diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14596 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2428,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.36986301369863e-07,
+      "loss": 1.7461,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.73972602739726e-07,
+      "loss": 1.8203,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1095890410958903e-07,
+      "loss": 1.8057,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.47945205479452e-07,
+      "loss": 1.7764,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.849315068493151e-07,
+      "loss": 1.709,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.219178082191781e-07,
+      "loss": 1.8711,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.589041095890411e-07,
+      "loss": 1.75,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.095890410958904e-06,
+      "loss": 1.8271,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2328767123287673e-06,
+      "loss": 1.8252,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3698630136986302e-06,
+      "loss": 1.8164,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5068493150684932e-06,
+      "loss": 1.8252,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6438356164383561e-06,
+      "loss": 1.8438,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7808219178082193e-06,
+      "loss": 1.7969,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9178082191780823e-06,
+      "loss": 1.8623,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0547945205479454e-06,
+      "loss": 1.7861,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.191780821917808e-06,
+      "loss": 1.8477,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3287671232876713e-06,
+      "loss": 1.9336,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4657534246575345e-06,
+      "loss": 1.7793,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6027397260273973e-06,
+      "loss": 1.7891,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7397260273972604e-06,
+      "loss": 1.7568,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.876712328767123e-06,
+      "loss": 1.7334,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0136986301369864e-06,
+      "loss": 1.6445,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1506849315068495e-06,
+      "loss": 1.7939,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2876712328767123e-06,
+      "loss": 1.7139,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4246575342465754e-06,
+      "loss": 1.7695,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5616438356164386e-06,
+      "loss": 1.6689,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6986301369863014e-06,
+      "loss": 1.6699,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8356164383561645e-06,
+      "loss": 1.6709,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.972602739726027e-06,
+      "loss": 1.5879,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.109589041095891e-06,
+      "loss": 1.5781,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.246575342465754e-06,
+      "loss": 1.6162,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.383561643835616e-06,
+      "loss": 1.6562,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.52054794520548e-06,
+      "loss": 1.5303,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.657534246575343e-06,
+      "loss": 1.5654,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7945205479452054e-06,
+      "loss": 1.6201,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.931506849315069e-06,
+      "loss": 1.585,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.068493150684932e-06,
+      "loss": 1.4912,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2054794520547945e-06,
+      "loss": 1.5381,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.342465753424658e-06,
+      "loss": 1.5752,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.479452054794521e-06,
+      "loss": 1.4893,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6164383561643845e-06,
+      "loss": 1.542,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.753424657534246e-06,
+      "loss": 1.5332,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.89041095890411e-06,
+      "loss": 1.5654,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.027397260273973e-06,
+      "loss": 1.5283,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.164383561643836e-06,
+      "loss": 1.4121,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.301369863013699e-06,
+      "loss": 1.4678,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.438356164383563e-06,
+      "loss": 1.4824,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.5753424657534245e-06,
+      "loss": 1.4893,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.712328767123288e-06,
+      "loss": 1.4404,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.849315068493151e-06,
+      "loss": 1.4053,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.9863013698630145e-06,
+      "loss": 1.4326,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.123287671232877e-06,
+      "loss": 1.4717,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.260273972602741e-06,
+      "loss": 1.3701,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.397260273972603e-06,
+      "loss": 1.4238,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.534246575342466e-06,
+      "loss": 1.3672,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.671232876712329e-06,
+      "loss": 1.4424,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.808219178082192e-06,
+      "loss": 1.3633,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.945205479452055e-06,
+      "loss": 1.4258,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.082191780821919e-06,
+      "loss": 1.2329,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.219178082191782e-06,
+      "loss": 1.2744,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.356164383561644e-06,
+      "loss": 1.3242,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.493150684931507e-06,
+      "loss": 1.2539,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.63013698630137e-06,
+      "loss": 1.3594,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.767123287671233e-06,
+      "loss": 1.3203,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.904109589041097e-06,
+      "loss": 1.2051,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.04109589041096e-06,
+      "loss": 1.3926,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.178082191780823e-06,
+      "loss": 1.2695,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.315068493150685e-06,
+      "loss": 1.3203,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.452054794520548e-06,
+      "loss": 1.335,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.589041095890411e-06,
+      "loss": 1.3076,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.726027397260275e-06,
+      "loss": 1.2944,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.863013698630138e-06,
+      "loss": 1.1631,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 1.3525,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9999955510465e-06,
+      "loss": 1.3223,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999982204193912e-06,
+      "loss": 1.3408,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99995995946599e-06,
+      "loss": 1.2974,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99992881690232e-06,
+      "loss": 1.3057,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999888776558325e-06,
+      "loss": 1.3076,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999839838505257e-06,
+      "loss": 1.2065,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999782002830206e-06,
+      "loss": 1.2686,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999715269636094e-06,
+      "loss": 1.1885,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999639639041681e-06,
+      "loss": 1.2812,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999555111181558e-06,
+      "loss": 1.2832,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999461686206146e-06,
+      "loss": 1.2002,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999359364281704e-06,
+      "loss": 1.291,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999248145590324e-06,
+      "loss": 1.2046,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999128030329924e-06,
+      "loss": 1.2798,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998999018714264e-06,
+      "loss": 1.2002,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998861110972929e-06,
+      "loss": 1.2798,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998714307351334e-06,
+      "loss": 1.1528,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998558608110733e-06,
+      "loss": 1.168,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998394013528201e-06,
+      "loss": 1.2451,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99822052389665e-06,
+      "loss": 1.2363,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99803813952482e-06,
+      "loss": 1.2661,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997846860737276e-06,
+      "loss": 1.1011,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997646687874416e-06,
+      "loss": 1.2026,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997437621292463e-06,
+      "loss": 1.1416,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99721966136347e-06,
+      "loss": 1.1865,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99699280847531e-06,
+      "loss": 1.1938,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99675706303169e-06,
+      "loss": 1.144,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996512425452138e-06,
+      "loss": 1.1548,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996258896172005e-06,
+      "loss": 1.2061,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995996475642466e-06,
+      "loss": 1.2695,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995725164330523e-06,
+      "loss": 1.2607,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995444962718994e-06,
+      "loss": 1.1401,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995155871306524e-06,
+      "loss": 1.2017,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99485789060757e-06,
+      "loss": 1.0908,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994551021152415e-06,
+      "loss": 1.1323,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994235263487158e-06,
+      "loss": 1.2026,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993910618173717e-06,
+      "loss": 1.1836,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993577085789824e-06,
+      "loss": 1.2466,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993234666929024e-06,
+      "loss": 1.2622,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992883362200683e-06,
+      "loss": 1.0928,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992523172229975e-06,
+      "loss": 1.2593,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992154097657888e-06,
+      "loss": 1.0977,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991776139141219e-06,
+      "loss": 1.1855,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991389297352577e-06,
+      "loss": 1.1621,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99099357298038e-06,
+      "loss": 1.3018,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990588966728846e-06,
+      "loss": 1.2124,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990175479318011e-06,
+      "loss": 1.2979,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989753111483707e-06,
+      "loss": 1.1841,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989321863977572e-06,
+      "loss": 1.2314,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988881737567046e-06,
+      "loss": 1.123,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988432733035369e-06,
+      "loss": 1.2017,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987974851181583e-06,
+      "loss": 1.0327,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987508092820524e-06,
+      "loss": 1.1426,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987032458782828e-06,
+      "loss": 1.1318,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986547949914925e-06,
+      "loss": 1.3086,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986054567079036e-06,
+      "loss": 1.2197,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985552311153178e-06,
+      "loss": 1.1074,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985041183031154e-06,
+      "loss": 1.0684,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98452118362256e-06,
+      "loss": 1.3213,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983992313852776e-06,
+      "loss": 1.1309,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98345457466297e-06,
+      "loss": 1.189,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982907967010092e-06,
+      "loss": 1.1865,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982352491866874e-06,
+      "loss": 1.1133,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981788150221829e-06,
+      "loss": 1.1011,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981214943079252e-06,
+      "loss": 1.1333,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980632871459209e-06,
+      "loss": 1.1514,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980041936397545e-06,
+      "loss": 1.1904,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979442138945876e-06,
+      "loss": 1.2168,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978833480171592e-06,
+      "loss": 1.293,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978215961157848e-06,
+      "loss": 1.1157,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977589583003574e-06,
+      "loss": 1.1201,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976954346823456e-06,
+      "loss": 1.1665,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976310253747952e-06,
+      "loss": 1.208,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975657304923275e-06,
+      "loss": 1.1235,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974995501511404e-06,
+      "loss": 1.127,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974324844690069e-06,
+      "loss": 1.1357,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97364533565276e-06,
+      "loss": 1.2285,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972956975608719e-06,
+      "loss": 1.2363,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972259765782937e-06,
+      "loss": 1.062,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971553707416157e-06,
+      "loss": 1.2173,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970838801764866e-06,
+      "loss": 1.0972,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970115050101298e-06,
+      "loss": 1.106,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96938245371343e-06,
+      "loss": 1.0703,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968641013904974e-06,
+      "loss": 1.2124,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967890731995383e-06,
+      "loss": 1.1367,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967131609319845e-06,
+      "loss": 1.2002,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96636364722928e-06,
+      "loss": 1.1309,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96558684709034e-06,
+      "loss": 1.1226,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964801210285403e-06,
+      "loss": 1.1133,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964006738212574e-06,
+      "loss": 1.1191,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963203432285683e-06,
+      "loss": 1.187,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962391293934277e-06,
+      "loss": 1.0942,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96157032460362e-06,
+      "loss": 1.1328,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960740525754696e-06,
+      "loss": 1.1465,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959901898864196e-06,
+      "loss": 1.1147,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959054445424532e-06,
+      "loss": 1.041,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958198166943809e-06,
+      "loss": 1.0957,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957333064945849e-06,
+      "loss": 1.1113,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95645914097017e-06,
+      "loss": 1.2236,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95557639657199e-06,
+      "loss": 1.2642,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954684833322228e-06,
+      "loss": 1.0527,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953784452807487e-06,
+      "loss": 1.0898,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952875256630072e-06,
+      "loss": 1.0698,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951957246407972e-06,
+      "loss": 1.1089,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951030423774858e-06,
+      "loss": 1.0786,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950094790380089e-06,
+      "loss": 1.0908,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9491503478887e-06,
+      "loss": 1.2061,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948197097981401e-06,
+      "loss": 1.0942,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94723504235458e-06,
+      "loss": 1.0986,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946264182720295e-06,
+      "loss": 1.0322,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945284520806267e-06,
+      "loss": 1.1289,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944296058355884e-06,
+      "loss": 1.041,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943298797128198e-06,
+      "loss": 1.0894,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942292738897914e-06,
+      "loss": 1.1216,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941277885455396e-06,
+      "loss": 1.0869,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940254238606657e-06,
+      "loss": 1.1475,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93922180017336e-06,
+      "loss": 1.0532,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938180571992814e-06,
+      "loss": 1.1606,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93713055591797e-06,
+      "loss": 1.0767,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936071753817416e-06,
+      "loss": 1.1333,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935004167575377e-06,
+      "loss": 1.1816,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933927799091708e-06,
+      "loss": 1.0547,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932842650281897e-06,
+      "loss": 1.1177,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931748723077053e-06,
+      "loss": 1.0605,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930646019423909e-06,
+      "loss": 0.9521,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929534541284814e-06,
+      "loss": 1.0547,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928414290637738e-06,
+      "loss": 1.084,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927285269476254e-06,
+      "loss": 1.0991,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92614747980955e-06,
+      "loss": 1.1309,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925000923662413e-06,
+      "loss": 1.0884,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923845603075235e-06,
+      "loss": 1.2085,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922681520104002e-06,
+      "loss": 1.1611,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921508676820296e-06,
+      "loss": 1.0332,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920327075311283e-06,
+      "loss": 1.2393,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919136717679723e-06,
+      "loss": 1.1348,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917937606043952e-06,
+      "loss": 1.0171,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91672974253789e-06,
+      "loss": 1.0254,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915513129311025e-06,
+      "loss": 0.9883,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91428776852842e-06,
+      "loss": 1.3428,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913053662370706e-06,
+      "loss": 1.1118,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911810813034073e-06,
+      "loss": 1.1646,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910559222730274e-06,
+      "loss": 1.1938,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909298893686617e-06,
+      "loss": 1.0967,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908029828145956e-06,
+      "loss": 1.0894,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906752028366702e-06,
+      "loss": 1.1929,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905465496622798e-06,
+      "loss": 1.0298,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904170235203737e-06,
+      "loss": 1.1577,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902866246414537e-06,
+      "loss": 1.1758,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901553532575759e-06,
+      "loss": 1.0547,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900232096023478e-06,
+      "loss": 1.0835,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.8989019391093e-06,
+      "loss": 1.1001,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897563064200348e-06,
+      "loss": 1.0664,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896215473679259e-06,
+      "loss": 1.0352,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89485916994418e-06,
+      "loss": 1.0786,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893494155408761e-06,
+      "loss": 1.1367,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892120432502161e-06,
+      "loss": 1.1416,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890738003669029e-06,
+      "loss": 1.1602,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889346871369511e-06,
+      "loss": 1.0664,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887947038079238e-06,
+      "loss": 0.9717,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88653850628933e-06,
+      "loss": 1.02,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885121278506382e-06,
+      "loss": 1.1685,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883695357252467e-06,
+      "loss": 1.0205,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882260745065128e-06,
+      "loss": 0.9971,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880817444497375e-06,
+      "loss": 1.106,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879365458117678e-06,
+      "loss": 1.1455,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877904788509964e-06,
+      "loss": 1.2119,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876435438273614e-06,
+      "loss": 1.0396,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874957410023458e-06,
+      "loss": 1.0459,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873470706389765e-06,
+      "loss": 1.1914,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871975330018246e-06,
+      "loss": 1.0771,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870471283570046e-06,
+      "loss": 1.0972,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868958569721738e-06,
+      "loss": 1.1128,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867437191165318e-06,
+      "loss": 1.105,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865907150608203e-06,
+      "loss": 1.0942,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864368450773227e-06,
+      "loss": 1.1245,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862821094398631e-06,
+      "loss": 1.0996,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86126508423806e-06,
+      "loss": 0.9609,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859700423060561e-06,
+      "loss": 1.084,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858127113650576e-06,
+      "loss": 0.9849,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856545158807938e-06,
+      "loss": 1.1973,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854954561347864e-06,
+      "loss": 0.9761,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853355324100953e-06,
+      "loss": 1.0239,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851747449913176e-06,
+      "loss": 1.0762,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850130941645876e-06,
+      "loss": 1.187,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848505802175762e-06,
+      "loss": 1.1025,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846872034394902e-06,
+      "loss": 1.0361,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845229641210719e-06,
+      "loss": 1.0327,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843578625545981e-06,
+      "loss": 1.187,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841918990338812e-06,
+      "loss": 1.1074,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840250738542663e-06,
+      "loss": 1.0303,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838573873126326e-06,
+      "loss": 1.1455,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836888397073919e-06,
+      "loss": 1.0479,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835194313384884e-06,
+      "loss": 1.1143,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83349162507398e-06,
+      "loss": 1.1699,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83178033517128e-06,
+      "loss": 1.0303,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83006044672216e-06,
+      "loss": 0.9976,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82833196278731e-06,
+      "loss": 1.0532,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8265948864427e-06,
+      "loss": 1.1606,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824849220779604e-06,
+      "loss": 1.0381,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823094968904572e-06,
+      "loss": 1.1318,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82133213393944e-06,
+      "loss": 1.082,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819560719021318e-06,
+      "loss": 0.979,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81778072730258e-06,
+      "loss": 0.9985,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815992161950867e-06,
+      "loss": 0.9629,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814195026149077e-06,
+      "loss": 0.9102,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812389323095359e-06,
+      "loss": 1.2002,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.81057505600311e-06,
+      "loss": 0.9727,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808752228100965e-06,
+      "loss": 1.0957,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806920842632795e-06,
+      "loss": 1.0713,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8050809028577e-06,
+      "loss": 1.0474,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80323241205e-06,
+      "loss": 1.0684,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801375373499236e-06,
+      "loss": 1.0981,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79950979051016e-06,
+      "loss": 1.1533,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797635666402729e-06,
+      "loss": 1.1875,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7957530045121e-06,
+      "loss": 0.9639,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793861808188622e-06,
+      "loss": 1.1216,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791962080797832e-06,
+      "loss": 1.002,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79005382572045e-06,
+      "loss": 1.1426,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788137046352374e-06,
+      "loss": 1.0283,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786211746104665e-06,
+      "loss": 1.0791,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784277928403554e-06,
+      "loss": 1.0908,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782335596690425e-06,
+      "loss": 1.0195,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780384754421817e-06,
+      "loss": 1.0537,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778425405069413e-06,
+      "loss": 0.957,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776457552120034e-06,
+      "loss": 0.979,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774481199075634e-06,
+      "loss": 0.9951,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772496349453292e-06,
+      "loss": 1.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770503006785214e-06,
+      "loss": 1.1709,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768501174618714e-06,
+      "loss": 1.0103,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766490856516216e-06,
+      "loss": 1.0752,
+      "step": 303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76447205605524e-06,
+      "loss": 1.1299,
+      "step": 304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76244477682841e-06,
+      "loss": 1.1074,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760409022443436e-06,
+      "loss": 1.1187,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758364796523105e-06,
+      "loss": 1.2227,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756312102705284e-06,
+      "loss": 1.0952,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754250944642909e-06,
+      "loss": 0.9912,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75218132600398e-06,
+      "loss": 1.0645,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75010325047155e-06,
+      "loss": 1.1128,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748016721743724e-06,
+      "loss": 1.002,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745921743533653e-06,
+      "loss": 1.0625,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743818319569516e-06,
+      "loss": 1.0781,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74170645359453e-06,
+      "loss": 1.0586,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739586149366932e-06,
+      "loss": 1.1577,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737457410659976e-06,
+      "loss": 1.064,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735320241261925e-06,
+      "loss": 1.0938,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733174644976047e-06,
+      "loss": 1.0791,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731020625620605e-06,
+      "loss": 0.9517,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728858187028852e-06,
+      "loss": 1.0879,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726687333049024e-06,
+      "loss": 0.9927,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724508067544328e-06,
+      "loss": 0.9912,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72232039439295e-06,
+      "loss": 1.1279,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720124317488031e-06,
+      "loss": 0.9292,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717919840737667e-06,
+      "loss": 1.188,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715706968064905e-06,
+      "loss": 1.0654,
+      "step": 327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713485703407732e-06,
+      "loss": 1.0205,
+      "step": 328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711256050719068e-06,
+      "loss": 1.0918,
+      "step": 329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709018013966764e-06,
+      "loss": 1.062,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706771597133587e-06,
+      "loss": 0.9946,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70451680421722e-06,
+      "loss": 1.0991,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702253639230246e-06,
+      "loss": 1.1221,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699982106200155e-06,
+      "loss": 1.2358,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697702209169325e-06,
+      "loss": 1.0996,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69541395219502e-06,
+      "loss": 1.0322,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693117339349376e-06,
+      "loss": 0.9771,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690812374719404e-06,
+      "loss": 1.0933,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688499062406975e-06,
+      "loss": 1.0884,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686177406528819e-06,
+      "loss": 1.0796,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683847411216509e-06,
+      "loss": 0.876,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681509080616464e-06,
+      "loss": 1.0132,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679162418889932e-06,
+      "loss": 1.0723,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676807430212988e-06,
+      "loss": 1.0649,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674444118776527e-06,
+      "loss": 1.1367,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672072488786254e-06,
+      "loss": 0.9346,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669692544462678e-06,
+      "loss": 1.1431,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667304290041102e-06,
+      "loss": 1.1011,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664907729771622e-06,
+      "loss": 1.0059,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662502867919111e-06,
+      "loss": 0.9712,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660089708763214e-06,
+      "loss": 1.0571,
+      "step": 351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657668256598347e-06,
+      "loss": 1.0884,
+      "step": 352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655238515733678e-06,
+      "loss": 0.9844,
+      "step": 353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652800490493132e-06,
+      "loss": 1.0942,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650354185215374e-06,
+      "loss": 0.9663,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647899604253801e-06,
+      "loss": 1.1211,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645436751976541e-06,
+      "loss": 1.0576,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642965632766437e-06,
+      "loss": 1.1108,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64048625102105e-06,
+      "loss": 1.019,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637998611152641e-06,
+      "loss": 1.0332,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635502717588168e-06,
+      "loss": 1.186,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632998574769277e-06,
+      "loss": 1.0483,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630486187152291e-06,
+      "loss": 1.1743,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627965559208212e-06,
+      "loss": 0.981,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6254366954227e-06,
+      "loss": 1.0332,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622899600296076e-06,
+      "loss": 1.0474,
+      "step": 366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620354278343306e-06,
+      "loss": 1.0723,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617800734093999e-06,
+      "loss": 1.0879,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615238972092391e-06,
+      "loss": 1.0732,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612668996897351e-06,
+      "loss": 1.0532,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610090813082356e-06,
+      "loss": 0.9697,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607504425235495e-06,
+      "loss": 1.0337,
+      "step": 372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604909837959456e-06,
+      "loss": 1.0464,
+      "step": 373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602307055871517e-06,
+      "loss": 1.0713,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59969608360354e-06,
+      "loss": 1.0015,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597076925801967e-06,
+      "loss": 0.9697,
+      "step": 376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594449587127799e-06,
+      "loss": 1.0298,
+      "step": 377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591814072256598e-06,
+      "loss": 0.9805,
+      "step": 378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58917038587848e-06,
+      "loss": 0.9692,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586518532698099e-06,
+      "loss": 1.062,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583858517434643e-06,
+      "loss": 0.9497,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581190344821827e-06,
+      "loss": 0.9961,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57851401960788e-06,
+      "loss": 0.9917,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575829546555541e-06,
+      "loss": 1.0244,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57313693044205e-06,
+      "loss": 1.0806,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570436176059133e-06,
+      "loss": 1.1147,
+      "step": 386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567727288213005e-06,
+      "loss": 1.1094,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565010271724353e-06,
+      "loss": 1.0659,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562285131428326e-06,
+      "loss": 0.9971,
+      "step": 389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559551872174535e-06,
+      "loss": 1.1084,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556810498827039e-06,
+      "loss": 0.9102,
+      "step": 391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554061016264329e-06,
+      "loss": 1.0322,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55130342937934e-06,
+      "loss": 1.0522,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54853774307942e-06,
+      "loss": 1.0034,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545763962286332e-06,
+      "loss": 1.0903,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542982091936244e-06,
+      "loss": 1.0835,
+      "step": 396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540192136979722e-06,
+      "loss": 1.0693,
+      "step": 397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537394102381719e-06,
+      "loss": 1.0762,
+      "step": 398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534587993121563e-06,
+      "loss": 1.0234,
+      "step": 399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531773814192953e-06,
+      "loss": 0.9917,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528951570603954e-06,
+      "loss": 0.9658,
+      "step": 401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526121267376976e-06,
+      "loss": 0.999,
+      "step": 402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523282909548773e-06,
+      "loss": 1.0176,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520436502170433e-06,
+      "loss": 1.0522,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517582050307371e-06,
+      "loss": 1.0347,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514719559039318e-06,
+      "loss": 0.9746,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511849033460307e-06,
+      "loss": 1.0405,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508970478678676e-06,
+      "loss": 1.0049,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506083899817043e-06,
+      "loss": 0.9824,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503189302012312e-06,
+      "loss": 0.9429,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500286690415657e-06,
+      "loss": 0.9409,
+      "step": 411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49737607019251e-06,
+      "loss": 1.0864,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494457446522555e-06,
+      "loss": 0.9834,
+      "step": 413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491530824599724e-06,
+      "loss": 1.0547,
+      "step": 414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488596209632179e-06,
+      "loss": 1.1895,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485653606842302e-06,
+      "loss": 1.1084,
+      "step": 416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482703021466698e-06,
+      "loss": 0.8652,
+      "step": 417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47974445875617e-06,
+      "loss": 1.0117,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476777923975725e-06,
+      "loss": 0.9292,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473803422404552e-06,
+      "loss": 1.1221,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470820959336018e-06,
+      "loss": 1.1587,
+      "step": 421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46783054007766e-06,
+      "loss": 1.0977,
+      "step": 422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464832169951171e-06,
+      "loss": 1.0508,
+      "step": 423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461825854292394e-06,
+      "loss": 0.9536,
+      "step": 424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458811598451317e-06,
+      "loss": 1.0884,
+      "step": 425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455789407792047e-06,
+      "loss": 1.0513,
+      "step": 426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452759287692824e-06,
+      "loss": 0.9458,
+      "step": 427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44972124354599e-06,
+      "loss": 1.0825,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446675280757993e-06,
+      "loss": 1.083,
+      "step": 429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443621404749374e-06,
+      "loss": 0.874,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44055962095475e-06,
+      "loss": 1.0464,
+      "step": 431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437489934822817e-06,
+      "loss": 1.0215,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434412351816329e-06,
+      "loss": 0.9717,
+      "step": 433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431326877412097e-06,
+      "loss": 1.0776,
+      "step": 434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428233517100976e-06,
+      "loss": 1.0771,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425132276387847e-06,
+      "loss": 0.9482,
+      "step": 436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422023160791626e-06,
+      "loss": 0.9419,
+      "step": 437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418906175845234e-06,
+      "loss": 1.0107,
+      "step": 438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415781327095601e-06,
+      "loss": 1.0962,
+      "step": 439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412648620103649e-06,
+      "loss": 1.0591,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409508060444283e-06,
+      "loss": 0.96,
+      "step": 441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40635965370639e-06,
+      "loss": 1.0308,
+      "step": 442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403203405492812e-06,
+      "loss": 1.0493,
+      "step": 443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400039321420348e-06,
+      "loss": 0.918,
+      "step": 444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396867407119748e-06,
+      "loss": 0.9395,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393687668235689e-06,
+      "loss": 1.042,
+      "step": 446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390500110426773e-06,
+      "loss": 1.1074,
+      "step": 447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387304739365524e-06,
+      "loss": 1.042,
+      "step": 448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38410156073836e-06,
+      "loss": 1.001,
+      "step": 449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3808905802456e-06,
+      "loss": 0.9482,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377671803601447e-06,
+      "loss": 1.0527,
+      "step": 451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374445236533974e-06,
+      "loss": 1.0449,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371210884785118e-06,
+      "loss": 0.9517,
+      "step": 453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367968754110675e-06,
+      "loss": 1.0479,
+      "step": 454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364718850280278e-06,
+      "loss": 1.0957,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361461179077395e-06,
+      "loss": 1.1719,
+      "step": 456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35819574629932e-06,
+      "loss": 1.1372,
+      "step": 457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.354922557757153e-06,
+      "loss": 1.0547,
+      "step": 458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351641619275802e-06,
+      "loss": 1.0322,
+      "step": 459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348352936693964e-06,
+      "loss": 1.0742,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345056515864115e-06,
+      "loss": 0.9922,
+      "step": 461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341752362652506e-06,
+      "loss": 0.9751,
+      "step": 462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338440482939146e-06,
+      "loss": 1.0171,
+      "step": 463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335120882617797e-06,
+      "loss": 0.9585,
+      "step": 464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331793567595953e-06,
+      "loss": 1.0513,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328458543794844e-06,
+      "loss": 0.9741,
+      "step": 466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325115817149419e-06,
+      "loss": 1.0928,
+      "step": 467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32176539360833e-06,
+      "loss": 1.0112,
+      "step": 468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31840727913393e-06,
+      "loss": 1.0078,
+      "step": 469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315041479702253e-06,
+      "loss": 0.9209,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311668001303018e-06,
+      "loss": 0.9551,
+      "step": 471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3082868499396e-06,
+      "loss": 1.1714,
+      "step": 472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.304898031629038e-06,
+      "loss": 0.9766,
+      "step": 473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301501552402006e-06,
+      "loss": 1.0278,
+      "step": 474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298097418302817e-06,
+      "loss": 0.9805,
+      "step": 475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294685635389402e-06,
+      "loss": 0.9824,
+      "step": 476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29126620973331e-06,
+      "loss": 0.9556,
+      "step": 477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.287839147419685e-06,
+      "loss": 1.0361,
+      "step": 478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284404454547268e-06,
+      "loss": 1.0239,
+      "step": 479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28096213722837e-06,
+      "loss": 1.0059,
+      "step": 480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277512201588871e-06,
+      "loss": 1.0342,
+      "step": 481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274054653768222e-06,
+      "loss": 0.9531,
+      "step": 482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270589499919405e-06,
+      "loss": 1.0771,
+      "step": 483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267116746208944e-06,
+      "loss": 1.0063,
+      "step": 484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263636398816883e-06,
+      "loss": 0.9932,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260148463936794e-06,
+      "loss": 0.9595,
+      "step": 486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25665294777573e-06,
+      "loss": 1.0137,
+      "step": 487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253149856554255e-06,
+      "loss": 0.9976,
+      "step": 488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2496391965064e-06,
+      "loss": 1.1753,
+      "step": 489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24612097387967e-06,
+      "loss": 0.96,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24259519493503e-06,
+      "loss": 1.0283,
+      "step": 491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239061865946892e-06,
+      "loss": 0.9746,
+      "step": 492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2355209932031e-06,
+      "loss": 1.0908,
+      "step": 493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231972583004926e-06,
+      "loss": 1.1357,
+      "step": 494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228416641667058e-06,
+      "loss": 0.9297,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224853175517578e-06,
+      "loss": 1.0259,
+      "step": 496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.221282190897969e-06,
+      "loss": 1.1641,
+      "step": 497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217703694163083e-06,
+      "loss": 1.0142,
+      "step": 498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214117691681152e-06,
+      "loss": 0.9541,
+      "step": 499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210524189833756e-06,
+      "loss": 1.041,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206923195015824e-06,
+      "loss": 1.186,
+      "step": 501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203314713635621e-06,
+      "loss": 1.0093,
+      "step": 502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199698752114732e-06,
+      "loss": 1.1753,
+      "step": 503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.196075316888056e-06,
+      "loss": 0.9785,
+      "step": 504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.192444414403792e-06,
+      "loss": 0.9248,
+      "step": 505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188806051123423e-06,
+      "loss": 1.0884,
+      "step": 506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185160233521715e-06,
+      "loss": 0.9941,
+      "step": 507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181506968086696e-06,
+      "loss": 1.0479,
+      "step": 508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.177846261319652e-06,
+      "loss": 1.1118,
+      "step": 509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174178119735104e-06,
+      "loss": 1.0522,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.170502549860813e-06,
+      "loss": 0.9824,
+      "step": 511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166819558237752e-06,
+      "loss": 0.9873,
+      "step": 512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163129151420107e-06,
+      "loss": 1.0283,
+      "step": 513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159431335975255e-06,
+      "loss": 1.0835,
+      "step": 514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155726118483759e-06,
+      "loss": 0.9175,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152013505539357e-06,
+      "loss": 1.0376,
+      "step": 516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148293503748947e-06,
+      "loss": 0.9312,
+      "step": 517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14456611973257e-06,
+      "loss": 0.9956,
+      "step": 518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140831360123415e-06,
+      "loss": 1.0205,
+      "step": 519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.137089231567789e-06,
+      "loss": 0.9966,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133339740725113e-06,
+      "loss": 0.978,
+      "step": 521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.129582894267911e-06,
+      "loss": 1.085,
+      "step": 522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125818698881798e-06,
+      "loss": 1.0635,
+      "step": 523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122047161265468e-06,
+      "loss": 1.0293,
+      "step": 524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118268288130676e-06,
+      "loss": 1.0439,
+      "step": 525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114482086202236e-06,
+      "loss": 0.9502,
+      "step": 526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110688562218001e-06,
+      "loss": 0.8887,
+      "step": 527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.106887722928858e-06,
+      "loss": 1.0894,
+      "step": 528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.103079575098708e-06,
+      "loss": 0.9976,
+      "step": 529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099264125504463e-06,
+      "loss": 1.0474,
+      "step": 530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09544138093602e-06,
+      "loss": 0.9082,
+      "step": 531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091611348196272e-06,
+      "loss": 1.0503,
+      "step": 532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087774034101069e-06,
+      "loss": 1.0513,
+      "step": 533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.083929445479224e-06,
+      "loss": 0.9629,
+      "step": 534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080077589172496e-06,
+      "loss": 1.0161,
+      "step": 535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.076218472035578e-06,
+      "loss": 0.9775,
+      "step": 536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072352100936083e-06,
+      "loss": 1.1035,
+      "step": 537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068478482754532e-06,
+      "loss": 0.9639,
+      "step": 538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064597624384343e-06,
+      "loss": 1.0332,
+      "step": 539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.060709532731822e-06,
+      "loss": 1.0654,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056814214716143e-06,
+      "loss": 1.0659,
+      "step": 541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05291167726934e-06,
+      "loss": 1.0503,
+      "step": 542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049001927336301e-06,
+      "loss": 0.9604,
+      "step": 543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 0.9561,
+      "step": 544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.041160817855196e-06,
+      "loss": 0.9111,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037229472261026e-06,
+      "loss": 0.9839,
+      "step": 546
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033290942088377e-06,
+      "loss": 1.0195,
+      "step": 547
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029345234346183e-06,
+      "loss": 1.0454,
+      "step": 548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025392356056154e-06,
+      "loss": 1.1416,
+      "step": 549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021432314252758e-06,
+      "loss": 1.019,
+      "step": 550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.017465115983212e-06,
+      "loss": 1.019,
+      "step": 551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013490768307468e-06,
+      "loss": 0.96,
+      "step": 552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.009509278298201e-06,
+      "loss": 1.0771,
+      "step": 553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005520653040796e-06,
+      "loss": 1.0586,
+      "step": 554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.001524899633337e-06,
+      "loss": 0.9927,
+      "step": 555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.997522025186592e-06,
+      "loss": 0.9746,
+      "step": 556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993512036824004e-06,
+      "loss": 1.0972,
+      "step": 557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.989494941681672e-06,
+      "loss": 0.9697,
+      "step": 558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.985470746908342e-06,
+      "loss": 0.9092,
+      "step": 559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.9814394596654e-06,
+      "loss": 0.9219,
+      "step": 560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977401087126847e-06,
+      "loss": 0.9214,
+      "step": 561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.973355636479294e-06,
+      "loss": 1.0913,
+      "step": 562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969303114921956e-06,
+      "loss": 1.0532,
+      "step": 563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96524352966662e-06,
+      "loss": 0.9263,
+      "step": 564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96117688793765e-06,
+      "loss": 1.0562,
+      "step": 565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957103196971965e-06,
+      "loss": 0.9458,
+      "step": 566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95302246401903e-06,
+      "loss": 0.9673,
+      "step": 567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.948934696340842e-06,
+      "loss": 0.9668,
+      "step": 568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944839901211917e-06,
+      "loss": 0.9136,
+      "step": 569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940738085919274e-06,
+      "loss": 1.0337,
+      "step": 570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.936629257762429e-06,
+      "loss": 0.9019,
+      "step": 571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.932513424053376e-06,
+      "loss": 0.9741,
+      "step": 572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.928390592116576e-06,
+      "loss": 1.0391,
+      "step": 573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.924260769288944e-06,
+      "loss": 0.9478,
+      "step": 574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920123962919834e-06,
+      "loss": 1.0283,
+      "step": 575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.915980180371034e-06,
+      "loss": 0.9648,
+      "step": 576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911829429016737e-06,
+      "loss": 0.9575,
+      "step": 577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907671716243547e-06,
+      "loss": 1.0708,
+      "step": 578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.903507049450452e-06,
+      "loss": 1.0049,
+      "step": 579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.899335436048813e-06,
+      "loss": 1.0229,
+      "step": 580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895156883462358e-06,
+      "loss": 0.9678,
+      "step": 581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89097139912716e-06,
+      "loss": 0.9634,
+      "step": 582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886778990491632e-06,
+      "loss": 0.957,
+      "step": 583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.882579665016502e-06,
+      "loss": 1.0576,
+      "step": 584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.878373430174812e-06,
+      "loss": 0.8521,
+      "step": 585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.874160293451903e-06,
+      "loss": 0.9912,
+      "step": 586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.869940262345393e-06,
+      "loss": 1.0454,
+      "step": 587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86571334436517e-06,
+      "loss": 0.9155,
+      "step": 588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86147954703338e-06,
+      "loss": 1.1035,
+      "step": 589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.857238877884409e-06,
+      "loss": 1.0034,
+      "step": 590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852991344464873e-06,
+      "loss": 0.9946,
+      "step": 591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848736954333603e-06,
+      "loss": 1.0308,
+      "step": 592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.844475715061634e-06,
+      "loss": 1.0518,
+      "step": 593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.840207634232185e-06,
+      "loss": 0.939,
+      "step": 594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.835932719440658e-06,
+      "loss": 0.9556,
+      "step": 595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.831650978294608e-06,
+      "loss": 1.0024,
+      "step": 596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827362418413744e-06,
+      "loss": 0.9819,
+      "step": 597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.823067047429908e-06,
+      "loss": 0.8784,
+      "step": 598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81876487298706e-06,
+      "loss": 0.9526,
+      "step": 599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81445590274127e-06,
+      "loss": 0.9136,
+      "step": 600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.810140144360701e-06,
+      "loss": 1.0503,
+      "step": 601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.8058176055256e-06,
+      "loss": 0.9839,
+      "step": 602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801488293928274e-06,
+      "loss": 1.0615,
+      "step": 603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.797152217273082e-06,
+      "loss": 0.9878,
+      "step": 604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79280938327643e-06,
+      "loss": 1.0112,
+      "step": 605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788459799666745e-06,
+      "loss": 0.9966,
+      "step": 606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784103474184463e-06,
+      "loss": 0.9668,
+      "step": 607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77974041458202e-06,
+      "loss": 0.9819,
+      "step": 608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.775370628623837e-06,
+      "loss": 0.9663,
+      "step": 609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7709941240863e-06,
+      "loss": 1.0435,
+      "step": 610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76661090875776e-06,
+      "loss": 0.9829,
+      "step": 611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.762220990438504e-06,
+      "loss": 0.9814,
+      "step": 612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.757824376940748e-06,
+      "loss": 0.9351,
+      "step": 613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.753421076088621e-06,
+      "loss": 1.0449,
+      "step": 614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74901109571816e-06,
+      "loss": 1.0142,
+      "step": 615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.744594443677284e-06,
+      "loss": 0.9634,
+      "step": 616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740171127825783e-06,
+      "loss": 1.0488,
+      "step": 617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.735741156035306e-06,
+      "loss": 0.8994,
+      "step": 618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73130453618935e-06,
+      "loss": 0.9893,
+      "step": 619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726861276183242e-06,
+      "loss": 0.9893,
+      "step": 620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.722411383924126e-06,
+      "loss": 1.0601,
+      "step": 621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.717954867330943e-06,
+      "loss": 0.9507,
+      "step": 622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71349173433443e-06,
+      "loss": 0.9551,
+      "step": 623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709021992877096e-06,
+      "loss": 1.0459,
+      "step": 624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70454565091321e-06,
+      "loss": 1.0229,
+      "step": 625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.700062716408784e-06,
+      "loss": 1.0664,
+      "step": 626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.695573197341566e-06,
+      "loss": 0.9766,
+      "step": 627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.691077101701024e-06,
+      "loss": 1.0376,
+      "step": 628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686574437488322e-06,
+      "loss": 1.0068,
+      "step": 629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68206521271632e-06,
+      "loss": 1.0381,
+      "step": 630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.677549435409548e-06,
+      "loss": 1.0005,
+      "step": 631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673027113604202e-06,
+      "loss": 1.0078,
+      "step": 632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.668498255348119e-06,
+      "loss": 1.0581,
+      "step": 633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.663962868700773e-06,
+      "loss": 1.0742,
+      "step": 634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659420961733255e-06,
+      "loss": 0.9878,
+      "step": 635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.654872542528256e-06,
+      "loss": 0.9722,
+      "step": 636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.650317619180057e-06,
+      "loss": 0.9814,
+      "step": 637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.645756199794519e-06,
+      "loss": 0.9463,
+      "step": 638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.641188292489054e-06,
+      "loss": 0.9917,
+      "step": 639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636613905392628e-06,
+      "loss": 1.064,
+      "step": 640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632033046645735e-06,
+      "loss": 1.167,
+      "step": 641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.627445724400387e-06,
+      "loss": 0.9395,
+      "step": 642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.622851946820094e-06,
+      "loss": 1.0498,
+      "step": 643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.618251722079861e-06,
+      "loss": 1.124,
+      "step": 644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.61364505836616e-06,
+      "loss": 1.0093,
+      "step": 645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.609031963876924e-06,
+      "loss": 1.04,
+      "step": 646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.604412446821531e-06,
+      "loss": 1.0107,
+      "step": 647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599786515420789e-06,
+      "loss": 0.9482,
+      "step": 648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.595154177906915e-06,
+      "loss": 1.0635,
+      "step": 649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.590515442523537e-06,
+      "loss": 0.8652,
+      "step": 650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.585870317525655e-06,
+      "loss": 0.9521,
+      "step": 651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.581218811179655e-06,
+      "loss": 1.0122,
+      "step": 652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.576560931763263e-06,
+      "loss": 0.9878,
+      "step": 653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57189668756556e-06,
+      "loss": 1.0239,
+      "step": 654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.567226086886948e-06,
+      "loss": 0.9258,
+      "step": 655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.56254913803914e-06,
+      "loss": 1.0576,
+      "step": 656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557865849345146e-06,
+      "loss": 0.9814,
+      "step": 657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553176229139262e-06,
+      "loss": 0.9375,
+      "step": 658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.548480285767044e-06,
+      "loss": 1.0762,
+      "step": 659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543778027585313e-06,
+      "loss": 0.9351,
+      "step": 660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.539069462962115e-06,
+      "loss": 1.0894,
+      "step": 661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.534354600276724e-06,
+      "loss": 0.939,
+      "step": 662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.529633447919622e-06,
+      "loss": 0.9922,
+      "step": 663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.524906014292488e-06,
+      "loss": 1.0767,
+      "step": 664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52017230780817e-06,
+      "loss": 0.9224,
+      "step": 665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.515432336890688e-06,
+      "loss": 0.9834,
+      "step": 666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.510686109975202e-06,
+      "loss": 1.1299,
+      "step": 667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.505933635508012e-06,
+      "loss": 0.9253,
+      "step": 668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501174921946533e-06,
+      "loss": 1.0151,
+      "step": 669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.496409977759281e-06,
+      "loss": 1.0264,
+      "step": 670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.491638811425865e-06,
+      "loss": 0.9287,
+      "step": 671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.486861431436962e-06,
+      "loss": 0.9775,
+      "step": 672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.48207784629431e-06,
+      "loss": 1.0356,
+      "step": 673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477288064510685e-06,
+      "loss": 1.0659,
+      "step": 674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.472492094609898e-06,
+      "loss": 0.9971,
+      "step": 675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.467689945126764e-06,
+      "loss": 0.8994,
+      "step": 676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.462881624607101e-06,
+      "loss": 0.9556,
+      "step": 677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.458067141607706e-06,
+      "loss": 0.9893,
+      "step": 678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.453246504696345e-06,
+      "loss": 1.0103,
+      "step": 679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44841972245173e-06,
+      "loss": 1.021,
+      "step": 680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.443586803463517e-06,
+      "loss": 1.0864,
+      "step": 681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.438747756332278e-06,
+      "loss": 0.9282,
+      "step": 682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.433902589669489e-06,
+      "loss": 1.0098,
+      "step": 683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42905131209752e-06,
+      "loss": 1.1221,
+      "step": 684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.424193932249614e-06,
+      "loss": 1.0112,
+      "step": 685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.419330458769874e-06,
+      "loss": 0.9922,
+      "step": 686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.414460900313249e-06,
+      "loss": 0.9839,
+      "step": 687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.409585265545509e-06,
+      "loss": 0.9648,
+      "step": 688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.404703563143249e-06,
+      "loss": 0.9692,
+      "step": 689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.399815801793852e-06,
+      "loss": 1.0215,
+      "step": 690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39492199019549e-06,
+      "loss": 0.9834,
+      "step": 691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.390022137057096e-06,
+      "loss": 0.9482,
+      "step": 692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.385116251098361e-06,
+      "loss": 0.9614,
+      "step": 693
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.380204341049706e-06,
+      "loss": 1.0288,
+      "step": 694
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.375286415652275e-06,
+      "loss": 0.8896,
+      "step": 695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.370362483657917e-06,
+      "loss": 0.9951,
+      "step": 696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36543255382917e-06,
+      "loss": 0.9971,
+      "step": 697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.360496634939243e-06,
+      "loss": 0.9639,
+      "step": 698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35555473577201e-06,
+      "loss": 1.0156,
+      "step": 699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35060686512198e-06,
+      "loss": 1.0806,
+      "step": 700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.345653031794292e-06,
+      "loss": 0.9419,
+      "step": 701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.340693244604696e-06,
+      "loss": 1.0557,
+      "step": 702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.335727512379535e-06,
+      "loss": 1.0229,
+      "step": 703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.330755843955736e-06,
+      "loss": 1.0156,
+      "step": 704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32577824818079e-06,
+      "loss": 0.9541,
+      "step": 705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.320794733912727e-06,
+      "loss": 1.061,
+      "step": 706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315805310020121e-06,
+      "loss": 1.062,
+      "step": 707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.310809985382059e-06,
+      "loss": 1.0117,
+      "step": 708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.305808768888123e-06,
+      "loss": 0.8916,
+      "step": 709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.300801669438391e-06,
+      "loss": 0.9551,
+      "step": 710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.2957886959434e-06,
+      "loss": 1.064,
+      "step": 711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.290769857324144e-06,
+      "loss": 0.998,
+      "step": 712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.285745162512057e-06,
+      "loss": 0.9751,
+      "step": 713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280714620448991e-06,
+      "loss": 0.9111,
+      "step": 714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.275678240087206e-06,
+      "loss": 0.8643,
+      "step": 715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.270636030389352e-06,
+      "loss": 1.002,
+      "step": 716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.265588000328449e-06,
+      "loss": 0.9229,
+      "step": 717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.260534158887878e-06,
+      "loss": 0.9482,
+      "step": 718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.255474515061362e-06,
+      "loss": 0.9448,
+      "step": 719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.250409077852947e-06,
+      "loss": 0.9951,
+      "step": 720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245337856276996e-06,
+      "loss": 0.98,
+      "step": 721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.240260859358156e-06,
+      "loss": 1.0054,
+      "step": 722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.235178096131355e-06,
+      "loss": 1.1089,
+      "step": 723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.23008957564179e-06,
+      "loss": 1.0508,
+      "step": 724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.224995306944892e-06,
+      "loss": 1.061,
+      "step": 725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.219895299106328e-06,
+      "loss": 1.0752,
+      "step": 726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.214789561201979e-06,
+      "loss": 1.0776,
+      "step": 727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.20967810231792e-06,
+      "loss": 1.0894,
+      "step": 728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.204560931550406e-06,
+      "loss": 1.0879,
+      "step": 729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.199438058005864e-06,
+      "loss": 1.0371,
+      "step": 730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19430949080086e-06,
+      "loss": 0.939,
+      "step": 731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.189175239062098e-06,
+      "loss": 0.9204,
+      "step": 732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184035311926397e-06,
+      "loss": 0.9033,
+      "step": 733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.178889718540677e-06,
+      "loss": 0.978,
+      "step": 734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17373846806194e-06,
+      "loss": 1.0688,
+      "step": 735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.168581569657253e-06,
+      "loss": 0.9771,
+      "step": 736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.163419032503738e-06,
+      "loss": 1.1191,
+      "step": 737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.158250865788552e-06,
+      "loss": 1.1001,
+      "step": 738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.153077078708867e-06,
+      "loss": 1.0522,
+      "step": 739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.147897680471855e-06,
+      "loss": 0.9839,
+      "step": 740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.142712680294681e-06,
+      "loss": 0.9399,
+      "step": 741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.137522087404474e-06,
+      "loss": 0.9824,
+      "step": 742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.132325911038318e-06,
+      "loss": 1.1328,
+      "step": 743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.127124160443229e-06,
+      "loss": 0.9482,
+      "step": 744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.121916844876145e-06,
+      "loss": 1.0278,
+      "step": 745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.116703973603911e-06,
+      "loss": 1.0444,
+      "step": 746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.111485555903253e-06,
+      "loss": 0.9189,
+      "step": 747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.106261601060773e-06,
+      "loss": 0.958,
+      "step": 748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10103211837292e-06,
+      "loss": 1.0522,
+      "step": 749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09579711714599e-06,
+      "loss": 0.9902,
+      "step": 750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.090556606696088e-06,
+      "loss": 0.9644,
+      "step": 751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.08531059634913e-06,
+      "loss": 0.9165,
+      "step": 752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.080059095440822e-06,
+      "loss": 0.9263,
+      "step": 753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.074802113316633e-06,
+      "loss": 0.9609,
+      "step": 754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.069539659331793e-06,
+      "loss": 1.0239,
+      "step": 755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.064271742851267e-06,
+      "loss": 0.9316,
+      "step": 756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05899837324974e-06,
+      "loss": 0.9253,
+      "step": 757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.053719559911605e-06,
+      "loss": 0.9854,
+      "step": 758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048435312230939e-06,
+      "loss": 1.0,
+      "step": 759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.043145639611488e-06,
+      "loss": 0.918,
+      "step": 760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.037850551466658e-06,
+      "loss": 0.9751,
+      "step": 761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.032550057219488e-06,
+      "loss": 1.0581,
+      "step": 762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.027244166302641e-06,
+      "loss": 1.043,
+      "step": 763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.021932888158382e-06,
+      "loss": 0.9976,
+      "step": 764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.01661623223856e-06,
+      "loss": 0.9478,
+      "step": 765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.011294208004596e-06,
+      "loss": 1.0244,
+      "step": 766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.005966824927469e-06,
+      "loss": 0.9536,
+      "step": 767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00063409248769e-06,
+      "loss": 0.9956,
+      "step": 768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99529602017529e-06,
+      "loss": 1.0449,
+      "step": 769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.989952617489802e-06,
+      "loss": 0.978,
+      "step": 770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.984603893940247e-06,
+      "loss": 0.9258,
+      "step": 771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.979249859045117e-06,
+      "loss": 1.0151,
+      "step": 772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973890522332348e-06,
+      "loss": 1.0327,
+      "step": 773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.968525893339319e-06,
+      "loss": 0.9941,
+      "step": 774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96315598161282e-06,
+      "loss": 0.9766,
+      "step": 775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.957780796709052e-06,
+      "loss": 0.9932,
+      "step": 776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95240034819359e-06,
+      "loss": 1.0166,
+      "step": 777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.94701464564138e-06,
+      "loss": 0.9619,
+      "step": 778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.94162369863672e-06,
+      "loss": 0.9448,
+      "step": 779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.936227516773238e-06,
+      "loss": 1.0366,
+      "step": 780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.930826109653882e-06,
+      "loss": 1.0967,
+      "step": 781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.925419486890888e-06,
+      "loss": 1.0166,
+      "step": 782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92000765810579e-06,
+      "loss": 0.9829,
+      "step": 783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914590632929372e-06,
+      "loss": 1.0811,
+      "step": 784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.909168421001675e-06,
+      "loss": 1.002,
+      "step": 785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90374103197196e-06,
+      "loss": 0.9805,
+      "step": 786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.898308475498717e-06,
+      "loss": 1.0161,
+      "step": 787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.892870761249616e-06,
+      "loss": 0.8418,
+      "step": 788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.887427898901514e-06,
+      "loss": 0.9561,
+      "step": 789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.881979898140428e-06,
+      "loss": 0.9595,
+      "step": 790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.876526768661518e-06,
+      "loss": 1.0063,
+      "step": 791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.871068520169071e-06,
+      "loss": 0.8882,
+      "step": 792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.865605162376485e-06,
+      "loss": 0.96,
+      "step": 793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.860136705006254e-06,
+      "loss": 1.0225,
+      "step": 794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.854663157789937e-06,
+      "loss": 0.9971,
+      "step": 795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84918453046816e-06,
+      "loss": 0.9209,
+      "step": 796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.843700832790585e-06,
+      "loss": 0.9829,
+      "step": 797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.838212074515899e-06,
+      "loss": 0.9727,
+      "step": 798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.832718265411795e-06,
+      "loss": 1.0581,
+      "step": 799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.827219415254952e-06,
+      "loss": 0.9609,
+      "step": 800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.821715533831022e-06,
+      "loss": 0.9165,
+      "step": 801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.816206630934611e-06,
+      "loss": 1.041,
+      "step": 802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.810692716369258e-06,
+      "loss": 1.0225,
+      "step": 803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.805173799947425e-06,
+      "loss": 1.0024,
+      "step": 804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.799649891490472e-06,
+      "loss": 0.9766,
+      "step": 805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.794121000828643e-06,
+      "loss": 0.9243,
+      "step": 806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78858713780105e-06,
+      "loss": 0.8877,
+      "step": 807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.783048312255653e-06,
+      "loss": 0.9023,
+      "step": 808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.777504534049242e-06,
+      "loss": 0.9272,
+      "step": 809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.771955813047422e-06,
+      "loss": 1.0151,
+      "step": 810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766402159124595e-06,
+      "loss": 0.9448,
+      "step": 811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.760843582163938e-06,
+      "loss": 1.0181,
+      "step": 812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.755280092057393e-06,
+      "loss": 1.0229,
+      "step": 813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.749711698705642e-06,
+      "loss": 1.0513,
+      "step": 814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.744138412018094e-06,
+      "loss": 0.9082,
+      "step": 815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.738560241912867e-06,
+      "loss": 1.0435,
+      "step": 816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.732977198316772e-06,
+      "loss": 0.9468,
+      "step": 817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.727389291165283e-06,
+      "loss": 0.9165,
+      "step": 818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.721796530402542e-06,
+      "loss": 0.9805,
+      "step": 819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.716198925981316e-06,
+      "loss": 0.9199,
+      "step": 820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.710596487863003e-06,
+      "loss": 0.8389,
+      "step": 821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.704989226017596e-06,
+      "loss": 0.9673,
+      "step": 822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699377150423673e-06,
+      "loss": 0.9629,
+      "step": 823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.693760271068378e-06,
+      "loss": 0.9487,
+      "step": 824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68813859794741e-06,
+      "loss": 0.9668,
+      "step": 825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.682512141064988e-06,
+      "loss": 0.9146,
+      "step": 826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.676880910433854e-06,
+      "loss": 0.8267,
+      "step": 827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67124491607524e-06,
+      "loss": 0.9688,
+      "step": 828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.665604168018856e-06,
+      "loss": 0.8994,
+      "step": 829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.659958676302875e-06,
+      "loss": 0.917,
+      "step": 830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.654308450973906e-06,
+      "loss": 1.002,
+      "step": 831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.648653502086987e-06,
+      "loss": 0.8892,
+      "step": 832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.642993839705557e-06,
+      "loss": 0.9673,
+      "step": 833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.637329473901449e-06,
+      "loss": 1.041,
+      "step": 834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.631660414754862e-06,
+      "loss": 1.0166,
+      "step": 835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.625986672354349e-06,
+      "loss": 0.856,
+      "step": 836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.620308256796795e-06,
+      "loss": 0.9312,
+      "step": 837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.614625178187402e-06,
+      "loss": 0.8735,
+      "step": 838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.608937446639672e-06,
+      "loss": 1.0303,
+      "step": 839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.603245072275389e-06,
+      "loss": 0.9497,
+      "step": 840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.597548065224594e-06,
+      "loss": 0.8867,
+      "step": 841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5918464356255735e-06,
+      "loss": 1.0796,
+      "step": 842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.586140193624842e-06,
+      "loss": 1.0205,
+      "step": 843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.580429349377123e-06,
+      "loss": 1.0049,
+      "step": 844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.57471391304533e-06,
+      "loss": 0.9053,
+      "step": 845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.568993894800545e-06,
+      "loss": 0.9961,
+      "step": 846
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.563269304822005e-06,
+      "loss": 1.1113,
+      "step": 847
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.557540153297086e-06,
+      "loss": 0.9253,
+      "step": 848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.551806450421278e-06,
+      "loss": 1.0703,
+      "step": 849
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.546068206398175e-06,
+      "loss": 0.8521,
+      "step": 850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.540325431439446e-06,
+      "loss": 0.8442,
+      "step": 851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.534578135764827e-06,
+      "loss": 0.8726,
+      "step": 852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.528826329602099e-06,
+      "loss": 0.9897,
+      "step": 853
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52307002318707e-06,
+      "loss": 1.0049,
+      "step": 854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.517309226763556e-06,
+      "loss": 0.9463,
+      "step": 855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.511543950583362e-06,
+      "loss": 0.9453,
+      "step": 856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.505774204906266e-06,
+      "loss": 1.0381,
+      "step": 857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.9297,
+      "step": 858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.494221346140234e-06,
+      "loss": 1.0542,
+      "step": 859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.488438253610551e-06,
+      "loss": 0.9648,
+      "step": 860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.482650732702436e-06,
+      "loss": 1.0078,
+      "step": 861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.476858793715252e-06,
+      "loss": 0.9648,
+      "step": 862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.471062446956226e-06,
+      "loss": 1.0938,
+      "step": 863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.465261702740431e-06,
+      "loss": 0.9741,
+      "step": 864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.459456571390762e-06,
+      "loss": 0.8853,
+      "step": 865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.453647063237921e-06,
+      "loss": 0.9956,
+      "step": 866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.447833188620405e-06,
+      "loss": 0.9683,
+      "step": 867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.442014957884473e-06,
+      "loss": 0.9253,
+      "step": 868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4361923813841416e-06,
+      "loss": 0.9375,
+      "step": 869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.430365469481161e-06,
+      "loss": 0.9673,
+      "step": 870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.424534232544993e-06,
+      "loss": 0.9775,
+      "step": 871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.418698680952802e-06,
+      "loss": 0.9478,
+      "step": 872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.412858825089423e-06,
+      "loss": 0.9097,
+      "step": 873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.407014675347356e-06,
+      "loss": 1.1445,
+      "step": 874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.401166242126743e-06,
+      "loss": 0.9673,
+      "step": 875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3953135358353446e-06,
+      "loss": 1.0454,
+      "step": 876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.389456566888529e-06,
+      "loss": 0.9233,
+      "step": 877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.38359534570925e-06,
+      "loss": 0.9214,
+      "step": 878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.377729882728026e-06,
+      "loss": 0.9976,
+      "step": 879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.37186018838293e-06,
+      "loss": 0.9023,
+      "step": 880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.365986273119555e-06,
+      "loss": 0.957,
+      "step": 881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.360108147391014e-06,
+      "loss": 1.0107,
+      "step": 882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3542258216579136e-06,
+      "loss": 1.0474,
+      "step": 883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3483393063883245e-06,
+      "loss": 0.9653,
+      "step": 884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.342448612057785e-06,
+      "loss": 0.8965,
+      "step": 885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.336553749149263e-06,
+      "loss": 1.0186,
+      "step": 886
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.330654728153148e-06,
+      "loss": 1.0049,
+      "step": 887
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.324751559567228e-06,
+      "loss": 0.9736,
+      "step": 888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.318844253896671e-06,
+      "loss": 0.9302,
+      "step": 889
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.312932821654009e-06,
+      "loss": 0.998,
+      "step": 890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.307017273359116e-06,
+      "loss": 1.0645,
+      "step": 891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.301097619539193e-06,
+      "loss": 0.9375,
+      "step": 892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.295173870728744e-06,
+      "loss": 1.0137,
+      "step": 893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.289246037469564e-06,
+      "loss": 1.0078,
+      "step": 894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.283314130310716e-06,
+      "loss": 0.9482,
+      "step": 895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.277378159808508e-06,
+      "loss": 0.9272,
+      "step": 896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.271438136526484e-06,
+      "loss": 1.0396,
+      "step": 897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.265494071035401e-06,
+      "loss": 0.9883,
+      "step": 898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.259545973913206e-06,
+      "loss": 0.9878,
+      "step": 899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.253593855745023e-06,
+      "loss": 1.0137,
+      "step": 900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.247637727123127e-06,
+      "loss": 1.0039,
+      "step": 901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2416775986469366e-06,
+      "loss": 1.0195,
+      "step": 902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.235713480922988e-06,
+      "loss": 0.9849,
+      "step": 903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.229745384564909e-06,
+      "loss": 0.9287,
+      "step": 904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.223773320193418e-06,
+      "loss": 0.958,
+      "step": 905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.217797298436285e-06,
+      "loss": 0.999,
+      "step": 906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.21181732992833e-06,
+      "loss": 0.8701,
+      "step": 907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.205833425311394e-06,
+      "loss": 1.0552,
+      "step": 908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.19984559523432e-06,
+      "loss": 1.0107,
+      "step": 909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.193853850352941e-06,
+      "loss": 0.9082,
+      "step": 910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1878582013300546e-06,
+      "loss": 0.9399,
+      "step": 911
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.181858658835405e-06,
+      "loss": 0.9624,
+      "step": 912
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.175855233545669e-06,
+      "loss": 1.0015,
+      "step": 913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.169847936144427e-06,
+      "loss": 1.0288,
+      "step": 914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.163836777322157e-06,
+      "loss": 0.9937,
+      "step": 915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.157821767776203e-06,
+      "loss": 1.0015,
+      "step": 916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.151802918210765e-06,
+      "loss": 1.0615,
+      "step": 917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.145780239336878e-06,
+      "loss": 0.9224,
+      "step": 918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.139753741872385e-06,
+      "loss": 0.9775,
+      "step": 919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.133723436541932e-06,
+      "loss": 0.9683,
+      "step": 920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.127689334076937e-06,
+      "loss": 0.8789,
+      "step": 921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.121651445215577e-06,
+      "loss": 1.0479,
+      "step": 922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.115609780702767e-06,
+      "loss": 0.9609,
+      "step": 923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.109564351290139e-06,
+      "loss": 0.9814,
+      "step": 924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.10351516773603e-06,
+      "loss": 1.0518,
+      "step": 925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.097462240805451e-06,
+      "loss": 0.9844,
+      "step": 926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.091405581270079e-06,
+      "loss": 0.998,
+      "step": 927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.085345199908234e-06,
+      "loss": 0.9956,
+      "step": 928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.079281107504858e-06,
+      "loss": 0.9473,
+      "step": 929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.073213314851497e-06,
+      "loss": 0.9453,
+      "step": 930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.067141832746279e-06,
+      "loss": 0.9224,
+      "step": 931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.061066671993905e-06,
+      "loss": 1.0205,
+      "step": 932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0549878434056155e-06,
+      "loss": 0.8926,
+      "step": 933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.048905357799181e-06,
+      "loss": 1.0083,
+      "step": 934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.04281922599888e-06,
+      "loss": 0.8081,
+      "step": 935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.036729458835481e-06,
+      "loss": 0.8882,
+      "step": 936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.030636067146217e-06,
+      "loss": 1.0371,
+      "step": 937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.024539061774778e-06,
+      "loss": 0.9409,
+      "step": 938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.018438453571279e-06,
+      "loss": 0.9946,
+      "step": 939
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.01233425339225e-06,
+      "loss": 0.9517,
+      "step": 940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.006226472100611e-06,
+      "loss": 0.979,
+      "step": 941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.000115120565657e-06,
+      "loss": 0.9595,
+      "step": 942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.994000209663037e-06,
+      "loss": 0.875,
+      "step": 943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.987881750274729e-06,
+      "loss": 0.9819,
+      "step": 944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.981759753289032e-06,
+      "loss": 0.9546,
+      "step": 945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.975634229600539e-06,
+      "loss": 1.0464,
+      "step": 946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.969505190110114e-06,
+      "loss": 0.8813,
+      "step": 947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9633726457248864e-06,
+      "loss": 1.0103,
+      "step": 948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.957236607358216e-06,
+      "loss": 0.9585,
+      "step": 949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.951097085929681e-06,
+      "loss": 0.8755,
+      "step": 950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.944954092365061e-06,
+      "loss": 1.0547,
+      "step": 951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.938807637596315e-06,
+      "loss": 0.9302,
+      "step": 952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.932657732561556e-06,
+      "loss": 0.9692,
+      "step": 953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.926504388205043e-06,
+      "loss": 0.9077,
+      "step": 954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.920347615477153e-06,
+      "loss": 0.9541,
+      "step": 955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.914187425334363e-06,
+      "loss": 1.0713,
+      "step": 956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.908023828739233e-06,
+      "loss": 1.041,
+      "step": 957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.901856836660386e-06,
+      "loss": 0.9912,
+      "step": 958
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.895686460072486e-06,
+      "loss": 0.9902,
+      "step": 959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.889512709956221e-06,
+      "loss": 1.0151,
+      "step": 960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.883335597298279e-06,
+      "loss": 0.9844,
+      "step": 961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8771551330913396e-06,
+      "loss": 0.9844,
+      "step": 962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.870971328334038e-06,
+      "loss": 0.915,
+      "step": 963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.864784194030958e-06,
+      "loss": 0.9961,
+      "step": 964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.858593741192612e-06,
+      "loss": 0.9775,
+      "step": 965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.852399980835413e-06,
+      "loss": 0.9897,
+      "step": 966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.846202923981661e-06,
+      "loss": 0.9829,
+      "step": 967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.840002581659522e-06,
+      "loss": 0.9653,
+      "step": 968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.833798964903013e-06,
+      "loss": 0.9644,
+      "step": 969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.827592084751975e-06,
+      "loss": 0.9429,
+      "step": 970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.821381952252053e-06,
+      "loss": 0.9731,
+      "step": 971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.815168578454686e-06,
+      "loss": 1.0845,
+      "step": 972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.808951974417077e-06,
+      "loss": 0.9004,
+      "step": 973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80273215120218e-06,
+      "loss": 1.002,
+      "step": 974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.796509119878678e-06,
+      "loss": 0.9858,
+      "step": 975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.790282891520958e-06,
+      "loss": 0.9033,
+      "step": 976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.784053477209105e-06,
+      "loss": 0.9917,
+      "step": 977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7778208880288644e-06,
+      "loss": 0.9531,
+      "step": 978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7715851350716375e-06,
+      "loss": 1.0283,
+      "step": 979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.765346229434455e-06,
+      "loss": 1.0249,
+      "step": 980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.759104182219958e-06,
+      "loss": 1.0146,
+      "step": 981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.752859004536376e-06,
+      "loss": 0.9434,
+      "step": 982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.746610707497511e-06,
+      "loss": 0.9941,
+      "step": 983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.740359302222716e-06,
+      "loss": 0.9907,
+      "step": 984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.734104799836878e-06,
+      "loss": 0.9268,
+      "step": 985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.727847211470391e-06,
+      "loss": 1.0078,
+      "step": 986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.721586548259142e-06,
+      "loss": 0.9854,
+      "step": 987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.715322821344495e-06,
+      "loss": 0.9814,
+      "step": 988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.709056041873256e-06,
+      "loss": 1.0063,
+      "step": 989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.702786220997674e-06,
+      "loss": 0.9888,
+      "step": 990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696513369875403e-06,
+      "loss": 0.8789,
+      "step": 991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6902374996694945e-06,
+      "loss": 0.9233,
+      "step": 992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.683958621548368e-06,
+      "loss": 0.8691,
+      "step": 993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6776767466857974e-06,
+      "loss": 1.0063,
+      "step": 994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.671391886260895e-06,
+      "loss": 0.9702,
+      "step": 995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.665104051458077e-06,
+      "loss": 0.9082,
+      "step": 996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.65881325346706e-06,
+      "loss": 0.957,
+      "step": 997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.652519503482829e-06,
+      "loss": 0.8896,
+      "step": 998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.646222812705625e-06,
+      "loss": 0.8652,
+      "step": 999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.639923192340923e-06,
+      "loss": 0.9028,
+      "step": 1000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.633620653599409e-06,
+      "loss": 0.8496,
+      "step": 1001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.627315207696966e-06,
+      "loss": 0.937,
+      "step": 1002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.621006865854645e-06,
+      "loss": 0.8994,
+      "step": 1003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.614695639298654e-06,
+      "loss": 0.9658,
+      "step": 1004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.608381539260338e-06,
+      "loss": 0.9956,
+      "step": 1005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.602064576976148e-06,
+      "loss": 0.9585,
+      "step": 1006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.595744763687635e-06,
+      "loss": 0.9512,
+      "step": 1007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58942211064142e-06,
+      "loss": 0.8765,
+      "step": 1008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.583096629089178e-06,
+      "loss": 0.9375,
+      "step": 1009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5767683302876214e-06,
+      "loss": 0.9263,
+      "step": 1010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.570437225498471e-06,
+      "loss": 0.9653,
+      "step": 1011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.564103325988442e-06,
+      "loss": 1.0127,
+      "step": 1012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.557766643029226e-06,
+      "loss": 0.9712,
+      "step": 1013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.551427187897465e-06,
+      "loss": 0.9761,
+      "step": 1014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 0.9717,
+      "step": 1015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.53874000624753e-06,
+      "loss": 1.0298,
+      "step": 1016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.532392302307228e-06,
+      "loss": 1.0454,
+      "step": 1017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.526041871350086e-06,
+      "loss": 0.8159,
+      "step": 1018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.519688724677214e-06,
+      "loss": 1.0137,
+      "step": 1019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.513332873594552e-06,
+      "loss": 0.9209,
+      "step": 1020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.506974329412855e-06,
+      "loss": 0.8604,
+      "step": 1021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5006131034476695e-06,
+      "loss": 0.8945,
+      "step": 1022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.494249207019317e-06,
+      "loss": 0.9917,
+      "step": 1023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.487882651452867e-06,
+      "loss": 0.9453,
+      "step": 1024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.481513448078125e-06,
+      "loss": 0.9927,
+      "step": 1025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.475141608229605e-06,
+      "loss": 0.9351,
+      "step": 1026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.468767143246515e-06,
+      "loss": 0.9194,
+      "step": 1027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.462390064472736e-06,
+      "loss": 0.9727,
+      "step": 1028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.456010383256799e-06,
+      "loss": 1.0771,
+      "step": 1029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.449628110951864e-06,
+      "loss": 0.9707,
+      "step": 1030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.443243258915703e-06,
+      "loss": 1.0093,
+      "step": 1031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.436855838510684e-06,
+      "loss": 0.9155,
+      "step": 1032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.43046586110374e-06,
+      "loss": 0.9399,
+      "step": 1033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.424073338066354e-06,
+      "loss": 1.0039,
+      "step": 1034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.417678280774542e-06,
+      "loss": 0.9414,
+      "step": 1035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.411280700608831e-06,
+      "loss": 1.0278,
+      "step": 1036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.404880608954231e-06,
+      "loss": 0.9849,
+      "step": 1037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3984780172002335e-06,
+      "loss": 0.9531,
+      "step": 1038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3920729367407645e-06,
+      "loss": 1.0234,
+      "step": 1039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3856653789741905e-06,
+      "loss": 0.9541,
+      "step": 1040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3792553553032785e-06,
+      "loss": 0.9272,
+      "step": 1041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.372842877135191e-06,
+      "loss": 0.9341,
+      "step": 1042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.366427955881453e-06,
+      "loss": 0.9453,
+      "step": 1043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.36001060295794e-06,
+      "loss": 1.0093,
+      "step": 1044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.353590829784853e-06,
+      "loss": 0.9355,
+      "step": 1045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.347168647786701e-06,
+      "loss": 0.9429,
+      "step": 1046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.340744068392279e-06,
+      "loss": 1.0371,
+      "step": 1047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3343171030346525e-06,
+      "loss": 1.0415,
+      "step": 1048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.327887763151125e-06,
+      "loss": 0.8677,
+      "step": 1049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.321456060183233e-06,
+      "loss": 1.0234,
+      "step": 1050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.315022005576713e-06,
+      "loss": 1.1011,
+      "step": 1051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.308585610781491e-06,
+      "loss": 0.979,
+      "step": 1052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.302146887251656e-06,
+      "loss": 0.9785,
+      "step": 1053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.295705846445439e-06,
+      "loss": 0.9893,
+      "step": 1054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.289262499825197e-06,
+      "loss": 0.9043,
+      "step": 1055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.282816858857389e-06,
+      "loss": 1.0532,
+      "step": 1056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.276368935012559e-06,
+      "loss": 0.9214,
+      "step": 1057
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.269918739765313e-06,
+      "loss": 0.9453,
+      "step": 1058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.263466284594295e-06,
+      "loss": 0.9106,
+      "step": 1059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.257011580982179e-06,
+      "loss": 0.9185,
+      "step": 1060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.250554640415631e-06,
+      "loss": 0.9316,
+      "step": 1061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2440954743853055e-06,
+      "loss": 1.0186,
+      "step": 1062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.237634094385814e-06,
+      "loss": 0.9751,
+      "step": 1063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.231170511915706e-06,
+      "loss": 0.9736,
+      "step": 1064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.224704738477456e-06,
+      "loss": 0.9111,
+      "step": 1065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.218236785577431e-06,
+      "loss": 0.9521,
+      "step": 1066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.211766664725879e-06,
+      "loss": 0.9282,
+      "step": 1067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.205294387436911e-06,
+      "loss": 0.9771,
+      "step": 1068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.198819965228468e-06,
+      "loss": 0.9653,
+      "step": 1069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.192343409622312e-06,
+      "loss": 0.9028,
+      "step": 1070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1858647321440014e-06,
+      "loss": 0.9087,
+      "step": 1071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1793839443228685e-06,
+      "loss": 0.8545,
+      "step": 1072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.172901057692007e-06,
+      "loss": 1.0005,
+      "step": 1073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.166416083788236e-06,
+      "loss": 0.9419,
+      "step": 1074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.159929034152098e-06,
+      "loss": 0.8833,
+      "step": 1075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.153439920327824e-06,
+      "loss": 0.9453,
+      "step": 1076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.14694875386332e-06,
+      "loss": 0.9336,
+      "step": 1077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.140455546310149e-06,
+      "loss": 0.8584,
+      "step": 1078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.133960309223496e-06,
+      "loss": 0.9585,
+      "step": 1079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1274630541621715e-06,
+      "loss": 0.9092,
+      "step": 1080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1209637926885635e-06,
+      "loss": 0.8945,
+      "step": 1081
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.114462536368639e-06,
+      "loss": 0.9863,
+      "step": 1082
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.107959296771915e-06,
+      "loss": 0.8418,
+      "step": 1083
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1014540854714324e-06,
+      "loss": 0.9478,
+      "step": 1084
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.094946914043747e-06,
+      "loss": 0.9883,
+      "step": 1085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0884377940688965e-06,
+      "loss": 0.9692,
+      "step": 1086
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.081926737130392e-06,
+      "loss": 0.9526,
+      "step": 1087
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.075413754815191e-06,
+      "loss": 1.0029,
+      "step": 1088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.068898858713673e-06,
+      "loss": 0.9795,
+      "step": 1089
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.062382060419628e-06,
+      "loss": 0.8687,
+      "step": 1090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0558633715302275e-06,
+      "loss": 0.9258,
+      "step": 1091
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.049342803646008e-06,
+      "loss": 1.0234,
+      "step": 1092
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.042820368370854e-06,
+      "loss": 0.9619,
+      "step": 1093
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0362960773119675e-06,
+      "loss": 0.9888,
+      "step": 1094
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0297699420798564e-06,
+      "loss": 0.8301,
+      "step": 1095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.023241974288308e-06,
+      "loss": 1.0015,
+      "step": 1096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.016712185554375e-06,
+      "loss": 0.8696,
+      "step": 1097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.010180587498347e-06,
+      "loss": 0.9849,
+      "step": 1098
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.003647191743734e-06,
+      "loss": 0.8643,
+      "step": 1099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.997112009917247e-06,
+      "loss": 1.0244,
+      "step": 1100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.990575053648772e-06,
+      "loss": 0.9844,
+      "step": 1101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.984036334571354e-06,
+      "loss": 1.0249,
+      "step": 1102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.977495864321179e-06,
+      "loss": 0.9492,
+      "step": 1103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.970953654537545e-06,
+      "loss": 1.0059,
+      "step": 1104
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9644097168628455e-06,
+      "loss": 0.9111,
+      "step": 1105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.957864062942551e-06,
+      "loss": 0.835,
+      "step": 1106
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9513167044251865e-06,
+      "loss": 1.0591,
+      "step": 1107
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944767652962309e-06,
+      "loss": 0.9497,
+      "step": 1108
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.938216920208488e-06,
+      "loss": 0.9775,
+      "step": 1109
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.931664517821286e-06,
+      "loss": 0.917,
+      "step": 1110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.925110457461236e-06,
+      "loss": 0.9731,
+      "step": 1111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9185547507918224e-06,
+      "loss": 0.9624,
+      "step": 1112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.911997409479461e-06,
+      "loss": 0.9453,
+      "step": 1113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.90543844519347e-06,
+      "loss": 0.9419,
+      "step": 1114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8988778696060636e-06,
+      "loss": 1.0498,
+      "step": 1115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8923156943923176e-06,
+      "loss": 0.9717,
+      "step": 1116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.885751931230159e-06,
+      "loss": 0.957,
+      "step": 1117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.879186591800339e-06,
+      "loss": 0.9038,
+      "step": 1118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.872619687786412e-06,
+      "loss": 0.9307,
+      "step": 1119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.866051230874719e-06,
+      "loss": 0.8696,
+      "step": 1120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8594812327543625e-06,
+      "loss": 0.9629,
+      "step": 1121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.85290970511719e-06,
+      "loss": 1.0405,
+      "step": 1122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8463366596577706e-06,
+      "loss": 0.9688,
+      "step": 1123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.839762108073372e-06,
+      "loss": 1.0347,
+      "step": 1124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.833186062063947e-06,
+      "loss": 1.0186,
+      "step": 1125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.826608533332101e-06,
+      "loss": 0.917,
+      "step": 1126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8200295335830836e-06,
+      "loss": 0.9834,
+      "step": 1127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8134490745247615e-06,
+      "loss": 0.9268,
+      "step": 1128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806867167867595e-06,
+      "loss": 0.9858,
+      "step": 1129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800283825324623e-06,
+      "loss": 0.8911,
+      "step": 1130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.793699058611442e-06,
+      "loss": 0.916,
+      "step": 1131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.787112879446177e-06,
+      "loss": 0.9756,
+      "step": 1132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.780525299549473e-06,
+      "loss": 0.9414,
+      "step": 1133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.773936330644462e-06,
+      "loss": 0.9863,
+      "step": 1134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.767345984456751e-06,
+      "loss": 0.979,
+      "step": 1135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7607542727144e-06,
+      "loss": 0.8818,
+      "step": 1136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.754161207147895e-06,
+      "loss": 0.8242,
+      "step": 1137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7475667994901316e-06,
+      "loss": 0.9336,
+      "step": 1138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.740971061476397e-06,
+      "loss": 0.9326,
+      "step": 1139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.734374004844342e-06,
+      "loss": 0.9268,
+      "step": 1140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.727775641333968e-06,
+      "loss": 1.0234,
+      "step": 1141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.721175982687597e-06,
+      "loss": 0.9072,
+      "step": 1142
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.714575040649863e-06,
+      "loss": 0.8374,
+      "step": 1143
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.707972826967675e-06,
+      "loss": 0.8989,
+      "step": 1144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.701369353390212e-06,
+      "loss": 1.062,
+      "step": 1145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.694764631668895e-06,
+      "loss": 0.9429,
+      "step": 1146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.68815867355736e-06,
+      "loss": 0.8965,
+      "step": 1147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.68155149081145e-06,
+      "loss": 0.938,
+      "step": 1148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.674943095189183e-06,
+      "loss": 0.8481,
+      "step": 1149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.668333498450736e-06,
+      "loss": 0.8813,
+      "step": 1150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.661722712358428e-06,
+      "loss": 0.9663,
+      "step": 1151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.655110748676687e-06,
+      "loss": 0.9653,
+      "step": 1152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.648497619172042e-06,
+      "loss": 0.9556,
+      "step": 1153
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.641883335613096e-06,
+      "loss": 0.9146,
+      "step": 1154
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.635267909770504e-06,
+      "loss": 0.9053,
+      "step": 1155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.628651353416957e-06,
+      "loss": 0.9121,
+      "step": 1156
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.622033678327153e-06,
+      "loss": 1.0386,
+      "step": 1157
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.615414896277786e-06,
+      "loss": 1.019,
+      "step": 1158
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.608795019047514e-06,
+      "loss": 0.8813,
+      "step": 1159
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6021740584169495e-06,
+      "loss": 0.9453,
+      "step": 1160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.595552026168631e-06,
+      "loss": 0.9639,
+      "step": 1161
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.588928934087003e-06,
+      "loss": 1.0303,
+      "step": 1162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5823047939584e-06,
+      "loss": 1.0181,
+      "step": 1163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.575679617571014e-06,
+      "loss": 1.0161,
+      "step": 1164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.569053416714887e-06,
+      "loss": 0.9512,
+      "step": 1165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.562426203181886e-06,
+      "loss": 0.8721,
+      "step": 1166
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.555797988765673e-06,
+      "loss": 0.9282,
+      "step": 1167
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.549168785261698e-06,
+      "loss": 0.9077,
+      "step": 1168
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5425386044671655e-06,
+      "loss": 1.0332,
+      "step": 1169
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.535907458181023e-06,
+      "loss": 0.9546,
+      "step": 1170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.529275358203938e-06,
+      "loss": 0.853,
+      "step": 1171
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.522642316338268e-06,
+      "loss": 0.9653,
+      "step": 1172
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.516008344388053e-06,
+      "loss": 0.8916,
+      "step": 1173
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.509373454158986e-06,
+      "loss": 0.9146,
+      "step": 1174
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.502737657458393e-06,
+      "loss": 0.9492,
+      "step": 1175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.496100966095217e-06,
+      "loss": 0.9702,
+      "step": 1176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.489463391879986e-06,
+      "loss": 0.9639,
+      "step": 1177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4828249466248095e-06,
+      "loss": 0.9102,
+      "step": 1178
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.476185642143337e-06,
+      "loss": 0.9092,
+      "step": 1179
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.469545490250753e-06,
+      "loss": 0.9824,
+      "step": 1180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.462904502763747e-06,
+      "loss": 0.959,
+      "step": 1181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.456262691500498e-06,
+      "loss": 0.9517,
+      "step": 1182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.44962006828065e-06,
+      "loss": 1.1494,
+      "step": 1183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.442976644925291e-06,
+      "loss": 0.9336,
+      "step": 1184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.436332433256932e-06,
+      "loss": 0.916,
+      "step": 1185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.429687445099493e-06,
+      "loss": 0.9443,
+      "step": 1186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.423041692278267e-06,
+      "loss": 0.9624,
+      "step": 1187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.416395186619915e-06,
+      "loss": 0.9336,
+      "step": 1188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.409747939952432e-06,
+      "loss": 0.8882,
+      "step": 1189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4030999641051366e-06,
+      "loss": 0.9731,
+      "step": 1190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.396451270908642e-06,
+      "loss": 0.9292,
+      "step": 1191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.38980187219484e-06,
+      "loss": 0.9912,
+      "step": 1192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.383151779796877e-06,
+      "loss": 0.9512,
+      "step": 1193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.376501005549132e-06,
+      "loss": 0.9653,
+      "step": 1194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3698495612872e-06,
+      "loss": 0.8599,
+      "step": 1195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3631974588478694e-06,
+      "loss": 1.0103,
+      "step": 1196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.356544710069095e-06,
+      "loss": 0.9551,
+      "step": 1197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3498913267899864e-06,
+      "loss": 0.9287,
+      "step": 1198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.34323732085078e-06,
+      "loss": 0.8584,
+      "step": 1199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.336582704092821e-06,
+      "loss": 0.8613,
+      "step": 1200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.329927488358544e-06,
+      "loss": 0.9287,
+      "step": 1201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3232716854914435e-06,
+      "loss": 0.9673,
+      "step": 1202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.316615307336065e-06,
+      "loss": 0.9409,
+      "step": 1203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.309958365737973e-06,
+      "loss": 0.8979,
+      "step": 1204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.303300872543739e-06,
+      "loss": 0.9092,
+      "step": 1205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.296642839600914e-06,
+      "loss": 1.0361,
+      "step": 1206
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.289984278758009e-06,
+      "loss": 0.9912,
+      "step": 1207
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.283325201864475e-06,
+      "loss": 0.918,
+      "step": 1208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2766656207706814e-06,
+      "loss": 0.8882,
+      "step": 1209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2700055473278935e-06,
+      "loss": 0.9194,
+      "step": 1210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.263344993388257e-06,
+      "loss": 0.8516,
+      "step": 1211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.256683970804768e-06,
+      "loss": 0.8823,
+      "step": 1212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.250022491431259e-06,
+      "loss": 0.9927,
+      "step": 1213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2433605671223745e-06,
+      "loss": 0.915,
+      "step": 1214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.236698209733549e-06,
+      "loss": 1.0161,
+      "step": 1215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2300354311209955e-06,
+      "loss": 0.9512,
+      "step": 1216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.223372243141665e-06,
+      "loss": 0.8745,
+      "step": 1217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.216708657653247e-06,
+      "loss": 0.9375,
+      "step": 1218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.210044686514129e-06,
+      "loss": 0.8843,
+      "step": 1219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.203380341583396e-06,
+      "loss": 0.9478,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.196715634720791e-06,
+      "loss": 0.8496,
+      "step": 1221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.190050577786699e-06,
+      "loss": 0.9268,
+      "step": 1222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.183385182642136e-06,
+      "loss": 1.0381,
+      "step": 1223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.17671946114871e-06,
+      "loss": 0.9077,
+      "step": 1224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.170053425168619e-06,
+      "loss": 0.9639,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1633870865646164e-06,
+      "loss": 0.9531,
+      "step": 1226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.156720457199992e-06,
+      "loss": 0.9209,
+      "step": 1227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.150053548938557e-06,
+      "loss": 0.9053,
+      "step": 1228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.143386373644618e-06,
+      "loss": 0.9443,
+      "step": 1229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.136718943182954e-06,
+      "loss": 0.9614,
+      "step": 1230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.130051269418804e-06,
+      "loss": 0.9307,
+      "step": 1231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1233833642178325e-06,
+      "loss": 0.9473,
+      "step": 1232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.116715239446121e-06,
+      "loss": 0.9653,
+      "step": 1233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.11004690697014e-06,
+      "loss": 1.0645,
+      "step": 1234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1033783786567295e-06,
+      "loss": 0.9614,
+      "step": 1235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.09670966637308e-06,
+      "loss": 0.8809,
+      "step": 1236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.090040781986706e-06,
+      "loss": 0.9482,
+      "step": 1237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.083371737365433e-06,
+      "loss": 0.9639,
+      "step": 1238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.076702544377366e-06,
+      "loss": 0.8799,
+      "step": 1239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.070033214890876e-06,
+      "loss": 0.9692,
+      "step": 1240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0633637607745815e-06,
+      "loss": 1.0024,
+      "step": 1241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.056694193897316e-06,
+      "loss": 0.9375,
+      "step": 1242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.050024526128118e-06,
+      "loss": 0.9502,
+      "step": 1243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.043354769336203e-06,
+      "loss": 0.8955,
+      "step": 1244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.036684935390946e-06,
+      "loss": 0.9912,
+      "step": 1245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.030015036161863e-06,
+      "loss": 0.9546,
+      "step": 1246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.023345083518577e-06,
+      "loss": 1.0039,
+      "step": 1247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.016675089330817e-06,
+      "loss": 0.9404,
+      "step": 1248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0100050654683766e-06,
+      "loss": 0.9497,
+      "step": 1249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.003335023801108e-06,
+      "loss": 0.9331,
+      "step": 1250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.996664976198894e-06,
+      "loss": 1.0239,
+      "step": 1251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.989994934531625e-06,
+      "loss": 0.9341,
+      "step": 1252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.983324910669184e-06,
+      "loss": 0.9004,
+      "step": 1253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.976654916481424e-06,
+      "loss": 1.0205,
+      "step": 1254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9699849638381396e-06,
+      "loss": 0.8838,
+      "step": 1255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.963315064609055e-06,
+      "loss": 0.8862,
+      "step": 1256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9566452306637995e-06,
+      "loss": 0.998,
+      "step": 1257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9499754738718835e-06,
+      "loss": 0.9883,
+      "step": 1258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9433058061026865e-06,
+      "loss": 0.9688,
+      "step": 1259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.93663623922542e-06,
+      "loss": 0.9507,
+      "step": 1260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929966785109125e-06,
+      "loss": 0.8428,
+      "step": 1261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.923297455622637e-06,
+      "loss": 0.9282,
+      "step": 1262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.916628262634568e-06,
+      "loss": 0.9736,
+      "step": 1263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.909959218013295e-06,
+      "loss": 0.896,
+      "step": 1264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.903290333626922e-06,
+      "loss": 0.8882,
+      "step": 1265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.896621621343272e-06,
+      "loss": 1.0562,
+      "step": 1266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.889953093029862e-06,
+      "loss": 0.9858,
+      "step": 1267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.883284760553881e-06,
+      "loss": 0.9565,
+      "step": 1268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.87661663578217e-06,
+      "loss": 0.8613,
+      "step": 1269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.869948730581198e-06,
+      "loss": 0.8711,
+      "step": 1270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.863281056817047e-06,
+      "loss": 0.9878,
+      "step": 1271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.856613626355384e-06,
+      "loss": 1.0088,
+      "step": 1272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.849946451061444e-06,
+      "loss": 0.9971,
+      "step": 1273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.84327954280001e-06,
+      "loss": 0.8833,
+      "step": 1274
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.836612913435385e-06,
+      "loss": 0.9409,
+      "step": 1275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.829946574831383e-06,
+      "loss": 0.9673,
+      "step": 1276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.823280538851291e-06,
+      "loss": 0.8535,
+      "step": 1277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.816614817357867e-06,
+      "loss": 0.9863,
+      "step": 1278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.809949422213303e-06,
+      "loss": 0.998,
+      "step": 1279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.80328436527921e-06,
+      "loss": 0.8687,
+      "step": 1280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7966196584166054e-06,
+      "loss": 0.8994,
+      "step": 1281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7899553134858715e-06,
+      "loss": 0.9214,
+      "step": 1282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7832913423467555e-06,
+      "loss": 1.0093,
+      "step": 1283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.776627756858336e-06,
+      "loss": 0.9614,
+      "step": 1284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.769964568879006e-06,
+      "loss": 0.8047,
+      "step": 1285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.763301790266452e-06,
+      "loss": 0.9941,
+      "step": 1286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.756639432877627e-06,
+      "loss": 0.9849,
+      "step": 1287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.749977508568742e-06,
+      "loss": 0.9009,
+      "step": 1288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.743316029195234e-06,
+      "loss": 0.8096,
+      "step": 1289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.736655006611744e-06,
+      "loss": 0.9146,
+      "step": 1290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.729994452672108e-06,
+      "loss": 0.9692,
+      "step": 1291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.723334379229321e-06,
+      "loss": 0.9175,
+      "step": 1292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7166747981355265e-06,
+      "loss": 1.0068,
+      "step": 1293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.710015721241993e-06,
+      "loss": 1.0068,
+      "step": 1294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7033571603990865e-06,
+      "loss": 0.8809,
+      "step": 1295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.696699127456262e-06,
+      "loss": 0.814,
+      "step": 1296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.690041634262028e-06,
+      "loss": 0.9951,
+      "step": 1297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.683384692663937e-06,
+      "loss": 0.7754,
+      "step": 1298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.676728314508558e-06,
+      "loss": 0.9995,
+      "step": 1299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.670072511641458e-06,
+      "loss": 0.9795,
+      "step": 1300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.66341729590718e-06,
+      "loss": 0.8706,
+      "step": 1301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.656762679149222e-06,
+      "loss": 0.9839,
+      "step": 1302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.650108673210014e-06,
+      "loss": 0.833,
+      "step": 1303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.643455289930907e-06,
+      "loss": 0.9902,
+      "step": 1304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.636802541152132e-06,
+      "loss": 1.0093,
+      "step": 1305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.630150438712801e-06,
+      "loss": 0.9106,
+      "step": 1306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.623498994450869e-06,
+      "loss": 0.8076,
+      "step": 1307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.616848220203124e-06,
+      "loss": 0.9487,
+      "step": 1308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.610198127805161e-06,
+      "loss": 0.9438,
+      "step": 1309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6035487290913585e-06,
+      "loss": 0.9731,
+      "step": 1310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.596900035894866e-06,
+      "loss": 0.8711,
+      "step": 1311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5902520600475694e-06,
+      "loss": 0.916,
+      "step": 1312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.583604813380087e-06,
+      "loss": 0.8765,
+      "step": 1313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.576958307721735e-06,
+      "loss": 1.0103,
+      "step": 1314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.570312554900508e-06,
+      "loss": 0.9224,
+      "step": 1315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.563667566743069e-06,
+      "loss": 0.9033,
+      "step": 1316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.557023355074711e-06,
+      "loss": 0.8789,
+      "step": 1317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.550379931719351e-06,
+      "loss": 0.9648,
+      "step": 1318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5437373084995026e-06,
+      "loss": 0.96,
+      "step": 1319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.537095497236254e-06,
+      "loss": 0.9419,
+      "step": 1320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.530454509749249e-06,
+      "loss": 0.9243,
+      "step": 1321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.523814357856664e-06,
+      "loss": 1.0322,
+      "step": 1322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.517175053375191e-06,
+      "loss": 0.9404,
+      "step": 1323
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5105366081200145e-06,
+      "loss": 0.9136,
+      "step": 1324
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.503899033904785e-06,
+      "loss": 1.1489,
+      "step": 1325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.497262342541609e-06,
+      "loss": 0.9448,
+      "step": 1326
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4906265458410155e-06,
+      "loss": 1.0024,
+      "step": 1327
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.483991655611949e-06,
+      "loss": 0.8301,
+      "step": 1328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.477357683661734e-06,
+      "loss": 0.9404,
+      "step": 1329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.470724641796064e-06,
+      "loss": 0.9619,
+      "step": 1330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.464092541818978e-06,
+      "loss": 0.9351,
+      "step": 1331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.457461395532835e-06,
+      "loss": 1.0112,
+      "step": 1332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.450831214738303e-06,
+      "loss": 0.8574,
+      "step": 1333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.444202011234328e-06,
+      "loss": 0.98,
+      "step": 1334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.437573796818115e-06,
+      "loss": 1.0547,
+      "step": 1335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.430946583285114e-06,
+      "loss": 0.9312,
+      "step": 1336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.424320382428988e-06,
+      "loss": 0.9448,
+      "step": 1337
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.417695206041603e-06,
+      "loss": 0.9683,
+      "step": 1338
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.411071065912998e-06,
+      "loss": 0.9932,
+      "step": 1339
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.40444797383137e-06,
+      "loss": 0.8794,
+      "step": 1340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.397825941583053e-06,
+      "loss": 0.9233,
+      "step": 1341
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.391204980952488e-06,
+      "loss": 0.9126,
+      "step": 1342
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.384585103722216e-06,
+      "loss": 0.9673,
+      "step": 1343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.377966321672849e-06,
+      "loss": 0.9043,
+      "step": 1344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.371348646583044e-06,
+      "loss": 0.9561,
+      "step": 1345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.364732090229497e-06,
+      "loss": 0.9219,
+      "step": 1346
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.358116664386905e-06,
+      "loss": 1.0513,
+      "step": 1347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.351502380827959e-06,
+      "loss": 0.9712,
+      "step": 1348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.344889251323316e-06,
+      "loss": 0.9688,
+      "step": 1349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.338277287641575e-06,
+      "loss": 0.9561,
+      "step": 1350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.331666501549266e-06,
+      "loss": 0.9375,
+      "step": 1351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.32505690481082e-06,
+      "loss": 1.0054,
+      "step": 1352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.318448509188552e-06,
+      "loss": 1.0366,
+      "step": 1353
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.311841326442642e-06,
+      "loss": 0.9722,
+      "step": 1354
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.305235368331106e-06,
+      "loss": 0.8335,
+      "step": 1355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2986306466097886e-06,
+      "loss": 0.9185,
+      "step": 1356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.292027173032326e-06,
+      "loss": 0.8491,
+      "step": 1357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.285424959350139e-06,
+      "loss": 0.9546,
+      "step": 1358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.278824017312404e-06,
+      "loss": 0.8252,
+      "step": 1359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.272224358666034e-06,
+      "loss": 0.9312,
+      "step": 1360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.26562599515566e-06,
+      "loss": 0.9473,
+      "step": 1361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.259028938523605e-06,
+      "loss": 0.8965,
+      "step": 1362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.252433200509869e-06,
+      "loss": 0.9307,
+      "step": 1363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.245838792852107e-06,
+      "loss": 1.0278,
+      "step": 1364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2392457272856e-06,
+      "loss": 1.0024,
+      "step": 1365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2326540155432495e-06,
+      "loss": 1.1211,
+      "step": 1366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.226063669355539e-06,
+      "loss": 0.981,
+      "step": 1367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2194747004505285e-06,
+      "loss": 0.9849,
+      "step": 1368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.212887120553824e-06,
+      "loss": 1.043,
+      "step": 1369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.20630094138856e-06,
+      "loss": 0.9678,
+      "step": 1370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1997161746753786e-06,
+      "loss": 0.9302,
+      "step": 1371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1931328321324076e-06,
+      "loss": 0.9771,
+      "step": 1372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.18655092547524e-06,
+      "loss": 0.9883,
+      "step": 1373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.179970466416918e-06,
+      "loss": 0.9458,
+      "step": 1374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.173391466667901e-06,
+      "loss": 0.9692,
+      "step": 1375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.166813937936054e-06,
+      "loss": 0.9795,
+      "step": 1376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1602378919266285e-06,
+      "loss": 0.8994,
+      "step": 1377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.15366334034223e-06,
+      "loss": 0.9741,
+      "step": 1378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.147090294882812e-06,
+      "loss": 0.9438,
+      "step": 1379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.140518767245639e-06,
+      "loss": 0.9985,
+      "step": 1380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1339487691252835e-06,
+      "loss": 0.8735,
+      "step": 1381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.127380312213591e-06,
+      "loss": 0.8979,
+      "step": 1382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1208134081996625e-06,
+      "loss": 0.9917,
+      "step": 1383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.114248068769843e-06,
+      "loss": 0.9209,
+      "step": 1384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.107684305607684e-06,
+      "loss": 0.8955,
+      "step": 1385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.101122130393938e-06,
+      "loss": 0.9692,
+      "step": 1386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.094561554806532e-06,
+      "loss": 1.0415,
+      "step": 1387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0880025905205415e-06,
+      "loss": 0.8516,
+      "step": 1388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.081445249208178e-06,
+      "loss": 0.9702,
+      "step": 1389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.074889542538765e-06,
+      "loss": 0.9341,
+      "step": 1390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.068335482178715e-06,
+      "loss": 0.8643,
+      "step": 1391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.061783079791513e-06,
+      "loss": 0.9282,
+      "step": 1392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0552323470376916e-06,
+      "loss": 0.9683,
+      "step": 1393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048683295574814e-06,
+      "loss": 0.9897,
+      "step": 1394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.04213593705745e-06,
+      "loss": 1.0381,
+      "step": 1395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.035590283137155e-06,
+      "loss": 0.874,
+      "step": 1396
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029046345462458e-06,
+      "loss": 0.9199,
+      "step": 1397
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.022504135678822e-06,
+      "loss": 0.8325,
+      "step": 1398
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.015963665428647e-06,
+      "loss": 0.8828,
+      "step": 1399
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0094249463512304e-06,
+      "loss": 0.9341,
+      "step": 1400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.002887990082754e-06,
+      "loss": 0.9512,
+      "step": 1401
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.996352808256267e-06,
+      "loss": 0.8564,
+      "step": 1402
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.989819412501654e-06,
+      "loss": 0.938,
+      "step": 1403
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.983287814445626e-06,
+      "loss": 0.9019,
+      "step": 1404
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.976758025711693e-06,
+      "loss": 0.9707,
+      "step": 1405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.970230057920145e-06,
+      "loss": 0.8662,
+      "step": 1406
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.963703922688035e-06,
+      "loss": 0.9165,
+      "step": 1407
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.957179631629148e-06,
+      "loss": 0.8955,
+      "step": 1408
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.950657196353994e-06,
+      "loss": 0.9707,
+      "step": 1409
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.944136628469775e-06,
+      "loss": 0.8926,
+      "step": 1410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.937617939580374e-06,
+      "loss": 1.1045,
+      "step": 1411
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.931101141286328e-06,
+      "loss": 0.8252,
+      "step": 1412
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9245862451848095e-06,
+      "loss": 0.9966,
+      "step": 1413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9180732628696085e-06,
+      "loss": 0.9341,
+      "step": 1414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.911562205931105e-06,
+      "loss": 0.897,
+      "step": 1415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.905053085956255e-06,
+      "loss": 0.8965,
+      "step": 1416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.898545914528569e-06,
+      "loss": 1.0142,
+      "step": 1417
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.892040703228087e-06,
+      "loss": 1.0581,
+      "step": 1418
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.885537463631362e-06,
+      "loss": 1.0103,
+      "step": 1419
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.879036207311438e-06,
+      "loss": 1.083,
+      "step": 1420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.87253694583783e-06,
+      "loss": 1.0293,
+      "step": 1421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8660396907765044e-06,
+      "loss": 0.999,
+      "step": 1422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.859544453689853e-06,
+      "loss": 0.8413,
+      "step": 1423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.853051246136681e-06,
+      "loss": 0.8799,
+      "step": 1424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.846560079672178e-06,
+      "loss": 1.0015,
+      "step": 1425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.840070965847904e-06,
+      "loss": 0.9922,
+      "step": 1426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.833583916211766e-06,
+      "loss": 0.8535,
+      "step": 1427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8270989423079955e-06,
+      "loss": 0.9741,
+      "step": 1428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.820616055677132e-06,
+      "loss": 0.9292,
+      "step": 1429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8141352678560007e-06,
+      "loss": 0.9756,
+      "step": 1430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.807656590377689e-06,
+      "loss": 0.9038,
+      "step": 1431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.801180034771534e-06,
+      "loss": 1.0635,
+      "step": 1432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7947056125630904e-06,
+      "loss": 1.0117,
+      "step": 1433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7882333352741218e-06,
+      "loss": 0.9951,
+      "step": 1434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7817632144225713e-06,
+      "loss": 1.0391,
+      "step": 1435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.775295261522546e-06,
+      "loss": 0.9355,
+      "step": 1436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7688294880842947e-06,
+      "loss": 0.8564,
+      "step": 1437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.762365905614187e-06,
+      "loss": 0.8398,
+      "step": 1438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7559045256146953e-06,
+      "loss": 0.9761,
+      "step": 1439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7494453595843692e-06,
+      "loss": 0.9785,
+      "step": 1440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7429884190178224e-06,
+      "loss": 0.9771,
+      "step": 1441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7365337154057056e-06,
+      "loss": 0.9834,
+      "step": 1442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7300812602346894e-06,
+      "loss": 0.8616,
+      "step": 1443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.723631064987443e-06,
+      "loss": 0.9146,
+      "step": 1444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7171831411426123e-06,
+      "loss": 1.0508,
+      "step": 1445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7107375001748047e-06,
+      "loss": 0.9092,
+      "step": 1446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7042941535545628e-06,
+      "loss": 0.8604,
+      "step": 1447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.697853112748345e-06,
+      "loss": 0.9077,
+      "step": 1448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6914143892185107e-06,
+      "loss": 1.0425,
+      "step": 1449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6849779944232885e-06,
+      "loss": 0.8525,
+      "step": 1450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6785439398167687e-06,
+      "loss": 0.8833,
+      "step": 1451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6721122368488772e-06,
+      "loss": 0.9204,
+      "step": 1452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.665682896965349e-06,
+      "loss": 0.9717,
+      "step": 1453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6592559316077213e-06,
+      "loss": 0.9658,
+      "step": 1454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6528313522133e-06,
+      "loss": 1.0928,
+      "step": 1455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6464091702151484e-06,
+      "loss": 1.1494,
+      "step": 1456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6399893970420614e-06,
+      "loss": 0.9771,
+      "step": 1457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6335720441185474e-06,
+      "loss": 0.8755,
+      "step": 1458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6271571228648108e-06,
+      "loss": 0.981,
+      "step": 1459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.620744644696722e-06,
+      "loss": 0.8779,
+      "step": 1460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.614334621025811e-06,
+      "loss": 0.9258,
+      "step": 1461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.607927063259237e-06,
+      "loss": 0.9678,
+      "step": 1462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.601521982799768e-06,
+      "loss": 0.9326,
+      "step": 1463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5951193910457692e-06,
+      "loss": 0.9561,
+      "step": 1464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.588719299391171e-06,
+      "loss": 0.9326,
+      "step": 1465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.582321719225459e-06,
+      "loss": 0.8623,
+      "step": 1466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.575926661933648e-06,
+      "loss": 0.9175,
+      "step": 1467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.569534138896262e-06,
+      "loss": 0.8989,
+      "step": 1468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5631441614893168e-06,
+      "loss": 0.8506,
+      "step": 1469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5567567410842972e-06,
+      "loss": 0.9204,
+      "step": 1470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5503718890481376e-06,
+      "loss": 0.8076,
+      "step": 1471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.543989616743203e-06,
+      "loss": 0.8735,
+      "step": 1472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.537609935527264e-06,
+      "loss": 1.0063,
+      "step": 1473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.531232856753486e-06,
+      "loss": 0.8765,
+      "step": 1474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5248583917703966e-06,
+      "loss": 0.9521,
+      "step": 1475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5184865519218762e-06,
+      "loss": 0.9121,
+      "step": 1476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.512117348547134e-06,
+      "loss": 0.8311,
+      "step": 1477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5057507929806843e-06,
+      "loss": 0.9263,
+      "step": 1478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4993868965523314e-06,
+      "loss": 1.0088,
+      "step": 1479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4930256705871467e-06,
+      "loss": 0.9663,
+      "step": 1480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4866671264054497e-06,
+      "loss": 0.877,
+      "step": 1481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.480311275322788e-06,
+      "loss": 0.8896,
+      "step": 1482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4739581286499147e-06,
+      "loss": 0.8906,
+      "step": 1483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.467607697692773e-06,
+      "loss": 0.9839,
+      "step": 1484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4612599937524704e-06,
+      "loss": 0.9385,
+      "step": 1485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 0.9517,
+      "step": 1486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.448572812102536e-06,
+      "loss": 0.9404,
+      "step": 1487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.442233356970775e-06,
+      "loss": 0.8484,
+      "step": 1488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4358966740115595e-06,
+      "loss": 0.853,
+      "step": 1489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4295627745015308e-06,
+      "loss": 0.8501,
+      "step": 1490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.423231669712379e-06,
+      "loss": 0.9443,
+      "step": 1491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.416903370910822e-06,
+      "loss": 0.9219,
+      "step": 1492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.410577889358582e-06,
+      "loss": 0.9312,
+      "step": 1493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.404255236312367e-06,
+      "loss": 0.9756,
+      "step": 1494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3979354230238537e-06,
+      "loss": 0.8506,
+      "step": 1495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3916184607396634e-06,
+      "loss": 0.9478,
+      "step": 1496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.385304360701347e-06,
+      "loss": 0.9463,
+      "step": 1497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3789931341453564e-06,
+      "loss": 0.8271,
+      "step": 1498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.372684792303035e-06,
+      "loss": 0.9194,
+      "step": 1499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.366379346400592e-06,
+      "loss": 0.9434,
+      "step": 1500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3600768076590772e-06,
+      "loss": 1.0405,
+      "step": 1501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.353777187294376e-06,
+      "loss": 0.9468,
+      "step": 1502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3474804965171726e-06,
+      "loss": 0.8691,
+      "step": 1503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3411867465329416e-06,
+      "loss": 0.9536,
+      "step": 1504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3348959485419242e-06,
+      "loss": 0.9619,
+      "step": 1505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.328608113739107e-06,
+      "loss": 1.0234,
+      "step": 1506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3223232533142034e-06,
+      "loss": 0.9824,
+      "step": 1507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3160413784516342e-06,
+      "loss": 0.9302,
+      "step": 1508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.309762500330507e-06,
+      "loss": 0.9194,
+      "step": 1509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3034866301245983e-06,
+      "loss": 1.0474,
+      "step": 1510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2972137790023273e-06,
+      "loss": 0.8228,
+      "step": 1511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2909439581267455e-06,
+      "loss": 0.9165,
+      "step": 1512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2846771786555075e-06,
+      "loss": 1.0532,
+      "step": 1513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2784134517408582e-06,
+      "loss": 0.9458,
+      "step": 1514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.272152788529611e-06,
+      "loss": 0.7786,
+      "step": 1515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.265895200163123e-06,
+      "loss": 0.9512,
+      "step": 1516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2596406977772855e-06,
+      "loss": 0.9448,
+      "step": 1517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.253389292502491e-06,
+      "loss": 0.9829,
+      "step": 1518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2471409954636256e-06,
+      "loss": 0.9819,
+      "step": 1519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2408958177800442e-06,
+      "loss": 0.8965,
+      "step": 1520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2346537705655456e-06,
+      "loss": 0.8853,
+      "step": 1521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.228414864928364e-06,
+      "loss": 0.9624,
+      "step": 1522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2221791119711372e-06,
+      "loss": 0.9253,
+      "step": 1523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2159465227908967e-06,
+      "loss": 0.8247,
+      "step": 1524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.209717108479042e-06,
+      "loss": 1.0083,
+      "step": 1525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2034908801213225e-06,
+      "loss": 0.9102,
+      "step": 1526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1972678487978205e-06,
+      "loss": 0.958,
+      "step": 1527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1910480255829235e-06,
+      "loss": 0.9453,
+      "step": 1528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1848314215453147e-06,
+      "loss": 0.9023,
+      "step": 1529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.178618047747948e-06,
+      "loss": 0.8701,
+      "step": 1530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.172407915248027e-06,
+      "loss": 0.8521,
+      "step": 1531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.166201035096988e-06,
+      "loss": 0.8638,
+      "step": 1532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1599974183404784e-06,
+      "loss": 0.8667,
+      "step": 1533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1537970760183406e-06,
+      "loss": 0.833,
+      "step": 1534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1476000191645894e-06,
+      "loss": 1.0366,
+      "step": 1535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.141406258807389e-06,
+      "loss": 0.9263,
+      "step": 1536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.135215805969043e-06,
+      "loss": 1.0,
+      "step": 1537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1290286716659635e-06,
+      "loss": 0.9932,
+      "step": 1538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.122844866908662e-06,
+      "loss": 0.8989,
+      "step": 1539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.116664402701721e-06,
+      "loss": 0.9341,
+      "step": 1540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1104872900437806e-06,
+      "loss": 0.9561,
+      "step": 1541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1043135399275152e-06,
+      "loss": 1.1265,
+      "step": 1542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0981431633396153e-06,
+      "loss": 1.0137,
+      "step": 1543
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.091976171260768e-06,
+      "loss": 0.9653,
+      "step": 1544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0858125746656394e-06,
+      "loss": 0.9375,
+      "step": 1545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.079652384522849e-06,
+      "loss": 0.9922,
+      "step": 1546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.073495611794959e-06,
+      "loss": 0.9438,
+      "step": 1547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.067342267438446e-06,
+      "loss": 1.02,
+      "step": 1548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.061192362403687e-06,
+      "loss": 0.9136,
+      "step": 1549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0550459076349403e-06,
+      "loss": 1.0151,
+      "step": 1550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0489029140703206e-06,
+      "loss": 0.8926,
+      "step": 1551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0427633926417876e-06,
+      "loss": 0.8784,
+      "step": 1552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0366273542751157e-06,
+      "loss": 0.8584,
+      "step": 1553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0304948098898866e-06,
+      "loss": 0.8477,
+      "step": 1554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.024365770399464e-06,
+      "loss": 0.8857,
+      "step": 1555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0182402467109696e-06,
+      "loss": 0.9561,
+      "step": 1556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.012118249725273e-06,
+      "loss": 0.9927,
+      "step": 1557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0059997903369658e-06,
+      "loss": 0.9482,
+      "step": 1558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.999884879434344e-06,
+      "loss": 0.8506,
+      "step": 1559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9937735278993913e-06,
+      "loss": 0.895,
+      "step": 1560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9876657466077523e-06,
+      "loss": 1.1304,
+      "step": 1561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.981561546428724e-06,
+      "loss": 0.9751,
+      "step": 1562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.975460938225225e-06,
+      "loss": 0.9502,
+      "step": 1563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.969363932853785e-06,
+      "loss": 0.9185,
+      "step": 1564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9632705411645224e-06,
+      "loss": 1.0317,
+      "step": 1565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9571807740011216e-06,
+      "loss": 0.9785,
+      "step": 1566
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.951094642200821e-06,
+      "loss": 0.8882,
+      "step": 1567
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.945012156594387e-06,
+      "loss": 0.8457,
+      "step": 1568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.938933328006097e-06,
+      "loss": 0.9004,
+      "step": 1569
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9328581672537227e-06,
+      "loss": 1.002,
+      "step": 1570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.926786685148506e-06,
+      "loss": 0.9004,
+      "step": 1571
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9207188924951447e-06,
+      "loss": 0.938,
+      "step": 1572
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.914654800091768e-06,
+      "loss": 0.9541,
+      "step": 1573
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.908594418729924e-06,
+      "loss": 0.9937,
+      "step": 1574
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.902537759194551e-06,
+      "loss": 0.9141,
+      "step": 1575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.896484832263973e-06,
+      "loss": 1.0005,
+      "step": 1576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.890435648709863e-06,
+      "loss": 0.9746,
+      "step": 1577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8843902192972345e-06,
+      "loss": 0.9639,
+      "step": 1578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8783485547844247e-06,
+      "loss": 0.9873,
+      "step": 1579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.872310665923065e-06,
+      "loss": 0.9712,
+      "step": 1580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.866276563458071e-06,
+      "loss": 0.9517,
+      "step": 1581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8602462581276166e-06,
+      "loss": 0.8555,
+      "step": 1582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.854219760663125e-06,
+      "loss": 0.9961,
+      "step": 1583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.848197081789237e-06,
+      "loss": 0.7983,
+      "step": 1584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8421782322237983e-06,
+      "loss": 0.9492,
+      "step": 1585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8361632226778456e-06,
+      "loss": 0.9902,
+      "step": 1586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8301520638555756e-06,
+      "loss": 0.96,
+      "step": 1587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.824144766454333e-06,
+      "loss": 0.9175,
+      "step": 1588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8181413411645964e-06,
+      "loss": 0.8877,
+      "step": 1589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.812141798669948e-06,
+      "loss": 0.8887,
+      "step": 1590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.806146149647062e-06,
+      "loss": 0.9561,
+      "step": 1591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8001544047656814e-06,
+      "loss": 0.894,
+      "step": 1592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.794166574688609e-06,
+      "loss": 0.9434,
+      "step": 1593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7881826700716724e-06,
+      "loss": 0.8623,
+      "step": 1594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7822027015637167e-06,
+      "loss": 1.0688,
+      "step": 1595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.776226679806585e-06,
+      "loss": 1.0425,
+      "step": 1596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.770254615435093e-06,
+      "loss": 0.9307,
+      "step": 1597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.764286519077014e-06,
+      "loss": 0.9258,
+      "step": 1598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7583224013530643e-06,
+      "loss": 0.9829,
+      "step": 1599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7523622728768757e-06,
+      "loss": 0.9829,
+      "step": 1600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.746406144254982e-06,
+      "loss": 0.98,
+      "step": 1601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7404540260867956e-06,
+      "loss": 0.8223,
+      "step": 1602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.734505928964601e-06,
+      "loss": 0.8965,
+      "step": 1603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.728561863473518e-06,
+      "loss": 1.0586,
+      "step": 1604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.722621840191494e-06,
+      "loss": 0.9795,
+      "step": 1605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7166858696892867e-06,
+      "loss": 0.9229,
+      "step": 1606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.710753962530438e-06,
+      "loss": 0.917,
+      "step": 1607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.704826129271257e-06,
+      "loss": 0.7964,
+      "step": 1608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6989023804608095e-06,
+      "loss": 0.8989,
+      "step": 1609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.692982726640887e-06,
+      "loss": 0.9937,
+      "step": 1610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6870671783459946e-06,
+      "loss": 0.8628,
+      "step": 1611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6811557461033313e-06,
+      "loss": 0.9189,
+      "step": 1612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.675248440432774e-06,
+      "loss": 0.9106,
+      "step": 1613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.669345271846854e-06,
+      "loss": 0.9595,
+      "step": 1614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6634462508507375e-06,
+      "loss": 0.8384,
+      "step": 1615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6575513879422167e-06,
+      "loss": 0.8555,
+      "step": 1616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.651660693611677e-06,
+      "loss": 0.9341,
+      "step": 1617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6457741783420885e-06,
+      "loss": 0.9326,
+      "step": 1618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6398918526089866e-06,
+      "loss": 0.9084,
+      "step": 1619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6340137268804473e-06,
+      "loss": 0.9634,
+      "step": 1620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6281398116170736e-06,
+      "loss": 0.9463,
+      "step": 1621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6222701172719744e-06,
+      "loss": 0.9062,
+      "step": 1622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.616404654290752e-06,
+      "loss": 0.9922,
+      "step": 1623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.610543433111473e-06,
+      "loss": 0.9131,
+      "step": 1624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6046864641646575e-06,
+      "loss": 0.8838,
+      "step": 1625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.598833757873259e-06,
+      "loss": 1.0098,
+      "step": 1626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5929853246526466e-06,
+      "loss": 0.8726,
+      "step": 1627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.587141174910579e-06,
+      "loss": 0.9092,
+      "step": 1628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5813013190472005e-06,
+      "loss": 1.0029,
+      "step": 1629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.575465767455009e-06,
+      "loss": 0.9829,
+      "step": 1630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.569634530518842e-06,
+      "loss": 0.9775,
+      "step": 1631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.56380761861586e-06,
+      "loss": 0.9868,
+      "step": 1632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5579850421155294e-06,
+      "loss": 1.0166,
+      "step": 1633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.552166811379599e-06,
+      "loss": 0.8208,
+      "step": 1634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.54635293676208e-06,
+      "loss": 0.8784,
+      "step": 1635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.540543428609241e-06,
+      "loss": 0.897,
+      "step": 1636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5347382972595724e-06,
+      "loss": 0.938,
+      "step": 1637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.528937553043775e-06,
+      "loss": 0.8921,
+      "step": 1638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5231412062847502e-06,
+      "loss": 0.9634,
+      "step": 1639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.517349267297567e-06,
+      "loss": 0.8296,
+      "step": 1640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5115617463894515e-06,
+      "loss": 0.8584,
+      "step": 1641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5057786538597674e-06,
+      "loss": 0.895,
+      "step": 1642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 0.9048,
+      "step": 1643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.494225795093737e-06,
+      "loss": 0.8467,
+      "step": 1644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.48845604941664e-06,
+      "loss": 0.8638,
+      "step": 1645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4826907732364466e-06,
+      "loss": 0.877,
+      "step": 1646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4769299768129317e-06,
+      "loss": 0.957,
+      "step": 1647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4711736703979015e-06,
+      "loss": 1.0264,
+      "step": 1648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.465421864235175e-06,
+      "loss": 0.8936,
+      "step": 1649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.459674568560557e-06,
+      "loss": 0.979,
+      "step": 1650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4539317936018287e-06,
+      "loss": 0.8848,
+      "step": 1651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.448193549578723e-06,
+      "loss": 0.9526,
+      "step": 1652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4424598467029164e-06,
+      "loss": 0.8779,
+      "step": 1653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.436730695177998e-06,
+      "loss": 0.981,
+      "step": 1654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4310061051994573e-06,
+      "loss": 1.042,
+      "step": 1655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.425286086954672e-06,
+      "loss": 0.9214,
+      "step": 1656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4195706506228785e-06,
+      "loss": 1.0835,
+      "step": 1657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.413859806375159e-06,
+      "loss": 0.8599,
+      "step": 1658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.408153564374429e-06,
+      "loss": 0.9712,
+      "step": 1659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4024519347754093e-06,
+      "loss": 0.8745,
+      "step": 1660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.396754927724614e-06,
+      "loss": 0.9297,
+      "step": 1661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.391062553360328e-06,
+      "loss": 0.9707,
+      "step": 1662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3853748218126e-06,
+      "loss": 0.9521,
+      "step": 1663
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3796917432032084e-06,
+      "loss": 0.8408,
+      "step": 1664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3740133276456528e-06,
+      "loss": 0.8267,
+      "step": 1665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3683395852451396e-06,
+      "loss": 1.0039,
+      "step": 1666
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3626705260985534e-06,
+      "loss": 0.9175,
+      "step": 1667
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3570061602944443e-06,
+      "loss": 0.9487,
+      "step": 1668
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.351346497913016e-06,
+      "loss": 0.8613,
+      "step": 1669
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3456915490260967e-06,
+      "loss": 0.9248,
+      "step": 1670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3400413236971286e-06,
+      "loss": 0.9875,
+      "step": 1671
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.334395831981145e-06,
+      "loss": 0.9126,
+      "step": 1672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3287550839247625e-06,
+      "loss": 0.9561,
+      "step": 1673
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.323119089566149e-06,
+      "loss": 1.0103,
+      "step": 1674
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3174878589350135e-06,
+      "loss": 0.9712,
+      "step": 1675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.311861402052593e-06,
+      "loss": 0.876,
+      "step": 1676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.306239728931624e-06,
+      "loss": 0.8423,
+      "step": 1677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3006228495763295e-06,
+      "loss": 0.8647,
+      "step": 1678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2950107739824067e-06,
+      "loss": 0.937,
+      "step": 1679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2894035121369996e-06,
+      "loss": 0.9077,
+      "step": 1680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.283801074018685e-06,
+      "loss": 0.9155,
+      "step": 1681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2782034695974613e-06,
+      "loss": 0.8394,
+      "step": 1682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.272610708834719e-06,
+      "loss": 0.8989,
+      "step": 1683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2670228016832325e-06,
+      "loss": 0.9165,
+      "step": 1684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.261439758087134e-06,
+      "loss": 0.9473,
+      "step": 1685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2558615879819085e-06,
+      "loss": 0.9961,
+      "step": 1686
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2502883012943614e-06,
+      "loss": 0.9282,
+      "step": 1687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.244719907942609e-06,
+      "loss": 1.022,
+      "step": 1688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2391564178360637e-06,
+      "loss": 0.9243,
+      "step": 1689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.233597840875407e-06,
+      "loss": 0.9414,
+      "step": 1690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.228044186952578e-06,
+      "loss": 0.9678,
+      "step": 1691
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.222495465950759e-06,
+      "loss": 1.0068,
+      "step": 1692
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2169516877443487e-06,
+      "loss": 0.9775,
+      "step": 1693
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2114128621989523e-06,
+      "loss": 0.9971,
+      "step": 1694
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2058789991713587e-06,
+      "loss": 0.9893,
+      "step": 1695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.20035010850953e-06,
+      "loss": 1.0122,
+      "step": 1696
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1948262000525778e-06,
+      "loss": 1.0088,
+      "step": 1697
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1893072836307433e-06,
+      "loss": 0.8945,
+      "step": 1698
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.183793369065391e-06,
+      "loss": 0.98,
+      "step": 1699
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.17828446616898e-06,
+      "loss": 0.8838,
+      "step": 1700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.172780584745049e-06,
+      "loss": 0.96,
+      "step": 1701
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.167281734588207e-06,
+      "loss": 0.9512,
+      "step": 1702
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.161787925484103e-06,
+      "loss": 0.9971,
+      "step": 1703
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1562991672094175e-06,
+      "loss": 0.981,
+      "step": 1704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1508154695318417e-06,
+      "loss": 1.0098,
+      "step": 1705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.145336842210065e-06,
+      "loss": 0.9766,
+      "step": 1706
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.139863294993749e-06,
+      "loss": 0.959,
+      "step": 1707
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1343948376235146e-06,
+      "loss": 0.9487,
+      "step": 1708
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.12893147983093e-06,
+      "loss": 0.9072,
+      "step": 1709
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.123473231338485e-06,
+      "loss": 1.0312,
+      "step": 1710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.118020101859573e-06,
+      "loss": 0.8926,
+      "step": 1711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1125721010984873e-06,
+      "loss": 0.8452,
+      "step": 1712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.107129238750386e-06,
+      "loss": 0.8335,
+      "step": 1713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.101691524501286e-06,
+      "loss": 1.021,
+      "step": 1714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0962589680280397e-06,
+      "loss": 0.9497,
+      "step": 1715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0908315789983286e-06,
+      "loss": 0.8555,
+      "step": 1716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.085409367070631e-06,
+      "loss": 0.9697,
+      "step": 1717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.079992341894212e-06,
+      "loss": 0.9868,
+      "step": 1718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.074580513109113e-06,
+      "loss": 1.0679,
+      "step": 1719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0691738903461218e-06,
+      "loss": 0.9956,
+      "step": 1720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.063772483226762e-06,
+      "loss": 0.9551,
+      "step": 1721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0583763013632814e-06,
+      "loss": 0.8813,
+      "step": 1722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.052985354358622e-06,
+      "loss": 1.002,
+      "step": 1723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0475996518064133e-06,
+      "loss": 0.9531,
+      "step": 1724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.04221920329095e-06,
+      "loss": 0.999,
+      "step": 1725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0368440183871812e-06,
+      "loss": 0.8975,
+      "step": 1726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0314741066606847e-06,
+      "loss": 0.9512,
+      "step": 1727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0261094776676537e-06,
+      "loss": 0.9473,
+      "step": 1728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0207501409548854e-06,
+      "loss": 0.8916,
+      "step": 1729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.015396106059754e-06,
+      "loss": 0.9868,
+      "step": 1730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0100473825101984e-06,
+      "loss": 0.8403,
+      "step": 1731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.004703979824712e-06,
+      "loss": 0.8525,
+      "step": 1732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9993659075123117e-06,
+      "loss": 0.9292,
+      "step": 1733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.994033175072533e-06,
+      "loss": 0.96,
+      "step": 1734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9887057919954056e-06,
+      "loss": 0.9893,
+      "step": 1735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9833837677614437e-06,
+      "loss": 0.9468,
+      "step": 1736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9780671118416217e-06,
+      "loss": 1.0229,
+      "step": 1737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9727558336973594e-06,
+      "loss": 0.9263,
+      "step": 1738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9674499427805126e-06,
+      "loss": 0.8853,
+      "step": 1739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.962149448533344e-06,
+      "loss": 0.8687,
+      "step": 1740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9568543603885136e-06,
+      "loss": 1.0591,
+      "step": 1741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.951564687769064e-06,
+      "loss": 0.9922,
+      "step": 1742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9462804400883973e-06,
+      "loss": 0.9116,
+      "step": 1743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.941001626750262e-06,
+      "loss": 1.0176,
+      "step": 1744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9357282571487346e-06,
+      "loss": 0.9033,
+      "step": 1745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.930460340668209e-06,
+      "loss": 0.9561,
+      "step": 1746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9251978866833696e-06,
+      "loss": 0.9722,
+      "step": 1747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.91994090455918e-06,
+      "loss": 1.0083,
+      "step": 1748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9146894036508714e-06,
+      "loss": 0.8745,
+      "step": 1749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.909443393303915e-06,
+      "loss": 1.0054,
+      "step": 1750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.904202882854012e-06,
+      "loss": 1.0039,
+      "step": 1751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.89896788162708e-06,
+      "loss": 0.9053,
+      "step": 1752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8937383989392294e-06,
+      "loss": 0.8975,
+      "step": 1753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.888514444096749e-06,
+      "loss": 0.9399,
+      "step": 1754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8832960263960908e-06,
+      "loss": 0.9707,
+      "step": 1755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8780831551238566e-06,
+      "loss": 0.9067,
+      "step": 1756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8728758395567743e-06,
+      "loss": 0.8462,
+      "step": 1757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8676740889616835e-06,
+      "loss": 0.9399,
+      "step": 1758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.862477912595526e-06,
+      "loss": 0.9614,
+      "step": 1759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8572873197053204e-06,
+      "loss": 0.958,
+      "step": 1760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8521023195281457e-06,
+      "loss": 0.9131,
+      "step": 1761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8469229212911361e-06,
+      "loss": 1.0596,
+      "step": 1762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8417491342114501e-06,
+      "loss": 0.9761,
+      "step": 1763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8365809674962643e-06,
+      "loss": 1.0137,
+      "step": 1764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8314184303427484e-06,
+      "loss": 0.9551,
+      "step": 1765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8262615319380628e-06,
+      "loss": 0.9521,
+      "step": 1766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8211102814593252e-06,
+      "loss": 0.8926,
+      "step": 1767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8159646880736036e-06,
+      "loss": 0.978,
+      "step": 1768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8108247609379036e-06,
+      "loss": 0.8096,
+      "step": 1769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8056905091991429e-06,
+      "loss": 0.9351,
+      "step": 1770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8005619419941372e-06,
+      "loss": 0.9141,
+      "step": 1771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7954390684495948e-06,
+      "loss": 0.9478,
+      "step": 1772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.790321897682083e-06,
+      "loss": 0.8623,
+      "step": 1773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.785210438798024e-06,
+      "loss": 0.8809,
+      "step": 1774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7801047008936733e-06,
+      "loss": 0.9224,
+      "step": 1775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7750046930551106e-06,
+      "loss": 0.9814,
+      "step": 1776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7699104243582133e-06,
+      "loss": 0.9453,
+      "step": 1777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7648219038686454e-06,
+      "loss": 0.9058,
+      "step": 1778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7597391406418468e-06,
+      "loss": 0.9351,
+      "step": 1779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7546621437230071e-06,
+      "loss": 0.9155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.749590922147053e-06,
+      "loss": 0.792,
+      "step": 1781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7445254849386405e-06,
+      "loss": 0.9561,
+      "step": 1782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.739465841112125e-06,
+      "loss": 0.9253,
+      "step": 1783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7344119996715547e-06,
+      "loss": 0.8247,
+      "step": 1784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7293639696106502e-06,
+      "loss": 0.9863,
+      "step": 1785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7243217599127953e-06,
+      "loss": 0.9175,
+      "step": 1786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7192853795510112e-06,
+      "loss": 0.9814,
+      "step": 1787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7142548374879448e-06,
+      "loss": 0.9683,
+      "step": 1788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.709230142675858e-06,
+      "loss": 0.853,
+      "step": 1789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7042113040566032e-06,
+      "loss": 0.9272,
+      "step": 1790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6991983305616106e-06,
+      "loss": 0.8862,
+      "step": 1791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.694191231111878e-06,
+      "loss": 0.9224,
+      "step": 1792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.689190014617944e-06,
+      "loss": 1.0742,
+      "step": 1793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6841946899798812e-06,
+      "loss": 1.0225,
+      "step": 1794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6792052660872749e-06,
+      "loss": 0.9355,
+      "step": 1795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6742217518192139e-06,
+      "loss": 0.9272,
+      "step": 1796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6692441560442656e-06,
+      "loss": 0.9233,
+      "step": 1797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6642724876204658e-06,
+      "loss": 0.9487,
+      "step": 1798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6593067553953069e-06,
+      "loss": 0.9492,
+      "step": 1799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6543469682057105e-06,
+      "loss": 0.9263,
+      "step": 1800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6493931348780211e-06,
+      "loss": 0.9414,
+      "step": 1801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6444452642279912e-06,
+      "loss": 0.8423,
+      "step": 1802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6395033650607583e-06,
+      "loss": 0.9658,
+      "step": 1803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6345674461708316e-06,
+      "loss": 0.8369,
+      "step": 1804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6296375163420846e-06,
+      "loss": 0.8948,
+      "step": 1805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6247135843477274e-06,
+      "loss": 1.0142,
+      "step": 1806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6197956589502967e-06,
+      "loss": 1.0083,
+      "step": 1807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6148837489016406e-06,
+      "loss": 0.9888,
+      "step": 1808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6099778629429052e-06,
+      "loss": 0.9341,
+      "step": 1809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6050780098045126e-06,
+      "loss": 1.0244,
+      "step": 1810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.600184198206149e-06,
+      "loss": 0.9395,
+      "step": 1811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.595296436856753e-06,
+      "loss": 0.8882,
+      "step": 1812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5904147344544928e-06,
+      "loss": 0.8872,
+      "step": 1813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5855390996867537e-06,
+      "loss": 0.9082,
+      "step": 1814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5806695412301271e-06,
+      "loss": 0.9712,
+      "step": 1815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5758060677503879e-06,
+      "loss": 0.9463,
+      "step": 1816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5709486879024827e-06,
+      "loss": 0.9019,
+      "step": 1817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5660974103305127e-06,
+      "loss": 0.8901,
+      "step": 1818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5612522436677246e-06,
+      "loss": 0.9126,
+      "step": 1819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5564131965364849e-06,
+      "loss": 0.9033,
+      "step": 1820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5515802775482703e-06,
+      "loss": 0.9307,
+      "step": 1821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5467534953036572e-06,
+      "loss": 0.9268,
+      "step": 1822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.541932858392296e-06,
+      "loss": 0.9443,
+      "step": 1823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5371183753929004e-06,
+      "loss": 0.9258,
+      "step": 1824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5323100548732378e-06,
+      "loss": 0.9917,
+      "step": 1825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5275079053901048e-06,
+      "loss": 1.0127,
+      "step": 1826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5227119354893172e-06,
+      "loss": 0.9419,
+      "step": 1827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.517922153705692e-06,
+      "loss": 0.8867,
+      "step": 1828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.51313856856304e-06,
+      "loss": 0.939,
+      "step": 1829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5083611885741372e-06,
+      "loss": 0.9331,
+      "step": 1830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5035900222407197e-06,
+      "loss": 0.9995,
+      "step": 1831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.498825078053469e-06,
+      "loss": 0.8853,
+      "step": 1832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.49406636449199e-06,
+      "loss": 0.917,
+      "step": 1833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4893138900247989e-06,
+      "loss": 0.8604,
+      "step": 1834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4845676631093143e-06,
+      "loss": 1.1045,
+      "step": 1835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4798276921918314e-06,
+      "loss": 0.8491,
+      "step": 1836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4750939857075147e-06,
+      "loss": 0.9385,
+      "step": 1837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4703665520803783e-06,
+      "loss": 0.939,
+      "step": 1838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4656453997232783e-06,
+      "loss": 0.9668,
+      "step": 1839
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4609305370378867e-06,
+      "loss": 0.9849,
+      "step": 1840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4562219724146887e-06,
+      "loss": 0.8481,
+      "step": 1841
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4515197142329552e-06,
+      "loss": 0.8599,
+      "step": 1842
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4468237708607397e-06,
+      "loss": 0.8535,
+      "step": 1843
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4421341506548553e-06,
+      "loss": 0.9219,
+      "step": 1844
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4374508619608608e-06,
+      "loss": 0.9053,
+      "step": 1845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.432773913113052e-06,
+      "loss": 0.8677,
+      "step": 1846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4281033124344401e-06,
+      "loss": 0.8569,
+      "step": 1847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.423439068236736e-06,
+      "loss": 0.9395,
+      "step": 1848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4187811888203468e-06,
+      "loss": 0.9092,
+      "step": 1849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.414129682474345e-06,
+      "loss": 0.9473,
+      "step": 1850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4094845574764654e-06,
+      "loss": 0.9272,
+      "step": 1851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4048458220930843e-06,
+      "loss": 0.96,
+      "step": 1852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4002134845792125e-06,
+      "loss": 0.9019,
+      "step": 1853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3955875531784695e-06,
+      "loss": 0.9795,
+      "step": 1854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.390968036123076e-06,
+      "loss": 1.0146,
+      "step": 1855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3863549416338406e-06,
+      "loss": 1.0664,
+      "step": 1856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3817482779201403e-06,
+      "loss": 0.9307,
+      "step": 1857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3771480531799054e-06,
+      "loss": 0.9351,
+      "step": 1858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3725542755996136e-06,
+      "loss": 0.9268,
+      "step": 1859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.367966953354265e-06,
+      "loss": 0.9766,
+      "step": 1860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.363386094607373e-06,
+      "loss": 0.856,
+      "step": 1861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3588117075109464e-06,
+      "loss": 0.9668,
+      "step": 1862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3542438002054832e-06,
+      "loss": 0.9292,
+      "step": 1863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3496823808199438e-06,
+      "loss": 1.0073,
+      "step": 1864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3451274574717449e-06,
+      "loss": 0.8633,
+      "step": 1865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3405790382667456e-06,
+      "loss": 0.939,
+      "step": 1866
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.336037131299227e-06,
+      "loss": 0.9604,
+      "step": 1867
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3315017446518808e-06,
+      "loss": 0.9346,
+      "step": 1868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3269728863957993e-06,
+      "loss": 1.0894,
+      "step": 1869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3224505645904534e-06,
+      "loss": 0.8911,
+      "step": 1870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3179347872836823e-06,
+      "loss": 0.8501,
+      "step": 1871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3134255625116782e-06,
+      "loss": 0.9131,
+      "step": 1872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3089228982989771e-06,
+      "loss": 0.9341,
+      "step": 1873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3044268026584338e-06,
+      "loss": 0.8647,
+      "step": 1874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2999372835912166e-06,
+      "loss": 0.9355,
+      "step": 1875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2954543490867917e-06,
+      "loss": 0.8584,
+      "step": 1876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2909780071229044e-06,
+      "loss": 0.9668,
+      "step": 1877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2865082656655698e-06,
+      "loss": 0.959,
+      "step": 1878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2820451326690576e-06,
+      "loss": 0.8901,
+      "step": 1879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.277588616075876e-06,
+      "loss": 0.9121,
+      "step": 1880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2731387238167586e-06,
+      "loss": 0.9473,
+      "step": 1881
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2686954638106497e-06,
+      "loss": 0.9058,
+      "step": 1882
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2642588439646951e-06,
+      "loss": 1.0078,
+      "step": 1883
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2598288721742191e-06,
+      "loss": 0.8472,
+      "step": 1884
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2554055563227163e-06,
+      "loss": 0.9507,
+      "step": 1885
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2509889042818396e-06,
+      "loss": 0.9756,
+      "step": 1886
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2465789239113797e-06,
+      "loss": 0.8711,
+      "step": 1887
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2421756230592535e-06,
+      "loss": 0.9268,
+      "step": 1888
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2377790095614973e-06,
+      "loss": 0.9639,
+      "step": 1889
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2333890912422409e-06,
+      "loss": 0.8809,
+      "step": 1890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2290058759137008e-06,
+      "loss": 0.9575,
+      "step": 1891
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2246293713761642e-06,
+      "loss": 0.8164,
+      "step": 1892
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2202595854179811e-06,
+      "loss": 0.8418,
+      "step": 1893
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.215896525815538e-06,
+      "loss": 0.9668,
+      "step": 1894
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2115402003332548e-06,
+      "loss": 0.8047,
+      "step": 1895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2071906167235698e-06,
+      "loss": 0.9277,
+      "step": 1896
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2028477827269186e-06,
+      "loss": 0.9927,
+      "step": 1897
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1985117060717278e-06,
+      "loss": 0.936,
+      "step": 1898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1941823944744007e-06,
+      "loss": 0.9365,
+      "step": 1899
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1898598556392987e-06,
+      "loss": 0.8784,
+      "step": 1900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1855440972587318e-06,
+      "loss": 0.9482,
+      "step": 1901
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1812351270129412e-06,
+      "loss": 0.9849,
+      "step": 1902
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1769329525700934e-06,
+      "loss": 0.9927,
+      "step": 1903
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1726375815862561e-06,
+      "loss": 0.8823,
+      "step": 1904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1683490217053916e-06,
+      "loss": 0.8384,
+      "step": 1905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1640672805593423e-06,
+      "loss": 0.9771,
+      "step": 1906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.159792365767815e-06,
+      "loss": 1.0552,
+      "step": 1907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1555242849383668e-06,
+      "loss": 0.9048,
+      "step": 1908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1512630456663976e-06,
+      "loss": 1.0137,
+      "step": 1909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1470086555351285e-06,
+      "loss": 0.8252,
+      "step": 1910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1427611221155926e-06,
+      "loss": 1.0352,
+      "step": 1911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1385204529666205e-06,
+      "loss": 0.9663,
+      "step": 1912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1342866556348303e-06,
+      "loss": 0.9707,
+      "step": 1913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.130059737654608e-06,
+      "loss": 0.9551,
+      "step": 1914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1258397065480963e-06,
+      "loss": 0.8589,
+      "step": 1915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1216265698251877e-06,
+      "loss": 0.9321,
+      "step": 1916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1174203349835e-06,
+      "loss": 0.8271,
+      "step": 1917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1132210095083696e-06,
+      "loss": 0.8994,
+      "step": 1918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1090286008728402e-06,
+      "loss": 1.0269,
+      "step": 1919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1048431165376433e-06,
+      "loss": 1.0425,
+      "step": 1920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1006645639511881e-06,
+      "loss": 0.9556,
+      "step": 1921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.096492950549549e-06,
+      "loss": 0.875,
+      "step": 1922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0923282837564537e-06,
+      "loss": 0.9404,
+      "step": 1923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.088170570983264e-06,
+      "loss": 0.9707,
+      "step": 1924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0840198196289676e-06,
+      "loss": 0.9365,
+      "step": 1925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0798760370801664e-06,
+      "loss": 0.9658,
+      "step": 1926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.075739230711058e-06,
+      "loss": 0.9111,
+      "step": 1927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0716094078834244e-06,
+      "loss": 0.936,
+      "step": 1928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0674865759466246e-06,
+      "loss": 0.9277,
+      "step": 1929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0633707422375716e-06,
+      "loss": 0.9463,
+      "step": 1930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0592619140807276e-06,
+      "loss": 0.9365,
+      "step": 1931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0551600987880839e-06,
+      "loss": 0.9746,
+      "step": 1932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0510653036591583e-06,
+      "loss": 0.9858,
+      "step": 1933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.046977535980971e-06,
+      "loss": 0.9502,
+      "step": 1934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0428968030280357e-06,
+      "loss": 0.9263,
+      "step": 1935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.038823112062351e-06,
+      "loss": 1.0488,
+      "step": 1936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0347564703333813e-06,
+      "loss": 0.9883,
+      "step": 1937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.030696885078044e-06,
+      "loss": 1.0581,
+      "step": 1938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0266443635207052e-06,
+      "loss": 0.9307,
+      "step": 1939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.022598912873155e-06,
+      "loss": 1.0381,
+      "step": 1940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.018560540334601e-06,
+      "loss": 0.8525,
+      "step": 1941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0145292530916584e-06,
+      "loss": 0.9795,
+      "step": 1942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.01050505831833e-06,
+      "loss": 0.9619,
+      "step": 1943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0064879631759976e-06,
+      "loss": 0.8896,
+      "step": 1944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0024779748134077e-06,
+      "loss": 0.8389,
+      "step": 1945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.984751003666638e-07,
+      "loss": 0.9409,
+      "step": 1946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.944793469592057e-07,
+      "loss": 0.9102,
+      "step": 1947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.904907217018e-07,
+      "loss": 0.8979,
+      "step": 1948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.865092316925328e-07,
+      "loss": 0.9277,
+      "step": 1949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.825348840167893e-07,
+      "loss": 0.8667,
+      "step": 1950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.785676857472421e-07,
+      "loss": 0.9351,
+      "step": 1951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.746076439438463e-07,
+      "loss": 0.9619,
+      "step": 1952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.70654765653818e-07,
+      "loss": 0.9258,
+      "step": 1953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.66709057911625e-07,
+      "loss": 0.9062,
+      "step": 1954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.627705277389749e-07,
+      "loss": 0.8643,
+      "step": 1955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.588391821448046e-07,
+      "loss": 0.9399,
+      "step": 1956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.549150281252633e-07,
+      "loss": 0.8408,
+      "step": 1957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.509980726637003e-07,
+      "loss": 0.8579,
+      "step": 1958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.470883227306598e-07,
+      "loss": 0.915,
+      "step": 1959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.431857852838583e-07,
+      "loss": 0.9614,
+      "step": 1960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.392904672681779e-07,
+      "loss": 0.8486,
+      "step": 1961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.354023756156571e-07,
+      "loss": 1.0107,
+      "step": 1962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.315215172454689e-07,
+      "loss": 0.9141,
+      "step": 1963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.276478990639182e-07,
+      "loss": 0.9404,
+      "step": 1964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.237815279644213e-07,
+      "loss": 0.9546,
+      "step": 1965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.19922410827504e-07,
+      "loss": 0.8579,
+      "step": 1966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.16070554520777e-07,
+      "loss": 0.9565,
+      "step": 1967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.122259658989319e-07,
+      "loss": 0.8916,
+      "step": 1968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.083886518037288e-07,
+      "loss": 0.8496,
+      "step": 1969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.045586190639799e-07,
+      "loss": 0.8389,
+      "step": 1970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.00735874495538e-07,
+      "loss": 1.0293,
+      "step": 1971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.96920424901292e-07,
+      "loss": 1.0117,
+      "step": 1972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.931122770711425e-07,
+      "loss": 0.8594,
+      "step": 1973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.893114377820001e-07,
+      "loss": 0.9004,
+      "step": 1974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.855179137977649e-07,
+      "loss": 0.9536,
+      "step": 1975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.817317118693253e-07,
+      "loss": 0.8613,
+      "step": 1976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.779528387345338e-07,
+      "loss": 0.8018,
+      "step": 1977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.741813011182015e-07,
+      "loss": 0.9224,
+      "step": 1978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.704171057320893e-07,
+      "loss": 0.9546,
+      "step": 1979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.666602592748885e-07,
+      "loss": 0.9307,
+      "step": 1980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.629107684322113e-07,
+      "loss": 0.937,
+      "step": 1981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.591686398765847e-07,
+      "loss": 0.939,
+      "step": 1982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.5543388026743e-07,
+      "loss": 0.9771,
+      "step": 1983
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.517064962510552e-07,
+      "loss": 0.9844,
+      "step": 1984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.479864944606431e-07,
+      "loss": 0.9541,
+      "step": 1985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.442738815162421e-07,
+      "loss": 0.9326,
+      "step": 1986
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.405686640247473e-07,
+      "loss": 0.9287,
+      "step": 1987
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.368708485798938e-07,
+      "loss": 0.8892,
+      "step": 1988
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.331804417622486e-07,
+      "loss": 0.8442,
+      "step": 1989
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.294974501391884e-07,
+      "loss": 0.9595,
+      "step": 1990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.25821880264896e-07,
+      "loss": 0.9629,
+      "step": 1991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.221537386803496e-07,
+      "loss": 0.769,
+      "step": 1992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.18493031913305e-07,
+      "loss": 0.897,
+      "step": 1993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.14839766478287e-07,
+      "loss": 0.978,
+      "step": 1994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.11193948876578e-07,
+      "loss": 0.9448,
+      "step": 1995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.075555855962097e-07,
+      "loss": 0.9233,
+      "step": 1996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.039246831119446e-07,
+      "loss": 0.9087,
+      "step": 1997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.003012478852679e-07,
+      "loss": 0.9766,
+      "step": 1998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.966852863643798e-07,
+      "loss": 0.9712,
+      "step": 1999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.930768049841769e-07,
+      "loss": 0.8066,
+      "step": 2000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.894758101662448e-07,
+      "loss": 0.9141,
+      "step": 2001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.858823083188494e-07,
+      "loss": 0.9697,
+      "step": 2002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.822963058369176e-07,
+      "loss": 0.9038,
+      "step": 2003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.787178091020337e-07,
+      "loss": 0.8667,
+      "step": 2004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.751468244824217e-07,
+      "loss": 1.0122,
+      "step": 2005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.71583358332943e-07,
+      "loss": 0.9556,
+      "step": 2006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.680274169950741e-07,
+      "loss": 1.0796,
+      "step": 2007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.644790067969005e-07,
+      "loss": 0.9014,
+      "step": 2008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.609381340531096e-07,
+      "loss": 0.979,
+      "step": 2009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.574048050649712e-07,
+      "loss": 0.9746,
+      "step": 2010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.53879026120331e-07,
+      "loss": 1.0249,
+      "step": 2011
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.503608034936022e-07,
+      "loss": 0.9268,
+      "step": 2012
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.46850143445747e-07,
+      "loss": 0.8506,
+      "step": 2013
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.433470522242702e-07,
+      "loss": 0.8916,
+      "step": 2014
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.398515360632075e-07,
+      "loss": 0.8433,
+      "step": 2015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.36363601183116e-07,
+      "loss": 1.0073,
+      "step": 2016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.328832537910585e-07,
+      "loss": 0.938,
+      "step": 2017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.294105000805962e-07,
+      "loss": 0.999,
+      "step": 2018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.259453462317789e-07,
+      "loss": 0.916,
+      "step": 2019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.224877984111289e-07,
+      "loss": 0.9761,
+      "step": 2020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.190378627716321e-07,
+      "loss": 0.917,
+      "step": 2021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.15595545452733e-07,
+      "loss": 0.9341,
+      "step": 2022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.121608525803142e-07,
+      "loss": 0.9268,
+      "step": 2023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.087337902666907e-07,
+      "loss": 0.9194,
+      "step": 2024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.053143646105987e-07,
+      "loss": 0.8872,
+      "step": 2025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.019025816971852e-07,
+      "loss": 0.9487,
+      "step": 2026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.984984475979961e-07,
+      "loss": 0.9761,
+      "step": 2027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.951019683709631e-07,
+      "loss": 0.9829,
+      "step": 2028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.917131500603996e-07,
+      "loss": 0.9043,
+      "step": 2029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.883319986969834e-07,
+      "loss": 0.9048,
+      "step": 2030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.849585202977466e-07,
+      "loss": 0.9971,
+      "step": 2031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.815927208660711e-07,
+      "loss": 0.8374,
+      "step": 2032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7823460639167e-07,
+      "loss": 0.8423,
+      "step": 2033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.748841828505815e-07,
+      "loss": 1.0513,
+      "step": 2034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.715414562051553e-07,
+      "loss": 0.9419,
+      "step": 2035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.682064324040482e-07,
+      "loss": 0.9072,
+      "step": 2036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.648791173822055e-07,
+      "loss": 0.8896,
+      "step": 2037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.615595170608541e-07,
+      "loss": 0.9282,
+      "step": 2038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.58247637347495e-07,
+      "loss": 0.8364,
+      "step": 2039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.54943484135886e-07,
+      "loss": 0.894,
+      "step": 2040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.516470633060368e-07,
+      "loss": 0.9692,
+      "step": 2041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.483583807241984e-07,
+      "loss": 0.8071,
+      "step": 2042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.450774422428474e-07,
+      "loss": 0.9175,
+      "step": 2043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.418042537006813e-07,
+      "loss": 0.9829,
+      "step": 2044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.385388209226046e-07,
+      "loss": 0.8667,
+      "step": 2045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.352811497197231e-07,
+      "loss": 0.9077,
+      "step": 2046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.320312458893262e-07,
+      "loss": 0.8452,
+      "step": 2047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.287891152148823e-07,
+      "loss": 0.9746,
+      "step": 2048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.255547634660275e-07,
+      "loss": 0.9858,
+      "step": 2049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.223281963985539e-07,
+      "loss": 1.0029,
+      "step": 2050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.191094197543995e-07,
+      "loss": 0.915,
+      "step": 2051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.158984392616407e-07,
+      "loss": 1.0815,
+      "step": 2052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.126952606344777e-07,
+      "loss": 1.0,
+      "step": 2053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.094998895732279e-07,
+      "loss": 0.8174,
+      "step": 2054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.063123317643127e-07,
+      "loss": 0.9243,
+      "step": 2055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.031325928802534e-07,
+      "loss": 0.9795,
+      "step": 2056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.99960678579653e-07,
+      "loss": 0.9678,
+      "step": 2057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.967965945071896e-07,
+      "loss": 0.9136,
+      "step": 2058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.936403462936113e-07,
+      "loss": 1.0659,
+      "step": 2059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.904919395557173e-07,
+      "loss": 0.9502,
+      "step": 2060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.873513798963521e-07,
+      "loss": 1.0381,
+      "step": 2061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.842186729044003e-07,
+      "loss": 1.02,
+      "step": 2062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.81093824154767e-07,
+      "loss": 0.9292,
+      "step": 2063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.779768392083745e-07,
+      "loss": 0.9531,
+      "step": 2064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.74867723612153e-07,
+      "loss": 0.9302,
+      "step": 2065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.717664828990266e-07,
+      "loss": 0.9463,
+      "step": 2066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.686731225879038e-07,
+      "loss": 0.9854,
+      "step": 2067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.655876481836719e-07,
+      "loss": 0.9033,
+      "step": 2068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.625100651771848e-07,
+      "loss": 0.8862,
+      "step": 2069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.594403790452513e-07,
+      "loss": 0.8838,
+      "step": 2070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.563785952506267e-07,
+      "loss": 0.9292,
+      "step": 2071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.533247192420066e-07,
+      "loss": 1.0205,
+      "step": 2072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.502787564540102e-07,
+      "loss": 0.9453,
+      "step": 2073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.47240712307176e-07,
+      "loss": 1.0479,
+      "step": 2074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.442105922079533e-07,
+      "loss": 0.9346,
+      "step": 2075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.411884015486851e-07,
+      "loss": 0.9336,
+      "step": 2076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.381741457076068e-07,
+      "loss": 0.8599,
+      "step": 2077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.351678300488306e-07,
+      "loss": 0.9673,
+      "step": 2078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.32169459922342e-07,
+      "loss": 0.8862,
+      "step": 2079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.291790406639836e-07,
+      "loss": 0.9541,
+      "step": 2080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.261965775954486e-07,
+      "loss": 0.9268,
+      "step": 2081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.232220760242757e-07,
+      "loss": 0.8115,
+      "step": 2082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.202555412438309e-07,
+      "loss": 0.856,
+      "step": 2083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.172969785333038e-07,
+      "loss": 0.9688,
+      "step": 2084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.143463931576992e-07,
+      "loss": 0.9355,
+      "step": 2085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.114037903678227e-07,
+      "loss": 1.0078,
+      "step": 2086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.084691754002763e-07,
+      "loss": 0.9727,
+      "step": 2087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.055425534774444e-07,
+      "loss": 0.9653,
+      "step": 2088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.026239298074909e-07,
+      "loss": 0.8926,
+      "step": 2089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.997133095843443e-07,
+      "loss": 0.9761,
+      "step": 2090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.96810697987688e-07,
+      "loss": 0.9653,
+      "step": 2091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.939161001829579e-07,
+      "loss": 0.8062,
+      "step": 2092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.910295213213262e-07,
+      "loss": 0.8564,
+      "step": 2093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.881509665396927e-07,
+      "loss": 0.8369,
+      "step": 2094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.852804409606832e-07,
+      "loss": 0.897,
+      "step": 2095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.824179496926296e-07,
+      "loss": 0.915,
+      "step": 2096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.795634978295688e-07,
+      "loss": 0.9727,
+      "step": 2097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7671709045122914e-07,
+      "loss": 0.8496,
+      "step": 2098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.73878732623026e-07,
+      "loss": 0.9595,
+      "step": 2099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7104842939604713e-07,
+      "loss": 0.9575,
+      "step": 2100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6822618580704694e-07,
+      "loss": 0.9126,
+      "step": 2101
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6541200687843903e-07,
+      "loss": 0.8848,
+      "step": 2102
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6260589761828347e-07,
+      "loss": 0.9302,
+      "step": 2103
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5980786302027846e-07,
+      "loss": 0.9854,
+      "step": 2104
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.570179080637571e-07,
+      "loss": 0.8774,
+      "step": 2105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.542360377136701e-07,
+      "loss": 0.8638,
+      "step": 2106
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5146225692058174e-07,
+      "loss": 0.8589,
+      "step": 2107
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.486965706206597e-07,
+      "loss": 0.8926,
+      "step": 2108
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4593898373567093e-07,
+      "loss": 1.0474,
+      "step": 2109
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.431895011729637e-07,
+      "loss": 0.916,
+      "step": 2110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.404481278254652e-07,
+      "loss": 0.9023,
+      "step": 2111
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3771486857167476e-07,
+      "loss": 1.0747,
+      "step": 2112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.349897282756488e-07,
+      "loss": 0.916,
+      "step": 2113
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.322727117869951e-07,
+      "loss": 0.9756,
+      "step": 2114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2956382394086783e-07,
+      "loss": 0.9004,
+      "step": 2115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2686306955795173e-07,
+      "loss": 0.8735,
+      "step": 2116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.241704534444596e-07,
+      "loss": 0.9028,
+      "step": 2117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2148598039212064e-07,
+      "loss": 0.9331,
+      "step": 2118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1880965517817396e-07,
+      "loss": 1.125,
+      "step": 2119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.161414825653576e-07,
+      "loss": 0.8633,
+      "step": 2120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1348146730190165e-07,
+      "loss": 0.9492,
+      "step": 2121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1082961412152065e-07,
+      "loss": 0.9688,
+      "step": 2122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.081859277434025e-07,
+      "loss": 0.8325,
+      "step": 2123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.055504128722021e-07,
+      "loss": 0.9243,
+      "step": 2124
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0292307419803333e-07,
+      "loss": 0.8984,
+      "step": 2125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.003039163964595e-07,
+      "loss": 0.9004,
+      "step": 2126
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9769294412848436e-07,
+      "loss": 0.9194,
+      "step": 2127
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9509016204054506e-07,
+      "loss": 0.9976,
+      "step": 2128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9249557476450593e-07,
+      "loss": 0.8608,
+      "step": 2129
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8990918691764525e-07,
+      "loss": 0.8813,
+      "step": 2130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8733100310265e-07,
+      "loss": 0.8916,
+      "step": 2131
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.847610279076097e-07,
+      "loss": 0.9067,
+      "step": 2132
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8219926590600365e-07,
+      "loss": 0.8857,
+      "step": 2133
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7964572165669456e-07,
+      "loss": 0.9985,
+      "step": 2134
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7710039970392464e-07,
+      "loss": 1.0391,
+      "step": 2135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.745633045773006e-07,
+      "loss": 1.0117,
+      "step": 2136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7203444079178977e-07,
+      "loss": 1.04,
+      "step": 2137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.695138128477094e-07,
+      "loss": 0.9458,
+      "step": 2138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.670014252307247e-07,
+      "loss": 1.0249,
+      "step": 2139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6449728241183256e-07,
+      "loss": 0.9243,
+      "step": 2140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6200138884735894e-07,
+      "loss": 0.9585,
+      "step": 2141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.595137489789502e-07,
+      "loss": 0.8748,
+      "step": 2142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.570343672335641e-07,
+      "loss": 0.9385,
+      "step": 2143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5456324802346064e-07,
+      "loss": 0.9907,
+      "step": 2144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.521003957461999e-07,
+      "loss": 0.9897,
+      "step": 2145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4964581478462654e-07,
+      "loss": 0.8389,
+      "step": 2146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4719950950686807e-07,
+      "loss": 0.9419,
+      "step": 2147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4476148426632215e-07,
+      "loss": 0.9785,
+      "step": 2148
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4233174340165486e-07,
+      "loss": 0.9507,
+      "step": 2149
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.39910291236788e-07,
+      "loss": 0.9453,
+      "step": 2150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.374971320808906e-07,
+      "loss": 0.939,
+      "step": 2151
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3509227022837876e-07,
+      "loss": 0.9272,
+      "step": 2152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.326957099588984e-07,
+      "loss": 0.8945,
+      "step": 2153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.303074555373226e-07,
+      "loss": 0.9478,
+      "step": 2154
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.27927511213747e-07,
+      "loss": 0.8594,
+      "step": 2155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2555588122347434e-07,
+      "loss": 0.9722,
+      "step": 2156
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2319256978701365e-07,
+      "loss": 0.9155,
+      "step": 2157
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2083758111006946e-07,
+      "loss": 0.8486,
+      "step": 2158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1849091938353725e-07,
+      "loss": 0.9873,
+      "step": 2159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.161525887834921e-07,
+      "loss": 0.9561,
+      "step": 2160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1382259347118195e-07,
+      "loss": 0.9585,
+      "step": 2161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1150093759302615e-07,
+      "loss": 0.8862,
+      "step": 2162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.091876252805981e-07,
+      "loss": 0.9038,
+      "step": 2163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0688266065062535e-07,
+      "loss": 0.939,
+      "step": 2164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.045860478049811e-07,
+      "loss": 0.9126,
+      "step": 2165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0229779083067455e-07,
+      "loss": 0.9253,
+      "step": 2166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0001789379984547e-07,
+      "loss": 0.9277,
+      "step": 2167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9774636076975507e-07,
+      "loss": 0.9648,
+      "step": 2168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9548319578278306e-07,
+      "loss": 0.9165,
+      "step": 2169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.932284028664145e-07,
+      "loss": 0.9585,
+      "step": 2170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9098198603323634e-07,
+      "loss": 0.8335,
+      "step": 2171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.887439492809319e-07,
+      "loss": 0.8774,
+      "step": 2172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8651429659226906e-07,
+      "loss": 0.9199,
+      "step": 2173
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.842930319350956e-07,
+      "loss": 0.8662,
+      "step": 2174
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8208015926233345e-07,
+      "loss": 0.8706,
+      "step": 2175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7987568251197027e-07,
+      "loss": 0.9521,
+      "step": 2176
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7767960560705076e-07,
+      "loss": 1.0283,
+      "step": 2177
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.754919324556721e-07,
+      "loss": 1.0107,
+      "step": 2178
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.733126669509789e-07,
+      "loss": 0.9043,
+      "step": 2179
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.711418129711496e-07,
+      "loss": 0.9668,
+      "step": 2180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.689793743793956e-07,
+      "loss": 0.9058,
+      "step": 2181
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6682535502395356e-07,
+      "loss": 0.9346,
+      "step": 2182
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6467975873807617e-07,
+      "loss": 0.9214,
+      "step": 2183
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6254258934002517e-07,
+      "loss": 1.0537,
+      "step": 2184
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.604138506330689e-07,
+      "loss": 0.8384,
+      "step": 2185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.582935464054709e-07,
+      "loss": 0.8354,
+      "step": 2186
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.561816804304845e-07,
+      "loss": 1.0078,
+      "step": 2187
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.54078256466348e-07,
+      "loss": 0.9634,
+      "step": 2188
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.519832782562753e-07,
+      "loss": 0.9399,
+      "step": 2189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4989674952845046e-07,
+      "loss": 0.9341,
+      "step": 2190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4781867399602033e-07,
+      "loss": 1.0225,
+      "step": 2191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4574905535709183e-07,
+      "loss": 0.814,
+      "step": 2192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.436878972947176e-07,
+      "loss": 0.8926,
+      "step": 2193
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.416352034768965e-07,
+      "loss": 0.9453,
+      "step": 2194
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.395909775565652e-07,
+      "loss": 0.8318,
+      "step": 2195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3755522317158998e-07,
+      "loss": 0.9351,
+      "step": 2196
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3552794394476053e-07,
+      "loss": 0.9224,
+      "step": 2197
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3350914348378606e-07,
+      "loss": 0.8247,
+      "step": 2198
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3149882538128664e-07,
+      "loss": 0.8857,
+      "step": 2199
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2949699321478612e-07,
+      "loss": 0.9692,
+      "step": 2200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2750365054670765e-07,
+      "loss": 0.9854,
+      "step": 2201
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2551880092436807e-07,
+      "loss": 1.1465,
+      "step": 2202
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2354244787996748e-07,
+      "loss": 0.835,
+      "step": 2203
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2157459493058754e-07,
+      "loss": 0.9014,
+      "step": 2204
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1961524557818304e-07,
+      "loss": 0.875,
+      "step": 2205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.176644033095754e-07,
+      "loss": 0.8711,
+      "step": 2206
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1572207159644643e-07,
+      "loss": 0.9131,
+      "step": 2207
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1378825389533508e-07,
+      "loss": 0.8721,
+      "step": 2208
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1186295364762687e-07,
+      "loss": 0.9277,
+      "step": 2209
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0994617427954945e-07,
+      "loss": 0.958,
+      "step": 2210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.080379192021681e-07,
+      "loss": 0.959,
+      "step": 2211
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0613819181137918e-07,
+      "loss": 1.0,
+      "step": 2212
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.042469954879006e-07,
+      "loss": 0.9351,
+      "step": 2213
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0236433359727082e-07,
+      "loss": 0.8931,
+      "step": 2214
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0049020948984033e-07,
+      "loss": 0.9316,
+      "step": 2215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.986246265007652e-07,
+      "loss": 0.8613,
+      "step": 2216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9676758795000194e-07,
+      "loss": 0.9653,
+      "step": 2217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9491909714230207e-07,
+      "loss": 0.9854,
+      "step": 2218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9307915736720585e-07,
+      "loss": 0.9858,
+      "step": 2219
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9124777189903575e-07,
+      "loss": 0.8774,
+      "step": 2220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8942494399688983e-07,
+      "loss": 0.8555,
+      "step": 2221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8761067690464097e-07,
+      "loss": 0.9585,
+      "step": 2222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8580497385092376e-07,
+      "loss": 0.897,
+      "step": 2223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8400783804913335e-07,
+      "loss": 1.0347,
+      "step": 2224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.822192726974209e-07,
+      "loss": 0.916,
+      "step": 2225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8043928097868312e-07,
+      "loss": 0.9731,
+      "step": 2226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7866786606055953e-07,
+      "loss": 0.813,
+      "step": 2227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7690503109542844e-07,
+      "loss": 0.877,
+      "step": 2228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7515077922039715e-07,
+      "loss": 0.8555,
+      "step": 2229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7340511355730006e-07,
+      "loss": 0.8086,
+      "step": 2230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7166803721269109e-07,
+      "loss": 0.9878,
+      "step": 2231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6993955327783919e-07,
+      "loss": 0.9312,
+      "step": 2232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6821966482872264e-07,
+      "loss": 0.8096,
+      "step": 2233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6650837492602155e-07,
+      "loss": 1.0288,
+      "step": 2234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6480568661511765e-07,
+      "loss": 0.9067,
+      "step": 2235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6311160292608208e-07,
+      "loss": 0.8721,
+      "step": 2236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6142612687367387e-07,
+      "loss": 1.0273,
+      "step": 2237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5974926145733705e-07,
+      "loss": 0.959,
+      "step": 2238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5808100966118844e-07,
+      "loss": 0.9976,
+      "step": 2239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5642137445401885e-07,
+      "loss": 0.8203,
+      "step": 2240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5477035878928293e-07,
+      "loss": 0.8779,
+      "step": 2241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5312796560509935e-07,
+      "loss": 0.9624,
+      "step": 2242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5149419782423902e-07,
+      "loss": 0.9253,
+      "step": 2243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4986905835412402e-07,
+      "loss": 0.9175,
+      "step": 2244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4825255008682483e-07,
+      "loss": 1.0967,
+      "step": 2245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.466446758990475e-07,
+      "loss": 0.9683,
+      "step": 2246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.450454386521355e-07,
+      "loss": 0.8955,
+      "step": 2247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4345484119206222e-07,
+      "loss": 0.9487,
+      "step": 2248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4187288634942454e-07,
+      "loss": 1.0107,
+      "step": 2249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.402995769394405e-07,
+      "loss": 0.8823,
+      "step": 2250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3873491576194165e-07,
+      "loss": 0.9604,
+      "step": 2251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.371789056013706e-07,
+      "loss": 0.8652,
+      "step": 2252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3563154922677347e-07,
+      "loss": 1.0527,
+      "step": 2253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.34092849391797e-07,
+      "loss": 0.9062,
+      "step": 2254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3256280883468309e-07,
+      "loss": 0.8872,
+      "step": 2255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3104143027826365e-07,
+      "loss": 1.002,
+      "step": 2256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.295287164299547e-07,
+      "loss": 0.8774,
+      "step": 2257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.28024669981755e-07,
+      "loss": 1.0347,
+      "step": 2258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2652929361023635e-07,
+      "loss": 0.8184,
+      "step": 2259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2504258997654395e-07,
+      "loss": 0.9453,
+      "step": 2260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2356456172638642e-07,
+      "loss": 0.8989,
+      "step": 2261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.22095211490037e-07,
+      "loss": 0.9155,
+      "step": 2262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.206345418823235e-07,
+      "loss": 0.9854,
+      "step": 2263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1918255550262548e-07,
+      "loss": 0.8564,
+      "step": 2264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1773925493487159e-07,
+      "loss": 0.8672,
+      "step": 2265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1630464274753284e-07,
+      "loss": 0.9844,
+      "step": 2266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1487872149361811e-07,
+      "loss": 1.063,
+      "step": 2267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1346149371067094e-07,
+      "loss": 0.9785,
+      "step": 2268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1205296192076275e-07,
+      "loss": 0.9707,
+      "step": 2269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1065312863049071e-07,
+      "loss": 0.9463,
+      "step": 2270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0926199633097156e-07,
+      "loss": 0.8843,
+      "step": 2271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0787956749784002e-07,
+      "loss": 0.8896,
+      "step": 2272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.065058445912398e-07,
+      "loss": 0.8091,
+      "step": 2273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.051408300558221e-07,
+      "loss": 0.8647,
+      "step": 2274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.037845263207421e-07,
+      "loss": 0.9878,
+      "step": 2275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0243693579965298e-07,
+      "loss": 0.8896,
+      "step": 2276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0109806089070085e-07,
+      "loss": 1.0127,
+      "step": 2277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.976790397652314e-08,
+      "loss": 0.9209,
+      "step": 2278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.844646742424247e-08,
+      "loss": 0.8784,
+      "step": 2279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.713375358546273e-08,
+      "loss": 0.9194,
+      "step": 2280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.582976479626471e-08,
+      "loss": 0.9473,
+      "step": 2281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.453450337720271e-08,
+      "loss": 1.0029,
+      "step": 2282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.324797163330012e-08,
+      "loss": 0.96,
+      "step": 2283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.197017185404444e-08,
+      "loss": 1.0117,
+      "step": 2284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.070110631338502e-08,
+      "loss": 0.8887,
+      "step": 2285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.944077726972644e-08,
+      "loss": 0.8545,
+      "step": 2286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.818918696592737e-08,
+      "loss": 0.9385,
+      "step": 2287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.694633762929505e-08,
+      "loss": 0.8242,
+      "step": 2288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.571223147158081e-08,
+      "loss": 1.0195,
+      "step": 2289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.448687068897676e-08,
+      "loss": 0.9321,
+      "step": 2290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.327025746211137e-08,
+      "loss": 0.9355,
+      "step": 2291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.206239395604832e-08,
+      "loss": 0.9639,
+      "step": 2292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.086328232027874e-08,
+      "loss": 0.9731,
+      "step": 2293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.967292468871846e-08,
+      "loss": 0.9272,
+      "step": 2294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.849132317970687e-08,
+      "loss": 0.9487,
+      "step": 2295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.731847989599916e-08,
+      "loss": 0.9712,
+      "step": 2296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.615439692476578e-08,
+      "loss": 0.9316,
+      "step": 2297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.499907633758797e-08,
+      "loss": 1.0112,
+      "step": 2298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.385252019045164e-08,
+      "loss": 0.9561,
+      "step": 2299
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.271473052374744e-08,
+      "loss": 0.9521,
+      "step": 2300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.15857093622635e-08,
+      "loss": 0.9438,
+      "step": 2301
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.046545871518651e-08,
+      "loss": 0.8311,
+      "step": 2302
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.935398057609344e-08,
+      "loss": 0.9429,
+      "step": 2303
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.825127692294875e-08,
+      "loss": 0.9023,
+      "step": 2304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.715734971810439e-08,
+      "loss": 0.8701,
+      "step": 2305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.607220090829314e-08,
+      "loss": 0.8828,
+      "step": 2306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.49958324246247e-08,
+      "loss": 0.9702,
+      "step": 2307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.39282461825852e-08,
+      "loss": 0.8779,
+      "step": 2308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.2869444082031e-08,
+      "loss": 0.8955,
+      "step": 2309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.181942800718654e-08,
+      "loss": 0.9985,
+      "step": 2310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.077819982664101e-08,
+      "loss": 0.9653,
+      "step": 2311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.974576139334442e-08,
+      "loss": 0.8682,
+      "step": 2312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8722114544605966e-08,
+      "loss": 0.9941,
+      "step": 2313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7707261102086795e-08,
+      "loss": 0.894,
+      "step": 2314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.670120287180281e-08,
+      "loss": 0.7964,
+      "step": 2315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5703941644116297e-08,
+      "loss": 0.9009,
+      "step": 2316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.471547919373377e-08,
+      "loss": 0.8643,
+      "step": 2317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.373581727970589e-08,
+      "loss": 0.8735,
+      "step": 2318
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2764957645419775e-08,
+      "loss": 0.8604,
+      "step": 2319
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.18029020185995e-08,
+      "loss": 0.9399,
+      "step": 2320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.084965211130166e-08,
+      "loss": 1.0005,
+      "step": 2321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9905209619912074e-08,
+      "loss": 0.9551,
+      "step": 2322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.896957622514298e-08,
+      "loss": 0.8008,
+      "step": 2323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8042753592029145e-08,
+      "loss": 0.8892,
+      "step": 2324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.712474336992845e-08,
+      "loss": 0.9902,
+      "step": 2325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6215547192514085e-08,
+      "loss": 0.9922,
+      "step": 2326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5315166677774025e-08,
+      "loss": 0.9043,
+      "step": 2327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.442360342800989e-08,
+      "loss": 1.0396,
+      "step": 2328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.354085902983085e-08,
+      "loss": 0.9761,
+      "step": 2329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.266693505415143e-08,
+      "loss": 1.0005,
+      "step": 2330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1801833056191453e-08,
+      "loss": 1.0723,
+      "step": 2331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.094555457547e-08,
+      "loss": 1.0039,
+      "step": 2332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.009810113580426e-08,
+      "loss": 0.9463,
+      "step": 2333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.925947424530618e-08,
+      "loss": 0.8721,
+      "step": 2334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8429675396381985e-08,
+      "loss": 0.9326,
+      "step": 2335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.760870606572487e-08,
+      "loss": 0.9761,
+      "step": 2336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.679656771431728e-08,
+      "loss": 0.959,
+      "step": 2337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.599326178742535e-08,
+      "loss": 0.9326,
+      "step": 2338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.519878971459723e-08,
+      "loss": 0.8296,
+      "step": 2339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.441315290966085e-08,
+      "loss": 0.8506,
+      "step": 2340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3636352770720636e-08,
+      "loss": 1.1372,
+      "step": 2341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2868390680156345e-08,
+      "loss": 1.0005,
+      "step": 2342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.210926800461811e-08,
+      "loss": 0.8926,
+      "step": 2343
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1358986095026964e-08,
+      "loss": 0.8838,
+      "step": 2344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0617546286570434e-08,
+      "loss": 0.9468,
+      "step": 2345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.988494989870139e-08,
+      "loss": 0.9648,
+      "step": 2346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.916119823513419e-08,
+      "loss": 0.9062,
+      "step": 2347
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8446292583844126e-08,
+      "loss": 0.9243,
+      "step": 2348
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.774023421706462e-08,
+      "loss": 0.9805,
+      "step": 2349
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7043024391282814e-08,
+      "loss": 1.0591,
+      "step": 2350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.635466434724121e-08,
+      "loss": 1.0107,
+      "step": 2351
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.567515530993214e-08,
+      "loss": 0.9355,
+      "step": 2352
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.500449848859776e-08,
+      "loss": 0.9004,
+      "step": 2353
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4342695076725598e-08,
+      "loss": 0.9619,
+      "step": 2354
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3689746252049117e-08,
+      "loss": 0.9819,
+      "step": 2355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3045653176544392e-08,
+      "loss": 0.9399,
+      "step": 2356
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2410416996426766e-08,
+      "loss": 0.8779,
+      "step": 2357
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.178403884215141e-08,
+      "loss": 0.8872,
+      "step": 2358
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1166519828408892e-08,
+      "loss": 0.8813,
+      "step": 2359
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0557861054124052e-08,
+      "loss": 0.8804,
+      "step": 2360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9958063602455445e-08,
+      "loss": 1.0181,
+      "step": 2361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.936712854079148e-08,
+      "loss": 0.8882,
+      "step": 2362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8785056920748723e-08,
+      "loss": 0.9106,
+      "step": 2363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.821184977817081e-08,
+      "loss": 0.8979,
+      "step": 2364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7647508133127324e-08,
+      "loss": 0.936,
+      "step": 2365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.709203298990991e-08,
+      "loss": 0.8862,
+      "step": 2366
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.654542533703174e-08,
+      "loss": 0.7925,
+      "step": 2367
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6007686147225254e-08,
+      "loss": 0.9219,
+      "step": 2368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5478816377441088e-08,
+      "loss": 1.0044,
+      "step": 2369
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4958816968846935e-08,
+      "loss": 0.9053,
+      "step": 2370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4447688846823127e-08,
+      "loss": 0.9951,
+      "step": 2371
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3945432920964286e-08,
+      "loss": 0.9009,
+      "step": 2372
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3452050085075441e-08,
+      "loss": 0.9189,
+      "step": 2373
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2967541217172031e-08,
+      "loss": 0.9141,
+      "step": 2374
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2491907179476015e-08,
+      "loss": 1.0103,
+      "step": 2375
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2025148818418097e-08,
+      "loss": 0.9077,
+      "step": 2376
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1567266964631618e-08,
+      "loss": 0.9634,
+      "step": 2377
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1118262432955329e-08,
+      "loss": 0.9785,
+      "step": 2378
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0678136022429509e-08,
+      "loss": 0.998,
+      "step": 2379
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.024688851629374e-08,
+      "loss": 0.9028,
+      "step": 2380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.82452068198969e-09,
+      "loss": 0.9526,
+      "step": 2381
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.411033271154447e-09,
+      "loss": 0.8057,
+      "step": 2382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.006427019622177e-09,
+      "loss": 0.8896,
+      "step": 2383
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.610702647423586e-09,
+      "loss": 0.8574,
+      "step": 2384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.223860858781463e-09,
+      "loss": 0.9175,
+      "step": 2385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.84590234211291e-09,
+      "loss": 0.9189,
+      "step": 2386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.476827770025452e-09,
+      "loss": 0.9639,
+      "step": 2387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.116637799317594e-09,
+      "loss": 0.8862,
+      "step": 2388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.765333070976599e-09,
+      "loss": 0.8848,
+      "step": 2389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.422914210177933e-09,
+      "loss": 0.8398,
+      "step": 2390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.089381826284158e-09,
+      "loss": 0.9507,
+      "step": 2391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.7647365128421505e-09,
+      "loss": 0.9341,
+      "step": 2392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.448978847585884e-09,
+      "loss": 1.0654,
+      "step": 2393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.142109392431427e-09,
+      "loss": 0.8745,
+      "step": 2394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.844128693477501e-09,
+      "loss": 0.9434,
+      "step": 2395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.555037281006036e-09,
+      "loss": 0.915,
+      "step": 2396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2748356694777285e-09,
+      "loss": 0.9502,
+      "step": 2397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.003524357534261e-09,
+      "loss": 0.8428,
+      "step": 2398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7411038279971945e-09,
+      "loss": 1.0234,
+      "step": 2399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.487574547863526e-09,
+      "loss": 0.9365,
+      "step": 2400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2429369683112387e-09,
+      "loss": 0.9497,
+      "step": 2401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0071915246909776e-09,
+      "loss": 0.9297,
+      "step": 2402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7803386365321537e-09,
+      "loss": 0.8452,
+      "step": 2403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5623787075373942e-09,
+      "loss": 0.9209,
+      "step": 2404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.353312125584761e-09,
+      "loss": 0.9512,
+      "step": 2405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1531392627244243e-09,
+      "loss": 0.9131,
+      "step": 2406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.961860475180877e-09,
+      "loss": 0.9248,
+      "step": 2407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7794761033496089e-09,
+      "loss": 0.9092,
+      "step": 2408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6059864717993256e-09,
+      "loss": 0.8955,
+      "step": 2409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4413918892686173e-09,
+      "loss": 0.8442,
+      "step": 2410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2856926486665145e-09,
+      "loss": 0.8481,
+      "step": 2411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1388890270730423e-09,
+      "loss": 0.9399,
+      "step": 2412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0009812857370016e-09,
+      "loss": 0.9414,
+      "step": 2413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.719696700765223e-10,
+      "loss": 0.8647,
+      "step": 2414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.518544096779545e-10,
+      "loss": 0.9414,
+      "step": 2415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.40635718296978e-10,
+      "loss": 0.8828,
+      "step": 2416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.383137938552718e-10,
+      "loss": 0.9102,
+      "step": 2417
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.448888184432898e-10,
+      "loss": 1.0342,
+      "step": 2418
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.603609583191503e-10,
+      "loss": 0.9604,
+      "step": 2419
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.847303639064158e-10,
+      "loss": 0.9634,
+      "step": 2420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1799716979575836e-10,
+      "loss": 0.9917,
+      "step": 2421
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6016149474440412e-10,
+      "loss": 0.8525,
+      "step": 2422
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1122344167613374e-10,
+      "loss": 0.9614,
+      "step": 2423
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.118309768017196e-11,
+      "loss": 0.9102,
+      "step": 2424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.004053401063246e-11,
+      "loss": 0.8062,
+      "step": 2425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7795806089293545e-11,
+      "loss": 0.9956,
+      "step": 2426
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4489535017122606e-12,
+      "loss": 0.8955,
+      "step": 2427
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.9678,
+      "step": 2428
+    },
+    {
+      "epoch": 1.0,
+      "step": 2428,
+      "total_flos": 35620199940096.0,
+      "train_loss": 0.9926470233345346,
+      "train_runtime": 12260.4584,
+      "train_samples_per_second": 4.751,
+      "train_steps_per_second": 0.198
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2428,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 35620199940096.0,
+  "trial_name": null,
+  "trial_params": null
+}