loss": 0.8332,
+      "step": 7045
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.3700645974476136e-06,
+      "loss": 0.9632,
+      "step": 7046
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.3693043957775326e-06,
+      "loss": 0.8834,
+      "step": 7047
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.3685441941074524e-06,
+      "loss": 0.908,
+      "step": 7048
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.367783992437372e-06,
+      "loss": 0.9137,
+      "step": 7049
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.367023790767292e-06,
+      "loss": 0.8765,
+      "step": 7050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.366263589097212e-06,
+      "loss": 0.8607,
+      "step": 7051
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.365503387427132e-06,
+      "loss": 0.6443,
+      "step": 7052
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.364743185757051e-06,
+      "loss": 0.7851,
+      "step": 7053
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.363982984086971e-06,
+      "loss": 0.9401,
+      "step": 7054
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.3632227824168906e-06,
+      "loss": 0.7581,
+      "step": 7055
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.3624625807468104e-06,
+      "loss": 0.9934,
+      "step": 7056
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.36170237907673e-06,
+      "loss": 0.7285,
+      "step": 7057
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.3609421774066484e-06,
+      "loss": 0.761,
+      "step": 7058
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.360181975736568e-06,
+      "loss": 0.9321,
+      "step": 7059
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.359421774066488e-06,
+      "loss": 0.9089,
+      "step": 7060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.358661572396408e-06,
+      "loss": 0.6578,
+      "step": 7061
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.357901370726328e-06,
+      "loss": 0.9309,
+      "step": 7062
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.357141169056247e-06,
+      "loss": 0.7831,
+      "step": 7063
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.356380967386167e-06,
+      "loss": 0.9166,
+      "step": 7064
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.355620765716087e-06,
+      "loss": 0.9534,
+      "step": 7065
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.3548605640460064e-06,
+      "loss": 0.7938,
+      "step": 7066
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.354100362375926e-06,
+      "loss": 0.8086,
+      "step": 7067
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.353340160705846e-06,
+      "loss": 0.6292,
+      "step": 7068
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.352579959035765e-06,
+      "loss": 0.7547,
+      "step": 7069
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.351819757365685e-06,
+      "loss": 0.8142,
+      "step": 7070
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.351059555695605e-06,
+      "loss": 0.8518,
+      "step": 7071
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.350299354025525e-06,
+      "loss": 0.9161,
+      "step": 7072
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.349539152355445e-06,
+      "loss": 0.9932,
+      "step": 7073
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.348778950685363e-06,
+      "loss": 0.7571,
+      "step": 7074
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.348018749015283e-06,
+      "loss": 0.9153,
+      "step": 7075
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.3472585473452025e-06,
+      "loss": 0.8111,
+      "step": 7076
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.346498345675122e-06,
+      "loss": 0.7382,
+      "step": 7077
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.345738144005042e-06,
+      "loss": 0.8647,
+      "step": 7078
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.344977942334961e-06,
+      "loss": 0.628,
+      "step": 7079
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.344217740664881e-06,
+      "loss": 0.7663,
+      "step": 7080
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.343457538994801e-06,
+      "loss": 0.8969,
+      "step": 7081
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.342697337324721e-06,
+      "loss": 0.9805,
+      "step": 7082
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.341937135654641e-06,
+      "loss": 0.8887,
+      "step": 7083
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.3411769339845605e-06,
+      "loss": 0.8621,
+      "step": 7084
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.3404167323144795e-06,
+      "loss": 0.8819,
+      "step": 7085
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.339656530644399e-06,
+      "loss": 0.8297,
+      "step": 7086
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.338896328974319e-06,
+      "loss": 0.765,
+      "step": 7087
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.338136127304239e-06,
+      "loss": 0.8621,
+      "step": 7088
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.337375925634159e-06,
+      "loss": 0.8819,
+      "step": 7089
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.336615723964078e-06,
+      "loss": 0.8117,
+      "step": 7090
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.335855522293998e-06,
+      "loss": 0.8116,
+      "step": 7091
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.335095320623918e-06,
+      "loss": 0.9071,
+      "step": 7092
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.3343351189538375e-06,
+      "loss": 0.885,
+      "step": 7093
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.333574917283757e-06,
+      "loss": 1.0367,
+      "step": 7094
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3328147156136755e-06,
+      "loss": 0.9081,
+      "step": 7095
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.332054513943595e-06,
+      "loss": 0.9062,
+      "step": 7096
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.331294312273515e-06,
+      "loss": 0.7042,
+      "step": 7097
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.330534110603435e-06,
+      "loss": 0.7391,
+      "step": 7098
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.329773908933355e-06,
+      "loss": 0.8332,
+      "step": 7099
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.329013707263275e-06,
+      "loss": 1.2117,
+      "step": 7100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.328253505593194e-06,
+      "loss": 0.7456,
+      "step": 7101
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.327493303923114e-06,
+      "loss": 0.805,
+      "step": 7102
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3267331022530335e-06,
+      "loss": 0.8362,
+      "step": 7103
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.325972900582953e-06,
+      "loss": 0.8694,
+      "step": 7104
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.325212698912873e-06,
+      "loss": 0.6756,
+      "step": 7105
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.324452497242792e-06,
+      "loss": 0.7673,
+      "step": 7106
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.323692295572712e-06,
+      "loss": 0.9124,
+      "step": 7107
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.322932093902632e-06,
+      "loss": 0.9012,
+      "step": 7108
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.322171892232552e-06,
+      "loss": 0.6749,
+      "step": 7109
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.321411690562472e-06,
+      "loss": 0.8492,
+      "step": 7110
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.32065148889239e-06,
+      "loss": 0.7884,
+      "step": 7111
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.31989128722231e-06,
+      "loss": 0.9452,
+      "step": 7112
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3191310855522295e-06,
+      "loss": 0.804,
+      "step": 7113
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.318370883882149e-06,
+      "loss": 0.9894,
+      "step": 7114
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.317610682212069e-06,
+      "loss": 0.7587,
+      "step": 7115
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.316850480541989e-06,
+      "loss": 0.9808,
+      "step": 7116
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.316090278871908e-06,
+      "loss": 0.8615,
+      "step": 7117
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.315330077201828e-06,
+      "loss": 0.8144,
+      "step": 7118
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.314569875531748e-06,
+      "loss": 0.7485,
+      "step": 7119
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.313809673861668e-06,
+      "loss": 0.8644,
+      "step": 7120
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3130494721915875e-06,
+      "loss": 0.7979,
+      "step": 7121
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3122892705215065e-06,
+      "loss": 0.946,
+      "step": 7122
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.311529068851426e-06,
+      "loss": 0.8427,
+      "step": 7123
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.310768867181346e-06,
+      "loss": 0.6796,
+      "step": 7124
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.310008665511266e-06,
+      "loss": 0.9059,
+      "step": 7125
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.309248463841186e-06,
+      "loss": 0.8039,
+      "step": 7126
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.308488262171104e-06,
+      "loss": 1.0561,
+      "step": 7127
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.307728060501024e-06,
+      "loss": 0.809,
+      "step": 7128
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.306967858830944e-06,
+      "loss": 0.7985,
+      "step": 7129
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.306207657160864e-06,
+      "loss": 0.8389,
+      "step": 7130
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.3054474554907835e-06,
+      "loss": 0.9507,
+      "step": 7131
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.304687253820703e-06,
+      "loss": 0.6804,
+      "step": 7132
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.303927052150622e-06,
+      "loss": 0.7295,
+      "step": 7133
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.303166850480542e-06,
+      "loss": 0.9673,
+      "step": 7134
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.302406648810462e-06,
+      "loss": 0.94,
+      "step": 7135
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.301646447140382e-06,
+      "loss": 0.7915,
+      "step": 7136
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.300886245470302e-06,
+      "loss": 0.8428,
+      "step": 7137
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.300126043800221e-06,
+      "loss": 0.8898,
+      "step": 7138
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.299365842130141e-06,
+      "loss": 0.7485,
+      "step": 7139
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.2986056404600605e-06,
+      "loss": 0.7505,
+      "step": 7140
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.29784543878998e-06,
+      "loss": 0.8896,
+      "step": 7141
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.2970852371199e-06,
+      "loss": 0.8794,
+      "step": 7142
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.296325035449818e-06,
+      "loss": 0.9035,
+      "step": 7143
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.295564833779738e-06,
+      "loss": 0.6531,
+      "step": 7144
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.294804632109658e-06,
+      "loss": 0.9478,
+      "step": 7145
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.294044430439578e-06,
+      "loss": 0.8455,
+      "step": 7146
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.293284228769498e-06,
+      "loss": 0.9247,
+      "step": 7147
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.292524027099418e-06,
+      "loss": 0.6634,
+      "step": 7148
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.291763825429337e-06,
+      "loss": 0.7174,
+      "step": 7149
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.2910036237592566e-06,
+      "loss": 0.9566,
+      "step": 7150
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.290243422089176e-06,
+      "loss": 0.9164,
+      "step": 7151
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.289483220419096e-06,
+      "loss": 0.907,
+      "step": 7152
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.288723018749016e-06,
+      "loss": 0.971,
+      "step": 7153
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.287962817078935e-06,
+      "loss": 0.9268,
+      "step": 7154
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.287202615408855e-06,
+      "loss": 0.7717,
+      "step": 7155
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.286442413738775e-06,
+      "loss": 0.8739,
+      "step": 7156
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.285682212068695e-06,
+      "loss": 0.8117,
+      "step": 7157
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.2849220103986146e-06,
+      "loss": 0.9334,
+      "step": 7158
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.284161808728533e-06,
+      "loss": 0.7978,
+      "step": 7159
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.283401607058453e-06,
+      "loss": 0.8708,
+      "step": 7160
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.2826414053883724e-06,
+      "loss": 0.7516,
+      "step": 7161
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.281881203718292e-06,
+      "loss": 1.0476,
+      "step": 7162
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.281121002048212e-06,
+      "loss": 0.9471,
+      "step": 7163
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.280360800378131e-06,
+      "loss": 0.8446,
+      "step": 7164
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.279600598708051e-06,
+      "loss": 0.8356,
+      "step": 7165
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.278840397037971e-06,
+      "loss": 0.6262,
+      "step": 7166
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.278080195367891e-06,
+      "loss": 0.9749,
+      "step": 7167
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.277319993697811e-06,
+      "loss": 0.7604,
+      "step": 7168
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.2765597920277304e-06,
+      "loss": 0.7375,
+      "step": 7169
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.2757995903576494e-06,
+      "loss": 0.9669,
+      "step": 7170
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.275039388687569e-06,
+      "loss": 0.9837,
+      "step": 7171
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.274279187017489e-06,
+      "loss": 0.8638,
+      "step": 7172
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.273518985347409e-06,
+      "loss": 0.8239,
+      "step": 7173
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.272758783677329e-06,
+      "loss": 0.6732,
+      "step": 7174
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.271998582007248e-06,
+      "loss": 0.8966,
+      "step": 7175
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.271238380337168e-06,
+      "loss": 0.9909,
+      "step": 7176
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.270478178667088e-06,
+      "loss": 0.8555,
+      "step": 7177
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.2697179769970074e-06,
+      "loss": 0.9426,
+      "step": 7178
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.2689577753269265e-06,
+      "loss": 0.7223,
+      "step": 7179
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.2681975736568455e-06,
+      "loss": 0.9047,
+      "step": 7180
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.267437371986765e-06,
+      "loss": 0.943,
+      "step": 7181
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.266677170316685e-06,
+      "loss": 0.9298,
+      "step": 7182
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.265916968646605e-06,
+      "loss": 0.8783,
+      "step": 7183
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.265156766976525e-06,
+      "loss": 1.0709,
+      "step": 7184
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.264396565306445e-06,
+      "loss": 0.7672,
+      "step": 7185
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.263636363636364e-06,
+      "loss": 1.0602,
+      "step": 7186
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.262876161966284e-06,
+      "loss": 0.7999,
+      "step": 7187
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.2621159602962035e-06,
+      "loss": 0.8678,
+      "step": 7188
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.261355758626123e-06,
+      "loss": 0.8714,
+      "step": 7189
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.260595556956043e-06,
+      "loss": 0.734,
+      "step": 7190
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.259835355285962e-06,
+      "loss": 1.0038,
+      "step": 7191
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.259075153615882e-06,
+      "loss": 0.8237,
+      "step": 7192
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.258314951945802e-06,
+      "loss": 0.8524,
+      "step": 7193
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.257554750275722e-06,
+      "loss": 1.0223,
+      "step": 7194
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.256794548605642e-06,
+      "loss": 0.8869,
+      "step": 7195
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.25603434693556e-06,
+      "loss": 0.7605,
+      "step": 7196
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.25527414526548e-06,
+      "loss": 0.7974,
+      "step": 7197
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.2545139435953995e-06,
+      "loss": 0.911,
+      "step": 7198
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.253753741925319e-06,
+      "loss": 0.9151,
+      "step": 7199
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.252993540255239e-06,
+      "loss": 0.8856,
+      "step": 7200
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.252233338585159e-06,
+      "loss": 0.859,
+      "step": 7201
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.251473136915078e-06,
+      "loss": 0.8751,
+      "step": 7202
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.250712935244998e-06,
+      "loss": 0.7183,
+      "step": 7203
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.249952733574918e-06,
+      "loss": 0.9995,
+      "step": 7204
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.249192531904838e-06,
+      "loss": 0.7626,
+      "step": 7205
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.2484323302347575e-06,
+      "loss": 0.7216,
+      "step": 7206
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.2476721285646765e-06,
+      "loss": 0.7973,
+      "step": 7207
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.246911926894596e-06,
+      "loss": 0.7801,
+      "step": 7208
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.246151725224516e-06,
+      "loss": 0.8308,
+      "step": 7209
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.245391523554436e-06,
+      "loss": 0.6486,
+      "step": 7210
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.244631321884356e-06,
+      "loss": 0.7906,
+      "step": 7211
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.243871120214274e-06,
+      "loss": 0.832,
+      "step": 7212
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.243110918544194e-06,
+      "loss": 0.9204,
+      "step": 7213
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.242350716874114e-06,
+      "loss": 0.9116,
+      "step": 7214
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.241590515204034e-06,
+      "loss": 0.9251,
+      "step": 7215
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.2408303135339535e-06,
+      "loss": 0.7061,
+      "step": 7216
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.240070111863873e-06,
+      "loss": 0.8403,
+      "step": 7217
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.239309910193792e-06,
+      "loss": 1.0354,
+      "step": 7218
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.238549708523712e-06,
+      "loss": 0.9817,
+      "step": 7219
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.237789506853632e-06,
+      "loss": 0.7643,
+      "step": 7220
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.237029305183552e-06,
+      "loss": 0.9007,
+      "step": 7221
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.236269103513472e-06,
+      "loss": 0.7196,
+      "step": 7222
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.235508901843391e-06,
+      "loss": 0.8592,
+      "step": 7223
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.234748700173311e-06,
+      "loss": 0.6459,
+      "step": 7224
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.2339884985032305e-06,
+      "loss": 1.024,
+      "step": 7225
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.23322829683315e-06,
+      "loss": 0.8314,
+      "step": 7226
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.23246809516307e-06,
+      "loss": 0.7534,
+      "step": 7227
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.231707893492988e-06,
+      "loss": 0.9029,
+      "step": 7228
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.230947691822908e-06,
+      "loss": 0.728,
+      "step": 7229
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.230187490152828e-06,
+      "loss": 0.7909,
+      "step": 7230
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.229427288482748e-06,
+      "loss": 0.8285,
+      "step": 7231
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.228667086812668e-06,
+      "loss": 0.8843,
+      "step": 7232
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.227906885142588e-06,
+      "loss": 0.8808,
+      "step": 7233
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.227146683472507e-06,
+      "loss": 0.8026,
+      "step": 7234
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.2263864818024265e-06,
+      "loss": 0.9939,
+      "step": 7235
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.225626280132346e-06,
+      "loss": 0.7643,
+      "step": 7236
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.224866078462266e-06,
+      "loss": 0.8986,
+      "step": 7237
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.224105876792186e-06,
+      "loss": 0.7175,
+      "step": 7238
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.223345675122105e-06,
+      "loss": 0.9708,
+      "step": 7239
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.222585473452025e-06,
+      "loss": 0.8111,
+      "step": 7240
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.221825271781945e-06,
+      "loss": 0.6726,
+      "step": 7241
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.221065070111865e-06,
+      "loss": 0.8466,
+      "step": 7242
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.2203048684417845e-06,
+      "loss": 0.6501,
+      "step": 7243
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.219544666771703e-06,
+      "loss": 0.8178,
+      "step": 7244
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.2187844651016226e-06,
+      "loss": 0.8023,
+      "step": 7245
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.218024263431542e-06,
+      "loss": 0.9502,
+      "step": 7246
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.217264061761462e-06,
+      "loss": 0.7904,
+      "step": 7247
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.216503860091382e-06,
+      "loss": 0.8166,
+      "step": 7248
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.215743658421302e-06,
+      "loss": 0.7344,
+      "step": 7249
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.214983456751221e-06,
+      "loss": 0.6624,
+      "step": 7250
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.214223255081141e-06,
+      "loss": 0.7572,
+      "step": 7251
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.213463053411061e-06,
+      "loss": 1.0869,
+      "step": 7252
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.2127028517409806e-06,
+      "loss": 0.799,
+      "step": 7253
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.2119426500709e-06,
+      "loss": 0.8939,
+      "step": 7254
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.211182448400819e-06,
+      "loss": 0.7587,
+      "step": 7255
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.210422246730739e-06,
+      "loss": 0.788,
+      "step": 7256
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.209662045060659e-06,
+      "loss": 0.7551,
+      "step": 7257
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.208901843390579e-06,
+      "loss": 0.7319,
+      "step": 7258
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.208141641720499e-06,
+      "loss": 1.158,
+      "step": 7259
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.207381440050418e-06,
+      "loss": 0.8723,
+      "step": 7260
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.206621238380338e-06,
+      "loss": 0.7889,
+      "step": 7261
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.205861036710257e-06,
+      "loss": 0.8193,
+      "step": 7262
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.2051008350401766e-06,
+      "loss": 0.8485,
+      "step": 7263
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.2043406333700964e-06,
+      "loss": 0.8216,
+      "step": 7264
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.203580431700016e-06,
+      "loss": 0.7433,
+      "step": 7265
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.202820230029935e-06,
+      "loss": 0.8845,
+      "step": 7266
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.202060028359855e-06,
+      "loss": 0.6992,
+      "step": 7267
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.201299826689775e-06,
+      "loss": 0.9265,
+      "step": 7268
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.200539625019695e-06,
+      "loss": 0.865,
+      "step": 7269
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.199779423349615e-06,
+      "loss": 0.8058,
+      "step": 7270
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.199019221679534e-06,
+      "loss": 0.7538,
+      "step": 7271
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.198259020009454e-06,
+      "loss": 0.7485,
+      "step": 7272
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.1974988183393734e-06,
+      "loss": 1.0023,
+      "step": 7273
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.196738616669293e-06,
+      "loss": 0.891,
+      "step": 7274
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.195978414999213e-06,
+      "loss": 1.004,
+      "step": 7275
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.195218213329132e-06,
+      "loss": 0.971,
+      "step": 7276
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.194458011659052e-06,
+      "loss": 0.8684,
+      "step": 7277
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.193697809988972e-06,
+      "loss": 0.8614,
+      "step": 7278
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.192937608318892e-06,
+      "loss": 0.8601,
+      "step": 7279
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.192177406648812e-06,
+      "loss": 0.7475,
+      "step": 7280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.1914172049787314e-06,
+      "loss": 0.9744,
+      "step": 7281
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.19065700330865e-06,
+      "loss": 0.8257,
+      "step": 7282
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.1898968016385695e-06,
+      "loss": 0.927,
+      "step": 7283
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.189136599968489e-06,
+      "loss": 0.7656,
+      "step": 7284
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.188376398298409e-06,
+      "loss": 0.8384,
+      "step": 7285
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.187616196628329e-06,
+      "loss": 0.8519,
+      "step": 7286
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.186855994958248e-06,
+      "loss": 0.9228,
+      "step": 7287
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.186095793288168e-06,
+      "loss": 0.9134,
+      "step": 7288
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.185335591618088e-06,
+      "loss": 0.7264,
+      "step": 7289
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.184575389948008e-06,
+      "loss": 1.0182,
+      "step": 7290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.1838151882779275e-06,
+      "loss": 0.9682,
+      "step": 7291
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.1830549866078465e-06,
+      "loss": 0.9255,
+      "step": 7292
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.182294784937766e-06,
+      "loss": 0.8189,
+      "step": 7293
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.181534583267686e-06,
+      "loss": 0.9016,
+      "step": 7294
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.180774381597606e-06,
+      "loss": 0.8546,
+      "step": 7295
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.180014179927526e-06,
+      "loss": 0.936,
+      "step": 7296
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.179253978257446e-06,
+      "loss": 0.8103,
+      "step": 7297
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.178493776587364e-06,
+      "loss": 0.927,
+      "step": 7298
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.177733574917284e-06,
+      "loss": 0.9824,
+      "step": 7299
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.176973373247204e-06,
+      "loss": 0.7717,
+      "step": 7300
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.1762131715771235e-06,
+      "loss": 1.0909,
+      "step": 7301
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.175452969907043e-06,
+      "loss": 0.7293,
+      "step": 7302
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.174692768236962e-06,
+      "loss": 0.9366,
+      "step": 7303
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.173932566566882e-06,
+      "loss": 0.9861,
+      "step": 7304
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.173172364896802e-06,
+      "loss": 0.931,
+      "step": 7305
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.172412163226722e-06,
+      "loss": 0.8935,
+      "step": 7306
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.171651961556642e-06,
+      "loss": 0.6846,
+      "step": 7307
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.170891759886561e-06,
+      "loss": 0.9046,
+      "step": 7308
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.170131558216481e-06,
+      "loss": 0.8824,
+      "step": 7309
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.1693713565464005e-06,
+      "loss": 0.883,
+      "step": 7310
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.16861115487632e-06,
+      "loss": 1.1139,
+      "step": 7311
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.16785095320624e-06,
+      "loss": 0.8644,
+      "step": 7312
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.16709075153616e-06,
+      "loss": 0.9586,
+      "step": 7313
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.166330549866078e-06,
+      "loss": 1.1771,
+      "step": 7314
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.165570348195998e-06,
+      "loss": 0.8004,
+      "step": 7315
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.164810146525918e-06,
+      "loss": 0.8934,
+      "step": 7316
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.164049944855838e-06,
+      "loss": 0.8373,
+      "step": 7317
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.163289743185758e-06,
+      "loss": 0.9491,
+      "step": 7318
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.162529541515677e-06,
+      "loss": 0.868,
+      "step": 7319
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.1617693398455965e-06,
+      "loss": 0.6879,
+      "step": 7320
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.161009138175516e-06,
+      "loss": 0.823,
+      "step": 7321
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.160248936505436e-06,
+      "loss": 0.8705,
+      "step": 7322
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.159488734835356e-06,
+      "loss": 0.7939,
+      "step": 7323
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.158728533165275e-06,
+      "loss": 0.6607,
+      "step": 7324
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.157968331495195e-06,
+      "loss": 0.8953,
+      "step": 7325
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.157208129825115e-06,
+      "loss": 0.8608,
+      "step": 7326
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.156447928155035e-06,
+      "loss": 0.8628,
+      "step": 7327
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.1556877264849545e-06,
+      "loss": 0.78,
+      "step": 7328
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.154927524814874e-06,
+      "loss": 0.9556,
+      "step": 7329
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.1541673231447925e-06,
+      "loss": 0.902,
+      "step": 7330
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.153407121474712e-06,
+      "loss": 0.8846,
+      "step": 7331
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.152646919804632e-06,
+      "loss": 0.8704,
+      "step": 7332
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.151886718134552e-06,
+      "loss": 0.9359,
+      "step": 7333
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.151126516464472e-06,
+      "loss": 0.9247,
+      "step": 7334
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.150366314794391e-06,
+      "loss": 0.8012,
+      "step": 7335
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.149606113124311e-06,
+      "loss": 0.9827,
+      "step": 7336
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.148845911454231e-06,
+      "loss": 0.8652,
+      "step": 7337
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.1480857097841505e-06,
+      "loss": 0.7202,
+      "step": 7338
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.14732550811407e-06,
+      "loss": 0.8811,
+      "step": 7339
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.146565306443989e-06,
+      "loss": 0.7365,
+      "step": 7340
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.145805104773909e-06,
+      "loss": 0.9621,
+      "step": 7341
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.145044903103829e-06,
+      "loss": 0.6829,
+      "step": 7342
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.144284701433749e-06,
+      "loss": 0.9927,
+      "step": 7343
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.143524499763669e-06,
+      "loss": 0.8052,
+      "step": 7344
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.142764298093589e-06,
+      "loss": 0.7426,
+      "step": 7345
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.142004096423507e-06,
+      "loss": 0.9502,
+      "step": 7346
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.141243894753427e-06,
+      "loss": 0.8227,
+      "step": 7347
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.1404836930833465e-06,
+      "loss": 0.8356,
+      "step": 7348
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.139723491413266e-06,
+      "loss": 0.7834,
+      "step": 7349
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.138963289743186e-06,
+      "loss": 1.0389,
+      "step": 7350
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.138203088073105e-06,
+      "loss": 0.9072,
+      "step": 7351
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.137442886403025e-06,
+      "loss": 0.8318,
+      "step": 7352
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.136682684732945e-06,
+      "loss": 0.7752,
+      "step": 7353
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.135922483062865e-06,
+      "loss": 0.7264,
+      "step": 7354
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.135162281392785e-06,
+      "loss": 0.7834,
+      "step": 7355
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.134402079722704e-06,
+      "loss": 0.8481,
+      "step": 7356
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1336418780526236e-06,
+      "loss": 0.8895,
+      "step": 7357
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.132881676382543e-06,
+      "loss": 0.8592,
+      "step": 7358
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.132121474712463e-06,
+      "loss": 0.9379,
+      "step": 7359
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.131361273042383e-06,
+      "loss": 0.7018,
+      "step": 7360
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.130601071372302e-06,
+      "loss": 0.7803,
+      "step": 7361
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.129840869702222e-06,
+      "loss": 0.7653,
+      "step": 7362
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.129080668032142e-06,
+      "loss": 0.8178,
+      "step": 7363
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.128320466362062e-06,
+      "loss": 0.771,
+      "step": 7364
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1275602646919816e-06,
+      "loss": 0.7652,
+      "step": 7365
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.126800063021901e-06,
+      "loss": 0.9608,
+      "step": 7366
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1260398613518196e-06,
+      "loss": 0.9617,
+      "step": 7367
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1252796596817394e-06,
+      "loss": 1.0548,
+      "step": 7368
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.124519458011659e-06,
+      "loss": 0.8694,
+      "step": 7369
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.123759256341579e-06,
+      "loss": 0.8703,
+      "step": 7370
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.122999054671499e-06,
+      "loss": 0.8434,
+      "step": 7371
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.122238853001418e-06,
+      "loss": 0.8808,
+      "step": 7372
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.121478651331338e-06,
+      "loss": 1.1054,
+      "step": 7373
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.120718449661258e-06,
+      "loss": 0.9261,
+      "step": 7374
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1199582479911776e-06,
+      "loss": 0.8251,
+      "step": 7375
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1191980463210974e-06,
+      "loss": 0.9394,
+      "step": 7376
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1184378446510164e-06,
+      "loss": 0.7225,
+      "step": 7377
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.117677642980936e-06,
+      "loss": 0.7096,
+      "step": 7378
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.116917441310856e-06,
+      "loss": 0.8136,
+      "step": 7379
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.116157239640776e-06,
+      "loss": 0.8208,
+      "step": 7380
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.115397037970696e-06,
+      "loss": 0.8171,
+      "step": 7381
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.114636836300616e-06,
+      "loss": 0.7048,
+      "step": 7382
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.113876634630534e-06,
+      "loss": 0.9066,
+      "step": 7383
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.113116432960454e-06,
+      "loss": 0.9434,
+      "step": 7384
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.112356231290374e-06,
+      "loss": 0.635,
+      "step": 7385
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1115960296202935e-06,
+      "loss": 1.0206,
+      "step": 7386
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.110835827950213e-06,
+      "loss": 0.8751,
+      "step": 7387
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.110075626280132e-06,
+      "loss": 0.8903,
+      "step": 7388
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.109315424610052e-06,
+      "loss": 0.8123,
+      "step": 7389
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.108555222939972e-06,
+      "loss": 0.7852,
+      "step": 7390
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.107795021269892e-06,
+      "loss": 0.7475,
+      "step": 7391
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.107034819599812e-06,
+      "loss": 0.9416,
+      "step": 7392
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.106274617929731e-06,
+      "loss": 0.7531,
+      "step": 7393
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.105514416259651e-06,
+      "loss": 1.032,
+      "step": 7394
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1047542145895705e-06,
+      "loss": 0.9633,
+      "step": 7395
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.10399401291949e-06,
+      "loss": 0.6855,
+      "step": 7396
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.10323381124941e-06,
+      "loss": 0.7351,
+      "step": 7397
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.10247360957933e-06,
+      "loss": 0.8089,
+      "step": 7398
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.101713407909248e-06,
+      "loss": 0.8969,
+      "step": 7399
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.100953206239168e-06,
+      "loss": 0.805,
+      "step": 7400
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.100193004569088e-06,
+      "loss": 0.8895,
+      "step": 7401
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.099432802899008e-06,
+      "loss": 0.9085,
+      "step": 7402
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.098672601228928e-06,
+      "loss": 0.7679,
+      "step": 7403
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.097912399558847e-06,
+      "loss": 0.7386,
+      "step": 7404
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.0971521978887665e-06,
+      "loss": 0.93,
+      "step": 7405
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.096391996218686e-06,
+      "loss": 0.9961,
+      "step": 7406
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.095631794548606e-06,
+      "loss": 0.9133,
+      "step": 7407
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.094871592878526e-06,
+      "loss": 0.9628,
+      "step": 7408
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.094111391208445e-06,
+      "loss": 0.9792,
+      "step": 7409
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.093351189538365e-06,
+      "loss": 0.7498,
+      "step": 7410
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.092590987868285e-06,
+      "loss": 0.8284,
+      "step": 7411
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.091830786198205e-06,
+      "loss": 0.9085,
+      "step": 7412
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0910705845281245e-06,
+      "loss": 0.8007,
+      "step": 7413
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.090310382858044e-06,
+      "loss": 0.8386,
+      "step": 7414
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0895501811879625e-06,
+      "loss": 0.7075,
+      "step": 7415
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.088789979517882e-06,
+      "loss": 0.8955,
+      "step": 7416
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.088029777847802e-06,
+      "loss": 0.853,
+      "step": 7417
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.087269576177722e-06,
+      "loss": 0.8787,
+      "step": 7418
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.086509374507642e-06,
+      "loss": 0.9958,
+      "step": 7419
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.085749172837561e-06,
+      "loss": 0.8807,
+      "step": 7420
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.084988971167481e-06,
+      "loss": 0.8053,
+      "step": 7421
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.084228769497401e-06,
+      "loss": 0.837,
+      "step": 7422
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0834685678273205e-06,
+      "loss": 0.8287,
+      "step": 7423
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.08270836615724e-06,
+      "loss": 0.9336,
+      "step": 7424
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.081948164487159e-06,
+      "loss": 0.7607,
+      "step": 7425
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.081187962817079e-06,
+      "loss": 0.6798,
+      "step": 7426
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.080427761146999e-06,
+      "loss": 0.9168,
+      "step": 7427
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.079667559476919e-06,
+      "loss": 0.7636,
+      "step": 7428
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.078907357806839e-06,
+      "loss": 0.9557,
+      "step": 7429
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.078147156136759e-06,
+      "loss": 0.9279,
+      "step": 7430
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.077386954466677e-06,
+      "loss": 0.9647,
+      "step": 7431
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.076626752796597e-06,
+      "loss": 1.0465,
+      "step": 7432
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0758665511265165e-06,
+      "loss": 0.8157,
+      "step": 7433
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.075106349456436e-06,
+      "loss": 0.8695,
+      "step": 7434
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.074346147786356e-06,
+      "loss": 0.878,
+      "step": 7435
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.073585946116275e-06,
+      "loss": 0.694,
+      "step": 7436
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.072825744446195e-06,
+      "loss": 0.9959,
+      "step": 7437
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.072065542776115e-06,
+      "loss": 0.6902,
+      "step": 7438
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.071305341106035e-06,
+      "loss": 0.7718,
+      "step": 7439
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.070545139435955e-06,
+      "loss": 0.9147,
+      "step": 7440
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.069784937765874e-06,
+      "loss": 0.8892,
+      "step": 7441
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0690247360957935e-06,
+      "loss": 0.8525,
+      "step": 7442
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.068264534425713e-06,
+      "loss": 0.6788,
+      "step": 7443
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.067504332755633e-06,
+      "loss": 0.6849,
+      "step": 7444
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.066744131085553e-06,
+      "loss": 0.8079,
+      "step": 7445
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.065983929415473e-06,
+      "loss": 0.9329,
+      "step": 7446
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.065223727745392e-06,
+      "loss": 0.9252,
+      "step": 7447
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.064463526075312e-06,
+      "loss": 0.7899,
+      "step": 7448
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.063703324405232e-06,
+      "loss": 0.7517,
+      "step": 7449
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0629431227351515e-06,
+      "loss": 0.8333,
+      "step": 7450
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0621829210650705e-06,
+      "loss": 0.8479,
+      "step": 7451
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0614227193949895e-06,
+      "loss": 0.7232,
+      "step": 7452
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.060662517724909e-06,
+      "loss": 0.9648,
+      "step": 7453
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.059902316054829e-06,
+      "loss": 0.7132,
+      "step": 7454
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.059142114384749e-06,
+      "loss": 0.7964,
+      "step": 7455
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.058381912714669e-06,
+      "loss": 0.6748,
+      "step": 7456
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.057621711044588e-06,
+      "loss": 0.875,
+      "step": 7457
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.056861509374508e-06,
+      "loss": 0.7689,
+      "step": 7458
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.056101307704428e-06,
+      "loss": 0.9258,
+      "step": 7459
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0553411060343475e-06,
+      "loss": 0.7367,
+      "step": 7460
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.054580904364267e-06,
+      "loss": 1.0238,
+      "step": 7461
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.053820702694187e-06,
+      "loss": 0.825,
+      "step": 7462
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.053060501024106e-06,
+      "loss": 0.9614,
+      "step": 7463
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.052300299354026e-06,
+      "loss": 0.7531,
+      "step": 7464
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.051540097683946e-06,
+      "loss": 0.9747,
+      "step": 7465
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.050779896013866e-06,
+      "loss": 0.9256,
+      "step": 7466
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.050019694343786e-06,
+      "loss": 0.7328,
+      "step": 7467
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.049259492673704e-06,
+      "loss": 0.8823,
+      "step": 7468
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.048499291003624e-06,
+      "loss": 0.7519,
+      "step": 7469
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.0477390893335436e-06,
+      "loss": 0.7621,
+      "step": 7470
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.046978887663463e-06,
+      "loss": 0.9036,
+      "step": 7471
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.046218685993383e-06,
+      "loss": 1.0112,
+      "step": 7472
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.045458484323302e-06,
+      "loss": 1.0184,
+      "step": 7473
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.044698282653222e-06,
+      "loss": 0.8452,
+      "step": 7474
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.043938080983142e-06,
+      "loss": 0.8969,
+      "step": 7475
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.043177879313062e-06,
+      "loss": 0.6856,
+      "step": 7476
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.042417677642982e-06,
+      "loss": 0.7705,
+      "step": 7477
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.0416574759729016e-06,
+      "loss": 0.858,
+      "step": 7478
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.040897274302821e-06,
+      "loss": 0.8738,
+      "step": 7479
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.0401370726327404e-06,
+      "loss": 1.0133,
+      "step": 7480
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.03937687096266e-06,
+      "loss": 0.9743,
+      "step": 7481
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.03861666929258e-06,
+      "loss": 0.962,
+      "step": 7482
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.0378564676225e-06,
+      "loss": 0.8048,
+      "step": 7483
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.037096265952418e-06,
+      "loss": 0.8655,
+      "step": 7484
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.036336064282338e-06,
+      "loss": 0.7626,
+      "step": 7485
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.035575862612258e-06,
+      "loss": 0.7866,
+      "step": 7486
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.034815660942178e-06,
+      "loss": 0.871,
+      "step": 7487
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.034055459272098e-06,
+      "loss": 0.9676,
+      "step": 7488
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.033295257602017e-06,
+      "loss": 0.8265,
+      "step": 7489
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.0325350559319365e-06,
+      "loss": 0.8938,
+      "step": 7490
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.031774854261856e-06,
+      "loss": 0.7476,
+      "step": 7491
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.031014652591776e-06,
+      "loss": 0.9202,
+      "step": 7492
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.030254450921696e-06,
+      "loss": 0.9518,
+      "step": 7493
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.029494249251616e-06,
+      "loss": 0.8767,
+      "step": 7494
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.028734047581535e-06,
+      "loss": 0.7331,
+      "step": 7495
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.027973845911455e-06,
+      "loss": 0.7875,
+      "step": 7496
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.027213644241375e-06,
+      "loss": 0.8111,
+      "step": 7497
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.0264534425712945e-06,
+      "loss": 0.7986,
+      "step": 7498
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.025693240901214e-06,
+      "loss": 0.864,
+      "step": 7499
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.0249330392311325e-06,
+      "loss": 1.008,
+      "step": 7500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.024172837561052e-06,
+      "loss": 0.9952,
+      "step": 7501
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.023412635890972e-06,
+      "loss": 0.7213,
+      "step": 7502
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.022652434220892e-06,
+      "loss": 0.7757,
+      "step": 7503
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.021892232550812e-06,
+      "loss": 0.874,
+      "step": 7504
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.021132030880731e-06,
+      "loss": 0.7199,
+      "step": 7505
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.020371829210651e-06,
+      "loss": 0.7223,
+      "step": 7506
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.019611627540571e-06,
+      "loss": 0.8317,
+      "step": 7507
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.0188514258704905e-06,
+      "loss": 0.6689,
+      "step": 7508
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.01809122420041e-06,
+      "loss": 0.9104,
+      "step": 7509
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.01733102253033e-06,
+      "loss": 0.9655,
+      "step": 7510
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.016570820860249e-06,
+      "loss": 1.0032,
+      "step": 7511
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.015810619190169e-06,
+      "loss": 0.8693,
+      "step": 7512
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.015050417520089e-06,
+      "loss": 0.8398,
+      "step": 7513
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.014290215850009e-06,
+      "loss": 0.8846,
+      "step": 7514
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.013530014179929e-06,
+      "loss": 0.8295,
+      "step": 7515
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.012769812509847e-06,
+      "loss": 1.0505,
+      "step": 7516
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.012009610839767e-06,
+      "loss": 0.9628,
+      "step": 7517
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.0112494091696865e-06,
+      "loss": 0.9166,
+      "step": 7518
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.010489207499606e-06,
+      "loss": 0.8797,
+      "step": 7519
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.009729005829526e-06,
+      "loss": 0.8825,
+      "step": 7520
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.008968804159445e-06,
+      "loss": 0.7771,
+      "step": 7521
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.008208602489365e-06,
+      "loss": 0.8978,
+      "step": 7522
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.007448400819285e-06,
+      "loss": 0.8686,
+      "step": 7523
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.006688199149205e-06,
+      "loss": 0.7899,
+      "step": 7524
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.005927997479125e-06,
+      "loss": 0.911,
+      "step": 7525
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.0051677958090445e-06,
+      "loss": 0.9107,
+      "step": 7526
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.0044075941389635e-06,
+      "loss": 0.8544,
+      "step": 7527
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.003647392468883e-06,
+      "loss": 0.9794,
+      "step": 7528
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.002887190798803e-06,
+      "loss": 0.9423,
+      "step": 7529
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.002126989128723e-06,
+      "loss": 0.8735,
+      "step": 7530
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.001366787458643e-06,
+      "loss": 0.7933,
+      "step": 7531
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.000606585788562e-06,
+      "loss": 0.8237,
+      "step": 7532
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.999846384118482e-06,
+      "loss": 0.7454,
+      "step": 7533
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.999086182448401e-06,
+      "loss": 0.9461,
+      "step": 7534
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.998325980778321e-06,
+      "loss": 0.6114,
+      "step": 7535
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.9975657791082405e-06,
+      "loss": 0.8502,
+      "step": 7536
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.9968055774381595e-06,
+      "loss": 0.9746,
+      "step": 7537
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.996045375768079e-06,
+      "loss": 0.8526,
+      "step": 7538
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.995285174097999e-06,
+      "loss": 0.9694,
+      "step": 7539
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.994524972427919e-06,
+      "loss": 0.8732,
+      "step": 7540
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.993764770757839e-06,
+      "loss": 0.8593,
+      "step": 7541
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.993004569087759e-06,
+      "loss": 0.6785,
+      "step": 7542
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.992244367417678e-06,
+      "loss": 0.8618,
+      "step": 7543
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.991484165747598e-06,
+      "loss": 0.9345,
+      "step": 7544
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.9907239640775175e-06,
+      "loss": 0.7108,
+      "step": 7545
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.989963762407437e-06,
+      "loss": 0.8529,
+      "step": 7546
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.989203560737357e-06,
+      "loss": 1.0515,
+      "step": 7547
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.988443359067276e-06,
+      "loss": 0.8863,
+      "step": 7548
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.987683157397196e-06,
+      "loss": 0.7441,
+      "step": 7549
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.986922955727116e-06,
+      "loss": 1.0487,
+      "step": 7550
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.986162754057036e-06,
+      "loss": 0.855,
+      "step": 7551
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.985402552386956e-06,
+      "loss": 0.7837,
+      "step": 7552
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.984642350716874e-06,
+      "loss": 0.8173,
+      "step": 7553
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.983882149046794e-06,
+      "loss": 0.8642,
+      "step": 7554
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.9831219473767135e-06,
+      "loss": 0.6396,
+      "step": 7555
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.982361745706633e-06,
+      "loss": 0.8652,
+      "step": 7556
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.981601544036553e-06,
+      "loss": 0.6609,
+      "step": 7557
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.980841342366473e-06,
+      "loss": 0.9913,
+      "step": 7558
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.980081140696392e-06,
+      "loss": 0.9274,
+      "step": 7559
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.979320939026312e-06,
+      "loss": 0.9286,
+      "step": 7560
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.978560737356232e-06,
+      "loss": 0.8826,
+      "step": 7561
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.977800535686152e-06,
+      "loss": 0.8373,
+      "step": 7562
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.9770403340160715e-06,
+      "loss": 0.899,
+      "step": 7563
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.9762801323459906e-06,
+      "loss": 0.9129,
+      "step": 7564
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.97551993067591e-06,
+      "loss": 0.9361,
+      "step": 7565
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.97475972900583e-06,
+      "loss": 0.9135,
+      "step": 7566
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.97399952733575e-06,
+      "loss": 1.0105,
+      "step": 7567
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.97323932566567e-06,
+      "loss": 0.8392,
+      "step": 7568
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.972479123995588e-06,
+      "loss": 0.8927,
+      "step": 7569
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.971718922325508e-06,
+      "loss": 1.0093,
+      "step": 7570
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.970958720655428e-06,
+      "loss": 1.0506,
+      "step": 7571
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.970198518985348e-06,
+      "loss": 0.7771,
+      "step": 7572
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.9694383173152676e-06,
+      "loss": 0.8524,
+      "step": 7573
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.9686781156451866e-06,
+      "loss": 1.0497,
+      "step": 7574
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.9679179139751064e-06,
+      "loss": 0.8645,
+      "step": 7575
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.967157712305026e-06,
+      "loss": 0.7667,
+      "step": 7576
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.966397510634946e-06,
+      "loss": 0.9466,
+      "step": 7577
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.965637308964866e-06,
+      "loss": 0.9013,
+      "step": 7578
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.964877107294786e-06,
+      "loss": 0.8308,
+      "step": 7579
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.964116905624705e-06,
+      "loss": 0.8116,
+      "step": 7580
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.963356703954625e-06,
+      "loss": 0.953,
+      "step": 7581
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.9625965022845446e-06,
+      "loss": 0.922,
+      "step": 7582
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.9618363006144644e-06,
+      "loss": 0.9046,
+      "step": 7583
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.961076098944384e-06,
+      "loss": 0.8861,
+      "step": 7584
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.9603158972743024e-06,
+      "loss": 0.8615,
+      "step": 7585
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.959555695604222e-06,
+      "loss": 0.9301,
+      "step": 7586
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.958795493934142e-06,
+      "loss": 0.8313,
+      "step": 7587
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.958035292264062e-06,
+      "loss": 0.7387,
+      "step": 7588
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.957275090593982e-06,
+      "loss": 0.8149,
+      "step": 7589
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.956514888923901e-06,
+      "loss": 0.9598,
+      "step": 7590
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.955754687253821e-06,
+      "loss": 0.7789,
+      "step": 7591
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.954994485583741e-06,
+      "loss": 0.8123,
+      "step": 7592
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.9542342839136604e-06,
+      "loss": 0.8316,
+      "step": 7593
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.95347408224358e-06,
+      "loss": 0.7241,
+      "step": 7594
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.9527138805735e-06,
+      "loss": 0.9028,
+      "step": 7595
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.951953678903419e-06,
+      "loss": 1.0374,
+      "step": 7596
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.951193477233339e-06,
+      "loss": 0.8897,
+      "step": 7597
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.950433275563259e-06,
+      "loss": 0.7574,
+      "step": 7598
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.949673073893179e-06,
+      "loss": 0.923,
+      "step": 7599
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.948912872223099e-06,
+      "loss": 1.0221,
+      "step": 7600
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.948152670553017e-06,
+      "loss": 0.8032,
+      "step": 7601
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.947392468882937e-06,
+      "loss": 0.8974,
+      "step": 7602
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9466322672128565e-06,
+      "loss": 0.9344,
+      "step": 7603
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.945872065542776e-06,
+      "loss": 0.9573,
+      "step": 7604
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.945111863872696e-06,
+      "loss": 0.8747,
+      "step": 7605
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.944351662202615e-06,
+      "loss": 0.9039,
+      "step": 7606
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.943591460532535e-06,
+      "loss": 0.7384,
+      "step": 7607
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.942831258862455e-06,
+      "loss": 0.9638,
+      "step": 7608
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.942071057192375e-06,
+      "loss": 0.6312,
+      "step": 7609
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.941310855522295e-06,
+      "loss": 0.6902,
+      "step": 7610
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9405506538522145e-06,
+      "loss": 0.8468,
+      "step": 7611
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9397904521821335e-06,
+      "loss": 0.8125,
+      "step": 7612
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.939030250512053e-06,
+      "loss": 0.9697,
+      "step": 7613
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.938270048841973e-06,
+      "loss": 0.7651,
+      "step": 7614
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.937509847171893e-06,
+      "loss": 0.8395,
+      "step": 7615
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.936749645501813e-06,
+      "loss": 0.8115,
+      "step": 7616
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.935989443831731e-06,
+      "loss": 0.7674,
+      "step": 7617
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.935229242161651e-06,
+      "loss": 0.825,
+      "step": 7618
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.934469040491571e-06,
+      "loss": 0.8487,
+      "step": 7619
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.933708838821491e-06,
+      "loss": 1.0357,
+      "step": 7620
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9329486371514105e-06,
+      "loss": 0.8254,
+      "step": 7621
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9321884354813295e-06,
+      "loss": 0.7428,
+      "step": 7622
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.931428233811249e-06,
+      "loss": 0.9734,
+      "step": 7623
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.930668032141169e-06,
+      "loss": 0.7789,
+      "step": 7624
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.929907830471089e-06,
+      "loss": 0.836,
+      "step": 7625
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.929147628801009e-06,
+      "loss": 0.8548,
+      "step": 7626
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.928387427130929e-06,
+      "loss": 0.8723,
+      "step": 7627
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.927627225460848e-06,
+      "loss": 0.8896,
+      "step": 7628
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.926867023790768e-06,
+      "loss": 0.8317,
+      "step": 7629
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9261068221206875e-06,
+      "loss": 0.8628,
+      "step": 7630
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.925346620450607e-06,
+      "loss": 1.0617,
+      "step": 7631
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.924586418780527e-06,
+      "loss": 0.8429,
+      "step": 7632
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.923826217110446e-06,
+      "loss": 0.82,
+      "step": 7633
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.923066015440366e-06,
+      "loss": 0.7366,
+      "step": 7634
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.922305813770286e-06,
+      "loss": 0.8687,
+      "step": 7635
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.921545612100206e-06,
+      "loss": 0.7983,
+      "step": 7636
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.920785410430126e-06,
+      "loss": 0.6901,
+      "step": 7637
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.920025208760044e-06,
+      "loss": 0.8153,
+      "step": 7638
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.919265007089964e-06,
+      "loss": 0.9213,
+      "step": 7639
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9185048054198835e-06,
+      "loss": 0.8728,
+      "step": 7640
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.917744603749803e-06,
+      "loss": 0.8816,
+      "step": 7641
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.916984402079723e-06,
+      "loss": 0.77,
+      "step": 7642
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.916224200409643e-06,
+      "loss": 0.8408,
+      "step": 7643
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.915463998739562e-06,
+      "loss": 0.851,
+      "step": 7644
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.914703797069482e-06,
+      "loss": 0.8632,
+      "step": 7645
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.913943595399402e-06,
+      "loss": 0.7836,
+      "step": 7646
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.913183393729322e-06,
+      "loss": 0.7448,
+      "step": 7647
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9124231920592415e-06,
+      "loss": 0.8033,
+      "step": 7648
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9116629903891605e-06,
+      "loss": 0.954,
+      "step": 7649
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.91090278871908e-06,
+      "loss": 0.8127,
+      "step": 7650
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.910142587049e-06,
+      "loss": 0.8794,
+      "step": 7651
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.90938238537892e-06,
+      "loss": 0.7428,
+      "step": 7652
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.90862218370884e-06,
+      "loss": 0.8066,
+      "step": 7653
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.907861982038758e-06,
+      "loss": 0.6945,
+      "step": 7654
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.907101780368678e-06,
+      "loss": 0.8238,
+      "step": 7655
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.906341578698598e-06,
+      "loss": 0.7465,
+      "step": 7656
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.905581377028518e-06,
+      "loss": 0.84,
+      "step": 7657
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9048211753584375e-06,
+      "loss": 0.7791,
+      "step": 7658
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.904060973688357e-06,
+      "loss": 0.6796,
+      "step": 7659
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.903300772018276e-06,
+      "loss": 0.9674,
+      "step": 7660
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.902540570348196e-06,
+      "loss": 0.9312,
+      "step": 7661
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.901780368678116e-06,
+      "loss": 0.8627,
+      "step": 7662
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.901020167008036e-06,
+      "loss": 0.9762,
+      "step": 7663
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.900259965337956e-06,
+      "loss": 0.7938,
+      "step": 7664
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.899499763667875e-06,
+      "loss": 0.7764,
+      "step": 7665
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.898739561997795e-06,
+      "loss": 0.8381,
+      "step": 7666
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.8979793603277145e-06,
+      "loss": 0.7721,
+      "step": 7667
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.897219158657634e-06,
+      "loss": 0.7476,
+      "step": 7668
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.896458956987554e-06,
+      "loss": 0.9261,
+      "step": 7669
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.895698755317472e-06,
+      "loss": 0.7198,
+      "step": 7670
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.894938553647392e-06,
+      "loss": 0.8752,
+      "step": 7671
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.894178351977312e-06,
+      "loss": 1.1448,
+      "step": 7672
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.893418150307232e-06,
+      "loss": 0.8903,
+      "step": 7673
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.892657948637152e-06,
+      "loss": 0.7769,
+      "step": 7674
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.891897746967072e-06,
+      "loss": 0.7921,
+      "step": 7675
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.891137545296991e-06,
+      "loss": 0.6638,
+      "step": 7676
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.8903773436269106e-06,
+      "loss": 0.8478,
+      "step": 7677
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.88961714195683e-06,
+      "loss": 0.9425,
+      "step": 7678
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.88885694028675e-06,
+      "loss": 0.8423,
+      "step": 7679
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.88809673861667e-06,
+      "loss": 0.9232,
+      "step": 7680
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.887336536946589e-06,
+      "loss": 0.8187,
+      "step": 7681
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.886576335276509e-06,
+      "loss": 0.8152,
+      "step": 7682
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.885816133606429e-06,
+      "loss": 0.7631,
+      "step": 7683
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.885055931936349e-06,
+      "loss": 0.8122,
+      "step": 7684
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.8842957302662686e-06,
+      "loss": 0.9631,
+      "step": 7685
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.883535528596187e-06,
+      "loss": 0.8285,
+      "step": 7686
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.882775326926107e-06,
+      "loss": 0.9738,
+      "step": 7687
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8820151252560264e-06,
+      "loss": 0.8266,
+      "step": 7688
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.881254923585946e-06,
+      "loss": 0.7785,
+      "step": 7689
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.880494721915866e-06,
+      "loss": 0.9912,
+      "step": 7690
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.879734520245786e-06,
+      "loss": 0.674,
+      "step": 7691
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.878974318575705e-06,
+      "loss": 1.0223,
+      "step": 7692
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.878214116905625e-06,
+      "loss": 0.8167,
+      "step": 7693
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.877453915235545e-06,
+      "loss": 1.0885,
+      "step": 7694
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.876693713565465e-06,
+      "loss": 0.8867,
+      "step": 7695
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8759335118953844e-06,
+      "loss": 1.0001,
+      "step": 7696
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8751733102253034e-06,
+      "loss": 1.0213,
+      "step": 7697
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.874413108555223e-06,
+      "loss": 0.8138,
+      "step": 7698
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.873652906885143e-06,
+      "loss": 0.9102,
+      "step": 7699
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.872892705215063e-06,
+      "loss": 1.0202,
+      "step": 7700
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.872132503544983e-06,
+      "loss": 0.7911,
+      "step": 7701
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.871372301874901e-06,
+      "loss": 0.8463,
+      "step": 7702
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.870612100204821e-06,
+      "loss": 0.7745,
+      "step": 7703
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.869851898534741e-06,
+      "loss": 0.8943,
+      "step": 7704
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.869091696864661e-06,
+      "loss": 1.0201,
+      "step": 7705
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8683314951945805e-06,
+      "loss": 0.744,
+      "step": 7706
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8675712935245e-06,
+      "loss": 0.8042,
+      "step": 7707
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.866811091854419e-06,
+      "loss": 0.8989,
+      "step": 7708
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.866050890184339e-06,
+      "loss": 0.884,
+      "step": 7709
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.865290688514259e-06,
+      "loss": 0.8163,
+      "step": 7710
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.864530486844179e-06,
+      "loss": 0.9982,
+      "step": 7711
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.863770285174099e-06,
+      "loss": 0.9315,
+      "step": 7712
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.863010083504018e-06,
+      "loss": 0.9638,
+      "step": 7713
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.862249881833938e-06,
+      "loss": 0.7913,
+      "step": 7714
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8614896801638575e-06,
+      "loss": 0.6782,
+      "step": 7715
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.860729478493777e-06,
+      "loss": 0.876,
+      "step": 7716
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.859969276823697e-06,
+      "loss": 0.8718,
+      "step": 7717
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.859209075153616e-06,
+      "loss": 0.7011,
+      "step": 7718
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.858448873483536e-06,
+      "loss": 0.8623,
+      "step": 7719
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.857688671813456e-06,
+      "loss": 0.9275,
+      "step": 7720
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.856928470143376e-06,
+      "loss": 0.8366,
+      "step": 7721
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.856168268473295e-06,
+      "loss": 0.9971,
+      "step": 7722
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.855408066803215e-06,
+      "loss": 0.9811,
+      "step": 7723
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.854647865133134e-06,
+      "loss": 1.0216,
+      "step": 7724
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.8538876634630535e-06,
+      "loss": 0.8638,
+      "step": 7725
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.853127461792973e-06,
+      "loss": 0.8851,
+      "step": 7726
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.852367260122893e-06,
+      "loss": 0.8202,
+      "step": 7727
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.851607058452813e-06,
+      "loss": 0.8566,
+      "step": 7728
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.850846856782732e-06,
+      "loss": 1.1411,
+      "step": 7729
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.850086655112652e-06,
+      "loss": 0.8106,
+      "step": 7730
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.849326453442572e-06,
+      "loss": 0.8719,
+      "step": 7731
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.848566251772492e-06,
+      "loss": 0.9167,
+      "step": 7732
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.8478060501024115e-06,
+      "loss": 0.6762,
+      "step": 7733
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.8470458484323305e-06,
+      "loss": 0.7951,
+      "step": 7734
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.84628564676225e-06,
+      "loss": 0.9481,
+      "step": 7735
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.84552544509217e-06,
+      "loss": 0.8747,
+      "step": 7736
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.84476524342209e-06,
+      "loss": 0.8178,
+      "step": 7737
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.84400504175201e-06,
+      "loss": 0.8511,
+      "step": 7738
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.84324484008193e-06,
+      "loss": 0.7814,
+      "step": 7739
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.842484638411848e-06,
+      "loss": 0.8204,
+      "step": 7740
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.841724436741768e-06,
+      "loss": 0.878,
+      "step": 7741
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.840964235071688e-06,
+      "loss": 0.8814,
+      "step": 7742
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.8402040334016075e-06,
+      "loss": 0.6918,
+      "step": 7743
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.839443831731527e-06,
+      "loss": 0.7977,
+      "step": 7744
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.838683630061446e-06,
+      "loss": 0.8249,
+      "step": 7745
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.837923428391366e-06,
+      "loss": 0.9669,
+      "step": 7746
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.837163226721286e-06,
+      "loss": 1.024,
+      "step": 7747
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.836403025051206e-06,
+      "loss": 0.9292,
+      "step": 7748
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.835642823381126e-06,
+      "loss": 0.7475,
+      "step": 7749
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.834882621711045e-06,
+      "loss": 0.8528,
+      "step": 7750
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.834122420040965e-06,
+      "loss": 0.7969,
+      "step": 7751
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.8333622183708845e-06,
+      "loss": 0.927,
+      "step": 7752
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.832602016700804e-06,
+      "loss": 0.8355,
+      "step": 7753
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.831841815030724e-06,
+      "loss": 0.7881,
+      "step": 7754
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.831081613360644e-06,
+      "loss": 0.8257,
+      "step": 7755
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.830321411690562e-06,
+      "loss": 0.8116,
+      "step": 7756
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.829561210020482e-06,
+      "loss": 0.9996,
+      "step": 7757
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.828801008350402e-06,
+      "loss": 0.8719,
+      "step": 7758
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.828040806680322e-06,
+      "loss": 0.7976,
+      "step": 7759
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.827280605010242e-06,
+      "loss": 0.6358,
+      "step": 7760
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.826520403340161e-06,
+      "loss": 0.8651,
+      "step": 7761
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.8257602016700805e-06,
+      "loss": 0.972,
+      "step": 7762
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.825e-06,
+      "loss": 0.8792,
+      "step": 7763
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.82423979832992e-06,
+      "loss": 0.87,
+      "step": 7764
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.823479596659839e-06,
+      "loss": 0.8767,
+      "step": 7765
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.822719394989759e-06,
+      "loss": 0.8434,
+      "step": 7766
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.821959193319679e-06,
+      "loss": 0.7945,
+      "step": 7767
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.821198991649599e-06,
+      "loss": 1.0979,
+      "step": 7768
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.820438789979519e-06,
+      "loss": 0.8677,
+      "step": 7769
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.819678588309438e-06,
+      "loss": 0.7148,
+      "step": 7770
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.8189183866393575e-06,
+      "loss": 0.9075,
+      "step": 7771
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.818158184969277e-06,
+      "loss": 0.9621,
+      "step": 7772
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.817397983299196e-06,
+      "loss": 0.797,
+      "step": 7773
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.816637781629116e-06,
+      "loss": 0.9085,
+      "step": 7774
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.815877579959036e-06,
+      "loss": 0.8681,
+      "step": 7775
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.815117378288956e-06,
+      "loss": 0.8717,
+      "step": 7776
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.814357176618876e-06,
+      "loss": 0.6828,
+      "step": 7777
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.813596974948795e-06,
+      "loss": 0.7433,
+      "step": 7778
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.812836773278715e-06,
+      "loss": 0.8495,
+      "step": 7779
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8120765716086346e-06,
+      "loss": 0.9635,
+      "step": 7780
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8113163699385536e-06,
+      "loss": 0.6043,
+      "step": 7781
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.810556168268473e-06,
+      "loss": 0.9017,
+      "step": 7782
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.809795966598393e-06,
+      "loss": 0.8116,
+      "step": 7783
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.809035764928313e-06,
+      "loss": 0.8275,
+      "step": 7784
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.808275563258233e-06,
+      "loss": 0.8111,
+      "step": 7785
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.807515361588152e-06,
+      "loss": 0.5963,
+      "step": 7786
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.806755159918072e-06,
+      "loss": 0.8062,
+      "step": 7787
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.805994958247992e-06,
+      "loss": 0.9461,
+      "step": 7788
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.805234756577911e-06,
+      "loss": 0.9085,
+      "step": 7789
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8044745549078306e-06,
+      "loss": 0.6808,
+      "step": 7790
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8037143532377504e-06,
+      "loss": 0.7728,
+      "step": 7791
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.80295415156767e-06,
+      "loss": 0.8149,
+      "step": 7792
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.80219394989759e-06,
+      "loss": 0.6784,
+      "step": 7793
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.801433748227509e-06,
+      "loss": 0.8815,
+      "step": 7794
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.800673546557429e-06,
+      "loss": 0.8023,
+      "step": 7795
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.799913344887349e-06,
+      "loss": 0.9492,
+      "step": 7796
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.799153143217268e-06,
+      "loss": 0.8247,
+      "step": 7797
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.798392941547188e-06,
+      "loss": 0.8823,
+      "step": 7798
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.797632739877108e-06,
+      "loss": 0.9506,
+      "step": 7799
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.7968725382070274e-06,
+      "loss": 0.8235,
+      "step": 7800
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.796112336536947e-06,
+      "loss": 0.9644,
+      "step": 7801
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.795352134866866e-06,
+      "loss": 0.7997,
+      "step": 7802
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.794591933196786e-06,
+      "loss": 0.8119,
+      "step": 7803
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.793831731526706e-06,
+      "loss": 0.9689,
+      "step": 7804
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.793071529856625e-06,
+      "loss": 0.9015,
+      "step": 7805
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.792311328186545e-06,
+      "loss": 0.8809,
+      "step": 7806
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.791551126516465e-06,
+      "loss": 0.9073,
+      "step": 7807
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.790790924846385e-06,
+      "loss": 1.1387,
+      "step": 7808
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.7900307231763044e-06,
+      "loss": 1.0129,
+      "step": 7809
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.7892705215062235e-06,
+      "loss": 0.8944,
+      "step": 7810
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.788510319836143e-06,
+      "loss": 0.7564,
+      "step": 7811
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.787750118166063e-06,
+      "loss": 0.8621,
+      "step": 7812
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.786989916495983e-06,
+      "loss": 0.8115,
+      "step": 7813
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.786229714825903e-06,
+      "loss": 0.8505,
+      "step": 7814
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.785469513155822e-06,
+      "loss": 0.9315,
+      "step": 7815
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.784709311485742e-06,
+      "loss": 0.967,
+      "step": 7816
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.783949109815662e-06,
+      "loss": 0.7382,
+      "step": 7817
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.783188908145581e-06,
+      "loss": 0.8752,
+      "step": 7818
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.7824287064755005e-06,
+      "loss": 0.8615,
+      "step": 7819
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.78166850480542e-06,
+      "loss": 0.7991,
+      "step": 7820
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.78090830313534e-06,
+      "loss": 1.0014,
+      "step": 7821
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.78014810146526e-06,
+      "loss": 0.8738,
+      "step": 7822
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.779387899795179e-06,
+      "loss": 0.8371,
+      "step": 7823
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.778627698125099e-06,
+      "loss": 1.0019,
+      "step": 7824
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.777867496455019e-06,
+      "loss": 0.876,
+      "step": 7825
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.777107294784938e-06,
+      "loss": 0.8039,
+      "step": 7826
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.776347093114858e-06,
+      "loss": 0.8985,
+      "step": 7827
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.7755868914447775e-06,
+      "loss": 0.8354,
+      "step": 7828
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.774826689774697e-06,
+      "loss": 0.6834,
+      "step": 7829
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.774066488104617e-06,
+      "loss": 0.7195,
+      "step": 7830
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.773306286434536e-06,
+      "loss": 1.0118,
+      "step": 7831
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.772546084764456e-06,
+      "loss": 0.7281,
+      "step": 7832
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.771785883094376e-06,
+      "loss": 0.8809,
+      "step": 7833
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.771025681424295e-06,
+      "loss": 0.9589,
+      "step": 7834
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.770265479754215e-06,
+      "loss": 0.9732,
+      "step": 7835
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.769505278084135e-06,
+      "loss": 0.7967,
+      "step": 7836
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.7687450764140545e-06,
+      "loss": 0.7309,
+      "step": 7837
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.767984874743974e-06,
+      "loss": 0.966,
+      "step": 7838
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.767224673073893e-06,
+      "loss": 0.9196,
+      "step": 7839
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.766464471403813e-06,
+      "loss": 0.8249,
+      "step": 7840
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.765704269733733e-06,
+      "loss": 0.8078,
+      "step": 7841
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.764944068063652e-06,
+      "loss": 0.973,
+      "step": 7842
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.764183866393572e-06,
+      "loss": 0.7221,
+      "step": 7843
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.763423664723492e-06,
+      "loss": 0.6824,
+      "step": 7844
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.762663463053412e-06,
+      "loss": 0.9088,
+      "step": 7845
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.7619032613833315e-06,
+      "loss": 0.8472,
+      "step": 7846
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.7611430597132505e-06,
+      "loss": 0.9726,
+      "step": 7847
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.76038285804317e-06,
+      "loss": 0.8009,
+      "step": 7848
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.75962265637309e-06,
+      "loss": 0.8553,
+      "step": 7849
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.758862454703009e-06,
+      "loss": 0.838,
+      "step": 7850
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.758102253032929e-06,
+      "loss": 0.9216,
+      "step": 7851
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.757342051362849e-06,
+      "loss": 0.8468,
+      "step": 7852
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.756581849692769e-06,
+      "loss": 0.7748,
+      "step": 7853
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.755821648022689e-06,
+      "loss": 0.8037,
+      "step": 7854
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.755061446352608e-06,
+      "loss": 0.7741,
+      "step": 7855
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.7543012446825275e-06,
+      "loss": 0.7363,
+      "step": 7856
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.753541043012447e-06,
+      "loss": 0.8118,
+      "step": 7857
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.752780841342366e-06,
+      "loss": 0.9283,
+      "step": 7858
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.752020639672286e-06,
+      "loss": 0.7826,
+      "step": 7859
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.751260438002206e-06,
+      "loss": 0.7619,
+      "step": 7860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.750500236332126e-06,
+      "loss": 1.0163,
+      "step": 7861
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.749740034662046e-06,
+      "loss": 0.9005,
+      "step": 7862
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.748979832991965e-06,
+      "loss": 0.9617,
+      "step": 7863
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.748219631321885e-06,
+      "loss": 0.8198,
+      "step": 7864
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.7474594296518045e-06,
+      "loss": 1.0015,
+      "step": 7865
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.7466992279817235e-06,
+      "loss": 1.0787,
+      "step": 7866
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.745939026311643e-06,
+      "loss": 0.9167,
+      "step": 7867
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.745178824641563e-06,
+      "loss": 0.8219,
+      "step": 7868
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.744418622971483e-06,
+      "loss": 0.7937,
+      "step": 7869
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.743658421301403e-06,
+      "loss": 0.9771,
+      "step": 7870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.742898219631322e-06,
+      "loss": 0.8841,
+      "step": 7871
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.742138017961242e-06,
+      "loss": 0.799,
+      "step": 7872
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.741377816291162e-06,
+      "loss": 0.7304,
+      "step": 7873
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.740617614621081e-06,
+      "loss": 0.7833,
+      "step": 7874
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.7398574129510005e-06,
+      "loss": 0.8565,
+      "step": 7875
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.73909721128092e-06,
+      "loss": 0.9789,
+      "step": 7876
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.73833700961084e-06,
+      "loss": 0.8072,
+      "step": 7877
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.73757680794076e-06,
+      "loss": 0.8629,
+      "step": 7878
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.736816606270679e-06,
+      "loss": 0.8207,
+      "step": 7879
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.736056404600599e-06,
+      "loss": 1.0422,
+      "step": 7880
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.735296202930519e-06,
+      "loss": 0.7772,
+      "step": 7881
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.734536001260438e-06,
+      "loss": 0.7442,
+      "step": 7882
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.733775799590358e-06,
+      "loss": 1.07,
+      "step": 7883
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.7330155979202776e-06,
+      "loss": 0.8082,
+      "step": 7884
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.732255396250197e-06,
+      "loss": 0.7782,
+      "step": 7885
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.731495194580117e-06,
+      "loss": 0.7869,
+      "step": 7886
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.730734992910036e-06,
+      "loss": 0.7449,
+      "step": 7887
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.729974791239956e-06,
+      "loss": 0.7581,
+      "step": 7888
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.729214589569876e-06,
+      "loss": 0.7076,
+      "step": 7889
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.728454387899795e-06,
+      "loss": 0.9416,
+      "step": 7890
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.727694186229715e-06,
+      "loss": 0.7289,
+      "step": 7891
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.726933984559635e-06,
+      "loss": 0.7595,
+      "step": 7892
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.7261737828895546e-06,
+      "loss": 0.92,
+      "step": 7893
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.725413581219474e-06,
+      "loss": 0.9746,
+      "step": 7894
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.7246533795493934e-06,
+      "loss": 0.8416,
+      "step": 7895
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.723893177879313e-06,
+      "loss": 0.8295,
+      "step": 7896
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.723132976209233e-06,
+      "loss": 0.6604,
+      "step": 7897
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.722372774539153e-06,
+      "loss": 0.7632,
+      "step": 7898
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.721612572869072e-06,
+      "loss": 0.988,
+      "step": 7899
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.720852371198992e-06,
+      "loss": 0.9664,
+      "step": 7900
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.720092169528912e-06,
+      "loss": 0.9576,
+      "step": 7901
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.7193319678588316e-06,
+      "loss": 0.8448,
+      "step": 7902
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.718571766188751e-06,
+      "loss": 0.8618,
+      "step": 7903
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.7178115645186704e-06,
+      "loss": 0.892,
+      "step": 7904
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.71705136284859e-06,
+      "loss": 0.9928,
+      "step": 7905
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.71629116117851e-06,
+      "loss": 0.9334,
+      "step": 7906
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.71553095950843e-06,
+      "loss": 0.8719,
+      "step": 7907
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.71477075783835e-06,
+      "loss": 0.9026,
+      "step": 7908
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.714010556168269e-06,
+      "loss": 0.9807,
+      "step": 7909
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.713250354498189e-06,
+      "loss": 0.7141,
+      "step": 7910
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.712490152828108e-06,
+      "loss": 0.8135,
+      "step": 7911
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.711729951158028e-06,
+      "loss": 0.7604,
+      "step": 7912
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.7109697494879475e-06,
+      "loss": 0.969,
+      "step": 7913
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.710209547817867e-06,
+      "loss": 0.8369,
+      "step": 7914
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.709449346147787e-06,
+      "loss": 0.7318,
+      "step": 7915
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.708689144477707e-06,
+      "loss": 0.9268,
+      "step": 7916
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.707928942807626e-06,
+      "loss": 0.8556,
+      "step": 7917
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.707168741137546e-06,
+      "loss": 0.7998,
+      "step": 7918
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.706408539467465e-06,
+      "loss": 0.9284,
+      "step": 7919
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.705648337797385e-06,
+      "loss": 0.9497,
+      "step": 7920
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.704888136127305e-06,
+      "loss": 1.019,
+      "step": 7921
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.7041279344572245e-06,
+      "loss": 0.7928,
+      "step": 7922
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.703367732787144e-06,
+      "loss": 1.0376,
+      "step": 7923
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.702607531117064e-06,
+      "loss": 0.7003,
+      "step": 7924
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.701847329446983e-06,
+      "loss": 0.9175,
+      "step": 7925
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.701087127776903e-06,
+      "loss": 0.8231,
+      "step": 7926
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.700326926106822e-06,
+      "loss": 0.7024,
+      "step": 7927
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.699566724436742e-06,
+      "loss": 0.8682,
+      "step": 7928
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.698806522766662e-06,
+      "loss": 0.8832,
+      "step": 7929
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.698046321096582e-06,
+      "loss": 0.8288,
+      "step": 7930
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.6972861194265015e-06,
+      "loss": 0.7652,
+      "step": 7931
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.696525917756421e-06,
+      "loss": 0.6844,
+      "step": 7932
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.69576571608634e-06,
+      "loss": 1.0774,
+      "step": 7933
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.69500551441626e-06,
+      "loss": 0.7793,
+      "step": 7934
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.694245312746179e-06,
+      "loss": 0.8763,
+      "step": 7935
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.693485111076099e-06,
+      "loss": 0.9255,
+      "step": 7936
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.692724909406019e-06,
+      "loss": 0.7017,
+      "step": 7937
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.691964707735939e-06,
+      "loss": 0.9466,
+      "step": 7938
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.691204506065859e-06,
+      "loss": 0.7282,
+      "step": 7939
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.6904443043957785e-06,
+      "loss": 0.8013,
+      "step": 7940
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.6896841027256975e-06,
+      "loss": 0.7394,
+      "step": 7941
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.688923901055617e-06,
+      "loss": 0.8524,
+      "step": 7942
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.688163699385536e-06,
+      "loss": 0.7871,
+      "step": 7943
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.687403497715456e-06,
+      "loss": 0.9311,
+      "step": 7944
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.686643296045376e-06,
+      "loss": 1.0336,
+      "step": 7945
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.685883094375296e-06,
+      "loss": 0.9005,
+      "step": 7946
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.685122892705216e-06,
+      "loss": 0.7626,
+      "step": 7947
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.684362691035136e-06,
+      "loss": 0.7817,
+      "step": 7948
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.683602489365055e-06,
+      "loss": 0.9964,
+      "step": 7949
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.6828422876949745e-06,
+      "loss": 0.7914,
+      "step": 7950
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.6820820860248935e-06,
+      "loss": 0.8223,
+      "step": 7951
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.681321884354813e-06,
+      "loss": 0.8739,
+      "step": 7952
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.680561682684733e-06,
+      "loss": 0.7927,
+      "step": 7953
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.679801481014653e-06,
+      "loss": 0.8906,
+      "step": 7954
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.679041279344573e-06,
+      "loss": 0.9578,
+      "step": 7955
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.678281077674493e-06,
+      "loss": 0.7782,
+      "step": 7956
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.677520876004412e-06,
+      "loss": 0.8128,
+      "step": 7957
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.676760674334332e-06,
+      "loss": 0.881,
+      "step": 7958
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.676000472664251e-06,
+      "loss": 0.9332,
+      "step": 7959
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.6752402709941705e-06,
+      "loss": 0.8142,
+      "step": 7960
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.67448006932409e-06,
+      "loss": 0.9688,
+      "step": 7961
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.67371986765401e-06,
+      "loss": 0.8661,
+      "step": 7962
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.67295966598393e-06,
+      "loss": 0.6413,
+      "step": 7963
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.67219946431385e-06,
+      "loss": 0.7926,
+      "step": 7964
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.671439262643769e-06,
+      "loss": 0.8287,
+      "step": 7965
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.670679060973689e-06,
+      "loss": 0.9631,
+      "step": 7966
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.669918859303608e-06,
+      "loss": 0.8444,
+      "step": 7967
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.669158657633528e-06,
+      "loss": 0.7427,
+      "step": 7968
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.6683984559634475e-06,
+      "loss": 0.8726,
+      "step": 7969
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.667638254293367e-06,
+      "loss": 0.8352,
+      "step": 7970
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.666878052623287e-06,
+      "loss": 0.6936,
+      "step": 7971
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.666117850953206e-06,
+      "loss": 0.8994,
+      "step": 7972
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.665357649283126e-06,
+      "loss": 0.7059,
+      "step": 7973
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.664597447613046e-06,
+      "loss": 0.7761,
+      "step": 7974
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.663837245942965e-06,
+      "loss": 0.8202,
+      "step": 7975
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.663077044272885e-06,
+      "loss": 0.7989,
+      "step": 7976
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.662316842602805e-06,
+      "loss": 0.9142,
+      "step": 7977
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.6615566409327245e-06,
+      "loss": 0.6514,
+      "step": 7978
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.660796439262644e-06,
+      "loss": 0.7764,
+      "step": 7979
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.660036237592563e-06,
+      "loss": 0.9271,
+      "step": 7980
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.659276035922483e-06,
+      "loss": 0.9773,
+      "step": 7981
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.658515834252403e-06,
+      "loss": 0.8183,
+      "step": 7982
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.657755632582322e-06,
+      "loss": 0.7965,
+      "step": 7983
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.656995430912242e-06,
+      "loss": 0.8455,
+      "step": 7984
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.656235229242162e-06,
+      "loss": 0.7598,
+      "step": 7985
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.655475027572082e-06,
+      "loss": 0.8422,
+      "step": 7986
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.6547148259020015e-06,
+      "loss": 0.7865,
+      "step": 7987
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.6539546242319206e-06,
+      "loss": 0.7612,
+      "step": 7988
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.65319442256184e-06,
+      "loss": 0.6204,
+      "step": 7989
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.65243422089176e-06,
+      "loss": 0.9303,
+      "step": 7990
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.65167401922168e-06,
+      "loss": 0.7732,
+      "step": 7991
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.6509138175516e-06,
+      "loss": 0.7493,
+      "step": 7992
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.650153615881519e-06,
+      "loss": 0.8477,
+      "step": 7993
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.649393414211439e-06,
+      "loss": 0.9806,
+      "step": 7994
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.648633212541359e-06,
+      "loss": 0.7832,
+      "step": 7995
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.647873010871278e-06,
+      "loss": 1.033,
+      "step": 7996
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.6471128092011976e-06,
+      "loss": 0.7274,
+      "step": 7997
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.6463526075311174e-06,
+      "loss": 0.9519,
+      "step": 7998
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.645592405861037e-06,
+      "loss": 0.7064,
+      "step": 7999
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.644832204190957e-06,
+      "loss": 0.8407,
+      "step": 8000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.644072002520877e-06,
+      "loss": 0.8235,
+      "step": 8001
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.643311800850796e-06,
+      "loss": 0.7929,
+      "step": 8002
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.642551599180716e-06,
+      "loss": 0.893,
+      "step": 8003
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.641791397510635e-06,
+      "loss": 0.844,
+      "step": 8004
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.641031195840555e-06,
+      "loss": 0.9555,
+      "step": 8005
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.640270994170475e-06,
+      "loss": 0.8085,
+      "step": 8006
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.6395107925003944e-06,
+      "loss": 0.9834,
+      "step": 8007
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.638750590830314e-06,
+      "loss": 0.9872,
+      "step": 8008
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.637990389160234e-06,
+      "loss": 0.7507,
+      "step": 8009
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.637230187490153e-06,
+      "loss": 0.8262,
+      "step": 8010
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.636469985820073e-06,
+      "loss": 0.8365,
+      "step": 8011
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.635709784149992e-06,
+      "loss": 0.9074,
+      "step": 8012
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.634949582479912e-06,
+      "loss": 0.7533,
+      "step": 8013
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.634189380809832e-06,
+      "loss": 0.7926,
+      "step": 8014
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.633429179139752e-06,
+      "loss": 0.9283,
+      "step": 8015
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.6326689774696714e-06,
+      "loss": 0.9612,
+      "step": 8016
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.631908775799591e-06,
+      "loss": 0.7289,
+      "step": 8017
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.63114857412951e-06,
+      "loss": 0.8463,
+      "step": 8018
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.63038837245943e-06,
+      "loss": 1.01,
+      "step": 8019
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.629628170789349e-06,
+      "loss": 0.8213,
+      "step": 8020
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.628867969119269e-06,
+      "loss": 0.8207,
+      "step": 8021
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.628107767449189e-06,
+      "loss": 0.9527,
+      "step": 8022
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.627347565779109e-06,
+      "loss": 0.9246,
+      "step": 8023
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.626587364109029e-06,
+      "loss": 0.8349,
+      "step": 8024
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.6258271624389485e-06,
+      "loss": 0.7517,
+      "step": 8025
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.6250669607688675e-06,
+      "loss": 0.8625,
+      "step": 8026
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.624306759098787e-06,
+      "loss": 0.8409,
+      "step": 8027
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.623546557428706e-06,
+      "loss": 1.0091,
+      "step": 8028
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.622786355758626e-06,
+      "loss": 1.0481,
+      "step": 8029
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.622026154088546e-06,
+      "loss": 0.7918,
+      "step": 8030
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.621265952418466e-06,
+      "loss": 0.8664,
+      "step": 8031
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.620505750748386e-06,
+      "loss": 0.8111,
+      "step": 8032
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.619745549078306e-06,
+      "loss": 0.8687,
+      "step": 8033
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.618985347408225e-06,
+      "loss": 1.0184,
+      "step": 8034
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.6182251457381445e-06,
+      "loss": 0.9349,
+      "step": 8035
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.6174649440680635e-06,
+      "loss": 0.9933,
+      "step": 8036
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.616704742397983e-06,
+      "loss": 0.8609,
+      "step": 8037
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.615944540727903e-06,
+      "loss": 0.8439,
+      "step": 8038
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.615184339057823e-06,
+      "loss": 0.8326,
+      "step": 8039
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.614424137387743e-06,
+      "loss": 0.8607,
+      "step": 8040
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.613663935717663e-06,
+      "loss": 0.7762,
+      "step": 8041
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.612903734047582e-06,
+      "loss": 0.7935,
+      "step": 8042
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.612143532377502e-06,
+      "loss": 0.8541,
+      "step": 8043
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.611383330707421e-06,
+      "loss": 0.7766,
+      "step": 8044
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.6106231290373405e-06,
+      "loss": 0.8227,
+      "step": 8045
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.60986292736726e-06,
+      "loss": 0.8815,
+      "step": 8046
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.60910272569718e-06,
+      "loss": 0.7603,
+      "step": 8047
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.6083425240271e-06,
+      "loss": 0.7511,
+      "step": 8048
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.60758232235702e-06,
+      "loss": 0.7922,
+      "step": 8049
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.606822120686939e-06,
+      "loss": 0.6877,
+      "step": 8050
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.606061919016859e-06,
+      "loss": 0.8912,
+      "step": 8051
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.605301717346778e-06,
+      "loss": 0.8454,
+      "step": 8052
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.604541515676698e-06,
+      "loss": 1.0367,
+      "step": 8053
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.6037813140066175e-06,
+      "loss": 0.832,
+      "step": 8054
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.603021112336537e-06,
+      "loss": 0.8397,
+      "step": 8055
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.602260910666457e-06,
+      "loss": 0.7839,
+      "step": 8056
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.601500708996377e-06,
+      "loss": 1.1081,
+      "step": 8057
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.600740507326296e-06,
+      "loss": 0.9391,
+      "step": 8058
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.599980305656216e-06,
+      "loss": 0.7302,
+      "step": 8059
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.599220103986135e-06,
+      "loss": 0.9646,
+      "step": 8060
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.598459902316055e-06,
+      "loss": 0.8611,
+      "step": 8061
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.597699700645975e-06,
+      "loss": 0.8704,
+      "step": 8062
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5969394989758945e-06,
+      "loss": 0.9751,
+      "step": 8063
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.596179297305814e-06,
+      "loss": 1.0047,
+      "step": 8064
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.595419095635734e-06,
+      "loss": 1.0204,
+      "step": 8065
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.594658893965653e-06,
+      "loss": 0.9432,
+      "step": 8066
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.593898692295573e-06,
+      "loss": 0.9441,
+      "step": 8067
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.593138490625492e-06,
+      "loss": 0.7979,
+      "step": 8068
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.592378288955412e-06,
+      "loss": 0.792,
+      "step": 8069
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.591618087285332e-06,
+      "loss": 0.9586,
+      "step": 8070
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.590857885615252e-06,
+      "loss": 0.8013,
+      "step": 8071
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5900976839451715e-06,
+      "loss": 0.6579,
+      "step": 8072
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.589337482275091e-06,
+      "loss": 0.9246,
+      "step": 8073
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.58857728060501e-06,
+      "loss": 0.8503,
+      "step": 8074
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.58781707893493e-06,
+      "loss": 0.9345,
+      "step": 8075
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.587056877264849e-06,
+      "loss": 0.8708,
+      "step": 8076
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.586296675594769e-06,
+      "loss": 0.6611,
+      "step": 8077
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.585536473924689e-06,
+      "loss": 0.5322,
+      "step": 8078
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.584776272254609e-06,
+      "loss": 0.8456,
+      "step": 8079
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.584016070584529e-06,
+      "loss": 0.8144,
+      "step": 8080
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5832558689144485e-06,
+      "loss": 0.8691,
+      "step": 8081
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5824956672443675e-06,
+      "loss": 0.8246,
+      "step": 8082
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.581735465574287e-06,
+      "loss": 0.9063,
+      "step": 8083
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.580975263904207e-06,
+      "loss": 0.7957,
+      "step": 8084
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.580215062234127e-06,
+      "loss": 0.6993,
+      "step": 8085
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.579454860564047e-06,
+      "loss": 0.8508,
+      "step": 8086
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.578694658893966e-06,
+      "loss": 0.8051,
+      "step": 8087
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.577934457223886e-06,
+      "loss": 0.8843,
+      "step": 8088
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.577174255553806e-06,
+      "loss": 0.9655,
+      "step": 8089
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.576414053883725e-06,
+      "loss": 0.8353,
+      "step": 8090
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.5756538522136446e-06,
+      "loss": 0.8045,
+      "step": 8091
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.574893650543564e-06,
+      "loss": 0.9215,
+      "step": 8092
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.574133448873484e-06,
+      "loss": 0.8422,
+      "step": 8093
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.573373247203404e-06,
+      "loss": 0.8395,
+      "step": 8094
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.572613045533323e-06,
+      "loss": 0.923,
+      "step": 8095
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.571852843863243e-06,
+      "loss": 0.7058,
+      "step": 8096
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.571092642193163e-06,
+      "loss": 0.8874,
+      "step": 8097
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.570332440523082e-06,
+      "loss": 0.884,
+      "step": 8098
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.569572238853002e-06,
+      "loss": 0.7437,
+      "step": 8099
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.5688120371829216e-06,
+      "loss": 1.008,
+      "step": 8100
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.568051835512841e-06,
+      "loss": 0.6702,
+      "step": 8101
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.567291633842761e-06,
+      "loss": 0.8432,
+      "step": 8102
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.56653143217268e-06,
+      "loss": 0.8716,
+      "step": 8103
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.5657712305026e-06,
+      "loss": 0.8926,
+      "step": 8104
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.56501102883252e-06,
+      "loss": 0.6575,
+      "step": 8105
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.564250827162439e-06,
+      "loss": 0.9113,
+      "step": 8106
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.563490625492359e-06,
+      "loss": 0.7302,
+      "step": 8107
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.562730423822279e-06,
+      "loss": 0.9871,
+      "step": 8108
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.5619702221521986e-06,
+      "loss": 0.7812,
+      "step": 8109
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.5612100204821184e-06,
+      "loss": 0.9668,
+      "step": 8110
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5604498188120374e-06,
+      "loss": 0.7557,
+      "step": 8111
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.559689617141957e-06,
+      "loss": 0.812,
+      "step": 8112
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.558929415471877e-06,
+      "loss": 0.8459,
+      "step": 8113
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.558169213801796e-06,
+      "loss": 0.7134,
+      "step": 8114
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.557409012131716e-06,
+      "loss": 1.0438,
+      "step": 8115
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.556648810461636e-06,
+      "loss": 0.9327,
+      "step": 8116
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.555888608791556e-06,
+      "loss": 0.883,
+      "step": 8117
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.555128407121476e-06,
+      "loss": 0.8882,
+      "step": 8118
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.554368205451395e-06,
+      "loss": 0.7883,
+      "step": 8119
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5536080037813144e-06,
+      "loss": 0.915,
+      "step": 8120
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.552847802111234e-06,
+      "loss": 0.5709,
+      "step": 8121
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.552087600441153e-06,
+      "loss": 0.8534,
+      "step": 8122
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.551327398771073e-06,
+      "loss": 0.8247,
+      "step": 8123
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.550567197100993e-06,
+      "loss": 0.6704,
+      "step": 8124
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.549806995430913e-06,
+      "loss": 0.8525,
+      "step": 8125
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.549046793760833e-06,
+      "loss": 0.8915,
+      "step": 8126
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.548286592090752e-06,
+      "loss": 0.7829,
+      "step": 8127
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.547526390420672e-06,
+      "loss": 0.7173,
+      "step": 8128
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5467661887505915e-06,
+      "loss": 0.8783,
+      "step": 8129
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5460059870805105e-06,
+      "loss": 0.7679,
+      "step": 8130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.54524578541043e-06,
+      "loss": 0.9584,
+      "step": 8131
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.54448558374035e-06,
+      "loss": 0.8918,
+      "step": 8132
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.54372538207027e-06,
+      "loss": 0.8312,
+      "step": 8133
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.54296518040019e-06,
+      "loss": 0.8985,
+      "step": 8134
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.542204978730109e-06,
+      "loss": 0.9007,
+      "step": 8135
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.541444777060029e-06,
+      "loss": 0.8181,
+      "step": 8136
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.540684575389949e-06,
+      "loss": 0.8898,
+      "step": 8137
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.539924373719868e-06,
+      "loss": 0.788,
+      "step": 8138
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.5391641720497875e-06,
+      "loss": 0.7528,
+      "step": 8139
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.538403970379707e-06,
+      "loss": 0.7769,
+      "step": 8140
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.537643768709627e-06,
+      "loss": 0.6505,
+      "step": 8141
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.536883567039547e-06,
+      "loss": 0.9249,
+      "step": 8142
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.536123365369466e-06,
+      "loss": 0.6577,
+      "step": 8143
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.535363163699386e-06,
+      "loss": 0.7654,
+      "step": 8144
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.534602962029306e-06,
+      "loss": 0.9716,
+      "step": 8145
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.533842760359225e-06,
+      "loss": 0.7997,
+      "step": 8146
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.533082558689145e-06,
+      "loss": 0.7842,
+      "step": 8147
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.5323223570190645e-06,
+      "loss": 0.7248,
+      "step": 8148
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.531562155348984e-06,
+      "loss": 0.817,
+      "step": 8149
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.530801953678904e-06,
+      "loss": 0.7984,
+      "step": 8150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.530041752008823e-06,
+      "loss": 0.815,
+      "step": 8151
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.529281550338743e-06,
+      "loss": 0.8552,
+      "step": 8152
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.528521348668663e-06,
+      "loss": 0.843,
+      "step": 8153
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.527761146998582e-06,
+      "loss": 0.873,
+      "step": 8154
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.527000945328502e-06,
+      "loss": 0.9318,
+      "step": 8155
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.526240743658422e-06,
+      "loss": 0.9248,
+      "step": 8156
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.5254805419883415e-06,
+      "loss": 0.8794,
+      "step": 8157
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.524720340318261e-06,
+      "loss": 0.8127,
+      "step": 8158
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.52396013864818e-06,
+      "loss": 0.6942,
+      "step": 8159
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.5231999369781e-06,
+      "loss": 0.7741,
+      "step": 8160
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.52243973530802e-06,
+      "loss": 0.8918,
+      "step": 8161
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.521679533637939e-06,
+      "loss": 0.8507,
+      "step": 8162
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.520919331967859e-06,
+      "loss": 0.929,
+      "step": 8163
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.520159130297779e-06,
+      "loss": 0.9505,
+      "step": 8164
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.519398928627699e-06,
+      "loss": 0.6689,
+      "step": 8165
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.5186387269576185e-06,
+      "loss": 0.9576,
+      "step": 8166
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.5178785252875375e-06,
+      "loss": 1.0422,
+      "step": 8167
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.517118323617457e-06,
+      "loss": 1.0184,
+      "step": 8168
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.516358121947377e-06,
+      "loss": 0.8911,
+      "step": 8169
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.515597920277297e-06,
+      "loss": 0.8369,
+      "step": 8170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.514837718607216e-06,
+      "loss": 0.8,
+      "step": 8171
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.514077516937136e-06,
+      "loss": 0.8911,
+      "step": 8172
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.513317315267056e-06,
+      "loss": 1.0183,
+      "step": 8173
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.512557113596976e-06,
+      "loss": 0.6979,
+      "step": 8174
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.511796911926895e-06,
+      "loss": 0.8454,
+      "step": 8175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.5110367102568145e-06,
+      "loss": 0.881,
+      "step": 8176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.510276508586734e-06,
+      "loss": 0.8254,
+      "step": 8177
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.509516306916654e-06,
+      "loss": 0.7636,
+      "step": 8178
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.508756105246574e-06,
+      "loss": 0.7166,
+      "step": 8179
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.507995903576493e-06,
+      "loss": 0.7116,
+      "step": 8180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.507235701906413e-06,
+      "loss": 0.8024,
+      "step": 8181
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.506475500236333e-06,
+      "loss": 0.797,
+      "step": 8182
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.505715298566252e-06,
+      "loss": 0.7689,
+      "step": 8183
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.504955096896172e-06,
+      "loss": 0.8075,
+      "step": 8184
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.5041948952260915e-06,
+      "loss": 1.0021,
+      "step": 8185
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.503434693556011e-06,
+      "loss": 0.9675,
+      "step": 8186
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.502674491885931e-06,
+      "loss": 0.9308,
+      "step": 8187
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.50191429021585e-06,
+      "loss": 0.8205,
+      "step": 8188
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.50115408854577e-06,
+      "loss": 1.0,
+      "step": 8189
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.50039388687569e-06,
+      "loss": 0.7452,
+      "step": 8190
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.499633685205609e-06,
+      "loss": 0.7328,
+      "step": 8191
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.498873483535529e-06,
+      "loss": 0.7872,
+      "step": 8192
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.498113281865449e-06,
+      "loss": 1.0464,
+      "step": 8193
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.4973530801953685e-06,
+      "loss": 0.9446,
+      "step": 8194
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.496592878525288e-06,
+      "loss": 0.7469,
+      "step": 8195
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.495832676855207e-06,
+      "loss": 0.8312,
+      "step": 8196
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.495072475185127e-06,
+      "loss": 0.9836,
+      "step": 8197
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.494312273515047e-06,
+      "loss": 0.8258,
+      "step": 8198
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.493552071844966e-06,
+      "loss": 0.9397,
+      "step": 8199
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.492791870174886e-06,
+      "loss": 0.6894,
+      "step": 8200
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.492031668504806e-06,
+      "loss": 0.7874,
+      "step": 8201
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.491271466834726e-06,
+      "loss": 0.8433,
+      "step": 8202
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4905112651646456e-06,
+      "loss": 0.9425,
+      "step": 8203
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4897510634945646e-06,
+      "loss": 0.9614,
+      "step": 8204
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.488990861824484e-06,
+      "loss": 0.7804,
+      "step": 8205
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.488230660154404e-06,
+      "loss": 0.9422,
+      "step": 8206
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.487470458484323e-06,
+      "loss": 0.6819,
+      "step": 8207
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.486710256814243e-06,
+      "loss": 0.8362,
+      "step": 8208
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.485950055144163e-06,
+      "loss": 0.949,
+      "step": 8209
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.485189853474083e-06,
+      "loss": 0.8263,
+      "step": 8210
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.484429651804003e-06,
+      "loss": 0.8904,
+      "step": 8211
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.483669450133922e-06,
+      "loss": 0.6086,
+      "step": 8212
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4829092484638416e-06,
+      "loss": 0.8281,
+      "step": 8213
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4821490467937614e-06,
+      "loss": 0.9099,
+      "step": 8214
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4813888451236804e-06,
+      "loss": 0.8723,
+      "step": 8215
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4806286434536e-06,
+      "loss": 1.0297,
+      "step": 8216
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.47986844178352e-06,
+      "loss": 0.9793,
+      "step": 8217
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.47910824011344e-06,
+      "loss": 0.8783,
+      "step": 8218
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.47834803844336e-06,
+      "loss": 0.95,
+      "step": 8219
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.477587836773279e-06,
+      "loss": 0.8535,
+      "step": 8220
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.476827635103199e-06,
+      "loss": 0.8958,
+      "step": 8221
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.476067433433119e-06,
+      "loss": 0.8759,
+      "step": 8222
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.475307231763038e-06,
+      "loss": 0.7039,
+      "step": 8223
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.4745470300929574e-06,
+      "loss": 0.8322,
+      "step": 8224
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.473786828422877e-06,
+      "loss": 0.7043,
+      "step": 8225
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.473026626752797e-06,
+      "loss": 0.9094,
+      "step": 8226
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.472266425082717e-06,
+      "loss": 0.911,
+      "step": 8227
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.471506223412636e-06,
+      "loss": 0.6777,
+      "step": 8228
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.470746021742556e-06,
+      "loss": 0.6893,
+      "step": 8229
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.469985820072476e-06,
+      "loss": 0.9822,
+      "step": 8230
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.469225618402395e-06,
+      "loss": 0.8994,
+      "step": 8231
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.468465416732315e-06,
+      "loss": 0.8294,
+      "step": 8232
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.4677052150622345e-06,
+      "loss": 0.8475,
+      "step": 8233
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.466945013392154e-06,
+      "loss": 0.7633,
+      "step": 8234
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.466184811722074e-06,
+      "loss": 0.6861,
+      "step": 8235
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.465424610051993e-06,
+      "loss": 0.6853,
+      "step": 8236
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.464664408381913e-06,
+      "loss": 0.7681,
+      "step": 8237
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.463904206711833e-06,
+      "loss": 0.9538,
+      "step": 8238
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.463144005041752e-06,
+      "loss": 0.7018,
+      "step": 8239
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.462383803371672e-06,
+      "loss": 0.74,
+      "step": 8240
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.461623601701592e-06,
+      "loss": 0.7699,
+      "step": 8241
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.4608634000315115e-06,
+      "loss": 0.959,
+      "step": 8242
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.460103198361431e-06,
+      "loss": 0.8373,
+      "step": 8243
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.45934299669135e-06,
+      "loss": 0.8511,
+      "step": 8244
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.45858279502127e-06,
+      "loss": 0.8868,
+      "step": 8245
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.45782259335119e-06,
+      "loss": 0.8926,
+      "step": 8246
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.457062391681109e-06,
+      "loss": 0.8617,
+      "step": 8247
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.456302190011029e-06,
+      "loss": 0.8337,
+      "step": 8248
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.455541988340949e-06,
+      "loss": 0.8542,
+      "step": 8249
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.454781786670869e-06,
+      "loss": 0.7792,
+      "step": 8250
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.4540215850007885e-06,
+      "loss": 0.6843,
+      "step": 8251
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.4532613833307075e-06,
+      "loss": 0.8089,
+      "step": 8252
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.452501181660627e-06,
+      "loss": 0.8999,
+      "step": 8253
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.451740979990547e-06,
+      "loss": 0.9499,
+      "step": 8254
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.450980778320466e-06,
+      "loss": 0.8578,
+      "step": 8255
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.450220576650386e-06,
+      "loss": 0.7313,
+      "step": 8256
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.449460374980306e-06,
+      "loss": 0.7421,
+      "step": 8257
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.448700173310226e-06,
+      "loss": 0.7212,
+      "step": 8258
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.447939971640146e-06,
+      "loss": 0.7775,
+      "step": 8259
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.447179769970065e-06,
+      "loss": 0.7751,
+      "step": 8260
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.4464195682999845e-06,
+      "loss": 0.8578,
+      "step": 8261
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.445659366629904e-06,
+      "loss": 0.7833,
+      "step": 8262
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.444899164959824e-06,
+      "loss": 0.8239,
+      "step": 8263
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.444138963289744e-06,
+      "loss": 1.1921,
+      "step": 8264
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.443378761619663e-06,
+      "loss": 0.6959,
+      "step": 8265
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.442618559949583e-06,
+      "loss": 0.8469,
+      "step": 8266
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.441858358279503e-06,
+      "loss": 0.7198,
+      "step": 8267
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.441098156609422e-06,
+      "loss": 0.9059,
+      "step": 8268
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.440337954939342e-06,
+      "loss": 0.792,
+      "step": 8269
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.4395777532692615e-06,
+      "loss": 0.8218,
+      "step": 8270
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.438817551599181e-06,
+      "loss": 0.687,
+      "step": 8271
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.438057349929101e-06,
+      "loss": 0.8808,
+      "step": 8272
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.43729714825902e-06,
+      "loss": 0.7924,
+      "step": 8273
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.43653694658894e-06,
+      "loss": 0.8822,
+      "step": 8274
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.43577674491886e-06,
+      "loss": 0.8024,
+      "step": 8275
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.435016543248779e-06,
+      "loss": 0.7545,
+      "step": 8276
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.434256341578699e-06,
+      "loss": 1.0064,
+      "step": 8277
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.433496139908619e-06,
+      "loss": 0.869,
+      "step": 8278
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.4327359382385385e-06,
+      "loss": 0.7387,
+      "step": 8279
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.431975736568458e-06,
+      "loss": 0.7746,
+      "step": 8280
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.431215534898377e-06,
+      "loss": 0.7595,
+      "step": 8281
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.430455333228297e-06,
+      "loss": 0.9629,
+      "step": 8282
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.429695131558217e-06,
+      "loss": 0.969,
+      "step": 8283
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.428934929888136e-06,
+      "loss": 0.755,
+      "step": 8284
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.428174728218056e-06,
+      "loss": 0.88,
+      "step": 8285
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.427414526547976e-06,
+      "loss": 0.9514,
+      "step": 8286
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.426654324877896e-06,
+      "loss": 0.8754,
+      "step": 8287
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4258941232078155e-06,
+      "loss": 0.7117,
+      "step": 8288
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4251339215377345e-06,
+      "loss": 0.8795,
+      "step": 8289
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.424373719867654e-06,
+      "loss": 0.842,
+      "step": 8290
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.423613518197574e-06,
+      "loss": 0.8026,
+      "step": 8291
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.422853316527493e-06,
+      "loss": 0.8012,
+      "step": 8292
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.422093114857413e-06,
+      "loss": 0.9381,
+      "step": 8293
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.421332913187333e-06,
+      "loss": 0.7438,
+      "step": 8294
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.420572711517253e-06,
+      "loss": 1.0118,
+      "step": 8295
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.419812509847173e-06,
+      "loss": 0.8812,
+      "step": 8296
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.419052308177092e-06,
+      "loss": 0.8056,
+      "step": 8297
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4182921065070115e-06,
+      "loss": 0.715,
+      "step": 8298
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.417531904836931e-06,
+      "loss": 0.8979,
+      "step": 8299
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.41677170316685e-06,
+      "loss": 1.0336,
+      "step": 8300
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.41601150149677e-06,
+      "loss": 0.764,
+      "step": 8301
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.41525129982669e-06,
+      "loss": 0.8581,
+      "step": 8302
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.41449109815661e-06,
+      "loss": 0.8497,
+      "step": 8303
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.41373089648653e-06,
+      "loss": 0.7648,
+      "step": 8304
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.412970694816449e-06,
+      "loss": 0.808,
+      "step": 8305
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.412210493146369e-06,
+      "loss": 0.8239,
+      "step": 8306
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4114502914762886e-06,
+      "loss": 0.9083,
+      "step": 8307
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.4106900898062076e-06,
+      "loss": 0.765,
+      "step": 8308
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.409929888136127e-06,
+      "loss": 0.7943,
+      "step": 8309
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.409169686466047e-06,
+      "loss": 1.095,
+      "step": 8310
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.408409484795967e-06,
+      "loss": 0.8944,
+      "step": 8311
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.407649283125887e-06,
+      "loss": 0.7196,
+      "step": 8312
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.406889081455806e-06,
+      "loss": 0.8838,
+      "step": 8313
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.406128879785726e-06,
+      "loss": 0.8298,
+      "step": 8314
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.405368678115646e-06,
+      "loss": 0.8884,
+      "step": 8315
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.404608476445565e-06,
+      "loss": 1.1734,
+      "step": 8316
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.4038482747754846e-06,
+      "loss": 0.952,
+      "step": 8317
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.4030880731054044e-06,
+      "loss": 0.8789,
+      "step": 8318
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.402327871435324e-06,
+      "loss": 0.8207,
+      "step": 8319
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.401567669765244e-06,
+      "loss": 0.8531,
+      "step": 8320
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.400807468095163e-06,
+      "loss": 0.7546,
+      "step": 8321
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.400047266425083e-06,
+      "loss": 0.8495,
+      "step": 8322
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.399287064755003e-06,
+      "loss": 0.8051,
+      "step": 8323
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.398526863084922e-06,
+      "loss": 0.8333,
+      "step": 8324
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.397766661414842e-06,
+      "loss": 0.7979,
+      "step": 8325
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.397006459744762e-06,
+      "loss": 0.8719,
+      "step": 8326
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.3962462580746814e-06,
+      "loss": 0.8628,
+      "step": 8327
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.395486056404601e-06,
+      "loss": 0.9509,
+      "step": 8328
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.39472585473452e-06,
+      "loss": 0.7953,
+      "step": 8329
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.39396565306444e-06,
+      "loss": 0.9218,
+      "step": 8330
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.39320545139436e-06,
+      "loss": 0.8992,
+      "step": 8331
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.392445249724279e-06,
+      "loss": 0.7103,
+      "step": 8332
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.391685048054199e-06,
+      "loss": 0.8335,
+      "step": 8333
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.390924846384119e-06,
+      "loss": 0.7467,
+      "step": 8334
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.390164644714039e-06,
+      "loss": 0.7329,
+      "step": 8335
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.3894044430439584e-06,
+      "loss": 0.7911,
+      "step": 8336
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.3886442413738775e-06,
+      "loss": 0.9228,
+      "step": 8337
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.387884039703797e-06,
+      "loss": 0.9102,
+      "step": 8338
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.387123838033717e-06,
+      "loss": 0.8506,
+      "step": 8339
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.386363636363636e-06,
+      "loss": 0.8485,
+      "step": 8340
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.385603434693556e-06,
+      "loss": 0.8528,
+      "step": 8341
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.384843233023476e-06,
+      "loss": 0.8113,
+      "step": 8342
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.384083031353396e-06,
+      "loss": 0.8289,
+      "step": 8343
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.383322829683316e-06,
+      "loss": 0.8554,
+      "step": 8344
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.382562628013235e-06,
+      "loss": 0.8515,
+      "step": 8345
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.3818024263431545e-06,
+      "loss": 0.7492,
+      "step": 8346
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.381042224673074e-06,
+      "loss": 0.7302,
+      "step": 8347
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.380282023002993e-06,
+      "loss": 0.8126,
+      "step": 8348
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.379521821332913e-06,
+      "loss": 0.9681,
+      "step": 8349
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.378761619662833e-06,
+      "loss": 0.8615,
+      "step": 8350
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.378001417992753e-06,
+      "loss": 0.8793,
+      "step": 8351
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.377241216322673e-06,
+      "loss": 0.9367,
+      "step": 8352
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.376481014652592e-06,
+      "loss": 1.0743,
+      "step": 8353
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.375720812982512e-06,
+      "loss": 0.8391,
+      "step": 8354
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.3749606113124315e-06,
+      "loss": 0.7315,
+      "step": 8355
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.374200409642351e-06,
+      "loss": 0.632,
+      "step": 8356
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.373440207972271e-06,
+      "loss": 0.8701,
+      "step": 8357
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.372680006302191e-06,
+      "loss": 0.8891,
+      "step": 8358
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.37191980463211e-06,
+      "loss": 0.9681,
+      "step": 8359
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.37115960296203e-06,
+      "loss": 0.8812,
+      "step": 8360
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.370399401291949e-06,
+      "loss": 0.6877,
+      "step": 8361
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.369639199621869e-06,
+      "loss": 0.8577,
+      "step": 8362
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.368878997951789e-06,
+      "loss": 0.8238,
+      "step": 8363
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.3681187962817085e-06,
+      "loss": 0.785,
+      "step": 8364
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.367358594611628e-06,
+      "loss": 1.043,
+      "step": 8365
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.366598392941548e-06,
+      "loss": 0.8222,
+      "step": 8366
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.365838191271467e-06,
+      "loss": 0.9624,
+      "step": 8367
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.365077989601387e-06,
+      "loss": 0.8355,
+      "step": 8368
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.364317787931306e-06,
+      "loss": 0.7699,
+      "step": 8369
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.363557586261226e-06,
+      "loss": 0.8252,
+      "step": 8370
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.362797384591146e-06,
+      "loss": 0.8161,
+      "step": 8371
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.362037182921066e-06,
+      "loss": 0.8665,
+      "step": 8372
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3612769812509855e-06,
+      "loss": 0.8856,
+      "step": 8373
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.360516779580905e-06,
+      "loss": 0.8508,
+      "step": 8374
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.359756577910824e-06,
+      "loss": 0.9104,
+      "step": 8375
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.358996376240744e-06,
+      "loss": 0.7327,
+      "step": 8376
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.358236174570663e-06,
+      "loss": 0.9421,
+      "step": 8377
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.357475972900583e-06,
+      "loss": 1.2236,
+      "step": 8378
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.356715771230503e-06,
+      "loss": 0.7572,
+      "step": 8379
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.355955569560423e-06,
+      "loss": 0.7092,
+      "step": 8380
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.355195367890343e-06,
+      "loss": 0.7611,
+      "step": 8381
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.354435166220262e-06,
+      "loss": 0.7655,
+      "step": 8382
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3536749645501815e-06,
+      "loss": 0.981,
+      "step": 8383
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.352914762880101e-06,
+      "loss": 0.8085,
+      "step": 8384
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.35215456121002e-06,
+      "loss": 1.0031,
+      "step": 8385
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.35139435953994e-06,
+      "loss": 0.8048,
+      "step": 8386
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.35063415786986e-06,
+      "loss": 0.6168,
+      "step": 8387
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.34987395619978e-06,
+      "loss": 0.8977,
+      "step": 8388
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3491137545297e-06,
+      "loss": 0.961,
+      "step": 8389
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.348353552859619e-06,
+      "loss": 0.8366,
+      "step": 8390
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.347593351189539e-06,
+      "loss": 0.9695,
+      "step": 8391
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3468331495194585e-06,
+      "loss": 0.8185,
+      "step": 8392
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.3460729478493775e-06,
+      "loss": 0.8534,
+      "step": 8393
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.345312746179297e-06,
+      "loss": 0.9336,
+      "step": 8394
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.344552544509217e-06,
+      "loss": 0.8024,
+      "step": 8395
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.343792342839137e-06,
+      "loss": 0.8257,
+      "step": 8396
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.343032141169057e-06,
+      "loss": 0.949,
+      "step": 8397
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.342271939498976e-06,
+      "loss": 0.7311,
+      "step": 8398
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.341511737828896e-06,
+      "loss": 0.6984,
+      "step": 8399
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.340751536158816e-06,
+      "loss": 0.7441,
+      "step": 8400
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.339991334488735e-06,
+      "loss": 0.7258,
+      "step": 8401
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.3392311328186545e-06,
+      "loss": 0.7354,
+      "step": 8402
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.338470931148574e-06,
+      "loss": 0.7841,
+      "step": 8403
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.337710729478494e-06,
+      "loss": 0.8175,
+      "step": 8404
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.336950527808414e-06,
+      "loss": 1.092,
+      "step": 8405
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.336190326138333e-06,
+      "loss": 0.8209,
+      "step": 8406
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.335430124468253e-06,
+      "loss": 0.965,
+      "step": 8407
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.334669922798173e-06,
+      "loss": 0.8899,
+      "step": 8408
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.333909721128092e-06,
+      "loss": 0.8053,
+      "step": 8409
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.333149519458012e-06,
+      "loss": 0.8908,
+      "step": 8410
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.3323893177879316e-06,
+      "loss": 1.0805,
+      "step": 8411
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.331629116117851e-06,
+      "loss": 0.8618,
+      "step": 8412
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.330868914447771e-06,
+      "loss": 0.9125,
+      "step": 8413
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.33010871277769e-06,
+      "loss": 0.822,
+      "step": 8414
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.32934851110761e-06,
+      "loss": 0.8627,
+      "step": 8415
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.32858830943753e-06,
+      "loss": 0.737,
+      "step": 8416
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.327828107767449e-06,
+      "loss": 0.8913,
+      "step": 8417
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.327067906097369e-06,
+      "loss": 0.8925,
+      "step": 8418
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.326307704427289e-06,
+      "loss": 0.6289,
+      "step": 8419
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.3255475027572086e-06,
+      "loss": 0.7948,
+      "step": 8420
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.324787301087128e-06,
+      "loss": 0.8715,
+      "step": 8421
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3240270994170474e-06,
+      "loss": 0.8004,
+      "step": 8422
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.323266897746967e-06,
+      "loss": 1.055,
+      "step": 8423
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.322506696076887e-06,
+      "loss": 0.9496,
+      "step": 8424
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.321746494406806e-06,
+      "loss": 0.8414,
+      "step": 8425
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.320986292736726e-06,
+      "loss": 0.7724,
+      "step": 8426
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.320226091066646e-06,
+      "loss": 0.7797,
+      "step": 8427
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.319465889396566e-06,
+      "loss": 0.9017,
+      "step": 8428
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3187056877264856e-06,
+      "loss": 0.7597,
+      "step": 8429
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.317945486056405e-06,
+      "loss": 0.9463,
+      "step": 8430
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3171852843863244e-06,
+      "loss": 0.7407,
+      "step": 8431
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.316425082716244e-06,
+      "loss": 0.9541,
+      "step": 8432
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.315664881046163e-06,
+      "loss": 0.8352,
+      "step": 8433
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.314904679376083e-06,
+      "loss": 0.7372,
+      "step": 8434
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.314144477706003e-06,
+      "loss": 0.8339,
+      "step": 8435
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.313384276035923e-06,
+      "loss": 0.8708,
+      "step": 8436
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.312624074365843e-06,
+      "loss": 0.765,
+      "step": 8437
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.311863872695762e-06,
+      "loss": 0.972,
+      "step": 8438
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.311103671025682e-06,
+      "loss": 0.7564,
+      "step": 8439
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3103434693556015e-06,
+      "loss": 0.7548,
+      "step": 8440
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.309583267685521e-06,
+      "loss": 0.7841,
+      "step": 8441
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.30882306601544e-06,
+      "loss": 0.7356,
+      "step": 8442
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.30806286434536e-06,
+      "loss": 0.8903,
+      "step": 8443
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.30730266267528e-06,
+      "loss": 0.9495,
+      "step": 8444
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3065424610052e-06,
+      "loss": 0.9904,
+      "step": 8445
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.305782259335119e-06,
+      "loss": 0.8475,
+      "step": 8446
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.305022057665039e-06,
+      "loss": 0.7917,
+      "step": 8447
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.304261855994959e-06,
+      "loss": 0.8741,
+      "step": 8448
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.3035016543248785e-06,
+      "loss": 0.7593,
+      "step": 8449
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.302741452654798e-06,
+      "loss": 0.8164,
+      "step": 8450
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.301981250984718e-06,
+      "loss": 0.8557,
+      "step": 8451
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.301221049314637e-06,
+      "loss": 0.8214,
+      "step": 8452
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.300460847644557e-06,
+      "loss": 0.8084,
+      "step": 8453
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.299700645974476e-06,
+      "loss": 0.8966,
+      "step": 8454
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.298940444304396e-06,
+      "loss": 1.162,
+      "step": 8455
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.298180242634316e-06,
+      "loss": 0.8213,
+      "step": 8456
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.297420040964236e-06,
+      "loss": 0.8693,
+      "step": 8457
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.2966598392941555e-06,
+      "loss": 0.7187,
+      "step": 8458
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.295899637624075e-06,
+      "loss": 0.9647,
+      "step": 8459
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.295139435953994e-06,
+      "loss": 0.7972,
+      "step": 8460
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.294379234283914e-06,
+      "loss": 0.7488,
+      "step": 8461
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.293619032613833e-06,
+      "loss": 0.8534,
+      "step": 8462
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.292858830943753e-06,
+      "loss": 0.7848,
+      "step": 8463
+    }
+  ],
+  "max_steps": 14105,
+  "num_train_epochs": 5,
+  "total_flos": 3.034749218850865e+18,
+  "trial_name": null,
+  "trial_params": null
+}