diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,68964 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999347244282948,
+  "eval_steps": 500,
+  "global_step": 11489,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.797101449275363e-08,
+      "loss": 1.281,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1594202898550726e-07,
+      "loss": 1.3388,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7391304347826088e-07,
+      "loss": 1.2981,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3188405797101452e-07,
+      "loss": 1.3312,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8985507246376816e-07,
+      "loss": 1.2763,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4782608695652175e-07,
+      "loss": 1.3213,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0579710144927534e-07,
+      "loss": 1.259,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6376811594202904e-07,
+      "loss": 1.2439,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.217391304347826e-07,
+      "loss": 1.2542,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.797101449275363e-07,
+      "loss": 1.2814,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.376811594202899e-07,
+      "loss": 1.2381,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.956521739130435e-07,
+      "loss": 1.223,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.536231884057972e-07,
+      "loss": 1.2625,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.115942028985507e-07,
+      "loss": 1.2828,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.695652173913044e-07,
+      "loss": 1.1916,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.275362318840581e-07,
+      "loss": 1.3293,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.855072463768117e-07,
+      "loss": 1.2874,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0434782608695653e-06,
+      "loss": 1.1793,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1014492753623188e-06,
+      "loss": 1.1953,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1594202898550726e-06,
+      "loss": 1.2812,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2173913043478262e-06,
+      "loss": 1.2396,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2753623188405798e-06,
+      "loss": 1.2422,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.2589,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.391304347826087e-06,
+      "loss": 1.2329,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4492753623188408e-06,
+      "loss": 1.2989,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5072463768115944e-06,
+      "loss": 1.1888,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.565217391304348e-06,
+      "loss": 1.1603,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6231884057971014e-06,
+      "loss": 1.2652,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6811594202898552e-06,
+      "loss": 1.2371,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7391304347826088e-06,
+      "loss": 1.2761,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7971014492753626e-06,
+      "loss": 1.1578,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8550724637681161e-06,
+      "loss": 1.2567,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9130434782608697e-06,
+      "loss": 1.2224,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9710144927536233e-06,
+      "loss": 1.1475,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.028985507246377e-06,
+      "loss": 1.2573,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0869565217391305e-06,
+      "loss": 1.1953,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.144927536231884e-06,
+      "loss": 1.1578,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2028985507246377e-06,
+      "loss": 1.1711,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2608695652173913e-06,
+      "loss": 1.1396,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3188405797101453e-06,
+      "loss": 1.1533,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.376811594202899e-06,
+      "loss": 1.0959,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4347826086956525e-06,
+      "loss": 1.1415,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.492753623188406e-06,
+      "loss": 1.1431,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5507246376811596e-06,
+      "loss": 1.119,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6086956521739132e-06,
+      "loss": 1.113,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.0807,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7246376811594204e-06,
+      "loss": 1.133,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.782608695652174e-06,
+      "loss": 1.1218,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.840579710144928e-06,
+      "loss": 1.0724,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8985507246376816e-06,
+      "loss": 1.0115,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.956521739130435e-06,
+      "loss": 1.0558,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.014492753623189e-06,
+      "loss": 1.0734,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0724637681159424e-06,
+      "loss": 0.9858,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.130434782608696e-06,
+      "loss": 1.0028,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.188405797101449e-06,
+      "loss": 0.9664,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2463768115942027e-06,
+      "loss": 1.0654,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3043478260869567e-06,
+      "loss": 0.9802,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3623188405797103e-06,
+      "loss": 0.9767,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.420289855072464e-06,
+      "loss": 0.9461,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4782608695652175e-06,
+      "loss": 1.0184,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.536231884057971e-06,
+      "loss": 0.9171,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.594202898550725e-06,
+      "loss": 0.9795,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6521739130434787e-06,
+      "loss": 0.9263,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7101449275362323e-06,
+      "loss": 0.9265,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.768115942028986e-06,
+      "loss": 0.928,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8260869565217395e-06,
+      "loss": 0.9257,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8840579710144935e-06,
+      "loss": 0.9361,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.942028985507247e-06,
+      "loss": 0.9015,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.9049,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.057971014492754e-06,
+      "loss": 0.9035,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.115942028985507e-06,
+      "loss": 0.8888,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.173913043478261e-06,
+      "loss": 0.9083,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.231884057971015e-06,
+      "loss": 0.9127,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.289855072463768e-06,
+      "loss": 0.9184,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 0.8722,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.405797101449275e-06,
+      "loss": 0.8999,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.463768115942029e-06,
+      "loss": 0.8845,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5217391304347826e-06,
+      "loss": 0.8282,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5797101449275366e-06,
+      "loss": 0.8938,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.637681159420291e-06,
+      "loss": 0.7787,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.695652173913044e-06,
+      "loss": 0.8601,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.753623188405798e-06,
+      "loss": 0.8369,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.811594202898551e-06,
+      "loss": 0.8715,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.869565217391305e-06,
+      "loss": 0.8254,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.927536231884059e-06,
+      "loss": 0.8445,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985507246376812e-06,
+      "loss": 0.8202,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.043478260869565e-06,
+      "loss": 0.8664,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.101449275362319e-06,
+      "loss": 0.8311,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1594202898550725e-06,
+      "loss": 0.8074,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2173913043478265e-06,
+      "loss": 0.7938,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.27536231884058e-06,
+      "loss": 0.8303,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.8182,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.391304347826088e-06,
+      "loss": 0.8199,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.449275362318841e-06,
+      "loss": 0.79,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.507246376811595e-06,
+      "loss": 0.7719,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.565217391304348e-06,
+      "loss": 0.7756,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.623188405797102e-06,
+      "loss": 0.789,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.681159420289856e-06,
+      "loss": 0.7963,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.739130434782609e-06,
+      "loss": 0.7726,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.797101449275363e-06,
+      "loss": 0.7904,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.855072463768116e-06,
+      "loss": 0.6929,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.91304347826087e-06,
+      "loss": 0.798,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9710144927536236e-06,
+      "loss": 0.772,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.028985507246378e-06,
+      "loss": 0.7072,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.086956521739132e-06,
+      "loss": 0.792,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.144927536231885e-06,
+      "loss": 0.7046,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.202898550724639e-06,
+      "loss": 0.7245,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.260869565217392e-06,
+      "loss": 0.7226,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.318840579710146e-06,
+      "loss": 0.692,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.376811594202898e-06,
+      "loss": 0.7305,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.434782608695652e-06,
+      "loss": 0.7199,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4927536231884055e-06,
+      "loss": 0.683,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5507246376811595e-06,
+      "loss": 0.6747,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6086956521739135e-06,
+      "loss": 0.7426,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.7729,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.724637681159421e-06,
+      "loss": 0.6586,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.782608695652174e-06,
+      "loss": 0.6623,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.840579710144928e-06,
+      "loss": 0.6761,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.898550724637682e-06,
+      "loss": 0.6416,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.956521739130435e-06,
+      "loss": 0.7035,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.014492753623189e-06,
+      "loss": 0.7101,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.072463768115942e-06,
+      "loss": 0.7036,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.130434782608696e-06,
+      "loss": 0.6692,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.18840579710145e-06,
+      "loss": 0.7021,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.246376811594203e-06,
+      "loss": 0.6915,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.304347826086957e-06,
+      "loss": 0.6724,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.3623188405797106e-06,
+      "loss": 0.6352,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.420289855072465e-06,
+      "loss": 0.6694,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.478260869565218e-06,
+      "loss": 0.6839,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.536231884057972e-06,
+      "loss": 0.6573,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.594202898550726e-06,
+      "loss": 0.6429,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.652173913043479e-06,
+      "loss": 0.6521,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.710144927536232e-06,
+      "loss": 0.672,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.768115942028987e-06,
+      "loss": 0.684,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.82608695652174e-06,
+      "loss": 0.6301,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.884057971014493e-06,
+      "loss": 0.6972,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.942028985507248e-06,
+      "loss": 0.649,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.614,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.057971014492754e-06,
+      "loss": 0.6547,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.115942028985508e-06,
+      "loss": 0.6516,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.173913043478263e-06,
+      "loss": 0.6922,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.231884057971014e-06,
+      "loss": 0.6155,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.289855072463769e-06,
+      "loss": 0.6591,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.347826086956522e-06,
+      "loss": 0.5749,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.405797101449275e-06,
+      "loss": 0.6809,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.46376811594203e-06,
+      "loss": 0.6469,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.521739130434783e-06,
+      "loss": 0.6208,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.579710144927536e-06,
+      "loss": 0.6333,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.63768115942029e-06,
+      "loss": 0.6497,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.695652173913044e-06,
+      "loss": 0.6343,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.753623188405798e-06,
+      "loss": 0.6336,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.81159420289855e-06,
+      "loss": 0.6391,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.869565217391306e-06,
+      "loss": 0.656,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.927536231884059e-06,
+      "loss": 0.6025,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.985507246376812e-06,
+      "loss": 0.6087,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.043478260869565e-06,
+      "loss": 0.6391,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.10144927536232e-06,
+      "loss": 0.6042,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.159420289855073e-06,
+      "loss": 0.618,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.217391304347826e-06,
+      "loss": 0.6254,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.275362318840581e-06,
+      "loss": 0.6055,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.5842,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.391304347826087e-06,
+      "loss": 0.5876,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.449275362318842e-06,
+      "loss": 0.5505,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.507246376811596e-06,
+      "loss": 0.6238,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.565217391304349e-06,
+      "loss": 0.6258,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.623188405797102e-06,
+      "loss": 0.5796,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.681159420289857e-06,
+      "loss": 0.6233,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.73913043478261e-06,
+      "loss": 0.5891,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.797101449275363e-06,
+      "loss": 0.533,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.855072463768118e-06,
+      "loss": 0.5612,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.913043478260871e-06,
+      "loss": 0.5883,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.971014492753624e-06,
+      "loss": 0.5592,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0028985507246377e-05,
+      "loss": 0.6008,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.008695652173913e-05,
+      "loss": 0.5742,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0144927536231885e-05,
+      "loss": 0.6027,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0202898550724639e-05,
+      "loss": 0.5544,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0260869565217393e-05,
+      "loss": 0.5552,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0318840579710145e-05,
+      "loss": 0.5355,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.03768115942029e-05,
+      "loss": 0.5635,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0434782608695653e-05,
+      "loss": 0.5936,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0492753623188408e-05,
+      "loss": 0.6378,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.055072463768116e-05,
+      "loss": 0.6266,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0608695652173914e-05,
+      "loss": 0.5826,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.5655,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0724637681159422e-05,
+      "loss": 0.5917,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0782608695652175e-05,
+      "loss": 0.6051,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.084057971014493e-05,
+      "loss": 0.5169,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0898550724637682e-05,
+      "loss": 0.5301,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0956521739130435e-05,
+      "loss": 0.6032,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.101449275362319e-05,
+      "loss": 0.5528,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1072463768115941e-05,
+      "loss": 0.5605,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1130434782608696e-05,
+      "loss": 0.5187,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.118840579710145e-05,
+      "loss": 0.6074,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1246376811594204e-05,
+      "loss": 0.4936,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1304347826086957e-05,
+      "loss": 0.5163,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1362318840579712e-05,
+      "loss": 0.5301,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1420289855072464e-05,
+      "loss": 0.5843,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1478260869565218e-05,
+      "loss": 0.5158,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1536231884057972e-05,
+      "loss": 0.5218,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1594202898550726e-05,
+      "loss": 0.5922,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1652173913043478e-05,
+      "loss": 0.5604,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1710144927536233e-05,
+      "loss": 0.5683,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1768115942028986e-05,
+      "loss": 0.5603,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.182608695652174e-05,
+      "loss": 0.5536,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1884057971014494e-05,
+      "loss": 0.5271,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1942028985507247e-05,
+      "loss": 0.5343,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 0.5708,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2057971014492755e-05,
+      "loss": 0.512,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2115942028985508e-05,
+      "loss": 0.5407,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2173913043478263e-05,
+      "loss": 0.5817,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2231884057971015e-05,
+      "loss": 0.5192,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.228985507246377e-05,
+      "loss": 0.5668,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2347826086956523e-05,
+      "loss": 0.5518,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2405797101449278e-05,
+      "loss": 0.5105,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2463768115942029e-05,
+      "loss": 0.4886,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2521739130434784e-05,
+      "loss": 0.5087,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2579710144927537e-05,
+      "loss": 0.5218,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2637681159420292e-05,
+      "loss": 0.5074,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2695652173913045e-05,
+      "loss": 0.4843,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2753623188405797e-05,
+      "loss": 0.5826,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2811594202898551e-05,
+      "loss": 0.5185,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2869565217391305e-05,
+      "loss": 0.4943,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.292753623188406e-05,
+      "loss": 0.5227,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2985507246376811e-05,
+      "loss": 0.4935,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3043478260869566e-05,
+      "loss": 0.5238,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3101449275362319e-05,
+      "loss": 0.4964,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3159420289855074e-05,
+      "loss": 0.5294,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3217391304347827e-05,
+      "loss": 0.5211,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3275362318840582e-05,
+      "loss": 0.5042,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.5404,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3391304347826088e-05,
+      "loss": 0.5153,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3449275362318841e-05,
+      "loss": 0.5426,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3507246376811596e-05,
+      "loss": 0.5087,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3565217391304348e-05,
+      "loss": 0.546,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3623188405797103e-05,
+      "loss": 0.5522,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3681159420289856e-05,
+      "loss": 0.4726,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.373913043478261e-05,
+      "loss": 0.5181,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3797101449275364e-05,
+      "loss": 0.5631,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3855072463768119e-05,
+      "loss": 0.5134,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.391304347826087e-05,
+      "loss": 0.5048,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3971014492753625e-05,
+      "loss": 0.5316,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4028985507246378e-05,
+      "loss": 0.4921,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4086956521739133e-05,
+      "loss": 0.5246,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4144927536231884e-05,
+      "loss": 0.4795,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.420289855072464e-05,
+      "loss": 0.5206,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4260869565217392e-05,
+      "loss": 0.4945,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4318840579710147e-05,
+      "loss": 0.4943,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.43768115942029e-05,
+      "loss": 0.4732,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4434782608695654e-05,
+      "loss": 0.5324,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4492753623188407e-05,
+      "loss": 0.495,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.455072463768116e-05,
+      "loss": 0.5218,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4608695652173915e-05,
+      "loss": 0.491,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.5297,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4724637681159421e-05,
+      "loss": 0.4621,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4782608695652174e-05,
+      "loss": 0.5151,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.484057971014493e-05,
+      "loss": 0.5214,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.489855072463768e-05,
+      "loss": 0.4911,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4956521739130436e-05,
+      "loss": 0.4973,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5014492753623189e-05,
+      "loss": 0.4772,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5072463768115944e-05,
+      "loss": 0.5078,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5130434782608697e-05,
+      "loss": 0.4955,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5188405797101452e-05,
+      "loss": 0.5032,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5246376811594203e-05,
+      "loss": 0.5245,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5304347826086958e-05,
+      "loss": 0.5183,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.536231884057971e-05,
+      "loss": 0.4649,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5420289855072464e-05,
+      "loss": 0.4971,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5478260869565217e-05,
+      "loss": 0.5172,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5536231884057974e-05,
+      "loss": 0.4908,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5594202898550727e-05,
+      "loss": 0.5,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.565217391304348e-05,
+      "loss": 0.4761,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5710144927536233e-05,
+      "loss": 0.4816,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5768115942028987e-05,
+      "loss": 0.4992,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.582608695652174e-05,
+      "loss": 0.4793,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5884057971014496e-05,
+      "loss": 0.5043,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5942028985507246e-05,
+      "loss": 0.5238,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.5021,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6057971014492756e-05,
+      "loss": 0.5228,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.611594202898551e-05,
+      "loss": 0.4978,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6173913043478262e-05,
+      "loss": 0.5126,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6231884057971015e-05,
+      "loss": 0.4785,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.628985507246377e-05,
+      "loss": 0.4718,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6347826086956525e-05,
+      "loss": 0.5302,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6405797101449278e-05,
+      "loss": 0.4767,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6463768115942028e-05,
+      "loss": 0.5161,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6521739130434785e-05,
+      "loss": 0.4822,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6579710144927538e-05,
+      "loss": 0.4611,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.663768115942029e-05,
+      "loss": 0.478,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6695652173913044e-05,
+      "loss": 0.5262,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6753623188405797e-05,
+      "loss": 0.467,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.681159420289855e-05,
+      "loss": 0.4926,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6869565217391307e-05,
+      "loss": 0.5052,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.692753623188406e-05,
+      "loss": 0.5066,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6985507246376813e-05,
+      "loss": 0.5472,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7043478260869566e-05,
+      "loss": 0.5043,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.710144927536232e-05,
+      "loss": 0.4749,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7159420289855073e-05,
+      "loss": 0.4778,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.721739130434783e-05,
+      "loss": 0.5175,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.727536231884058e-05,
+      "loss": 0.4886,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.4759,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.739130434782609e-05,
+      "loss": 0.5197,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7449275362318842e-05,
+      "loss": 0.4998,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7507246376811595e-05,
+      "loss": 0.5023,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.756521739130435e-05,
+      "loss": 0.5188,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.76231884057971e-05,
+      "loss": 0.4687,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7681159420289858e-05,
+      "loss": 0.5156,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.773913043478261e-05,
+      "loss": 0.5075,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7797101449275364e-05,
+      "loss": 0.5442,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7855072463768118e-05,
+      "loss": 0.4504,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.791304347826087e-05,
+      "loss": 0.4805,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7971014492753624e-05,
+      "loss": 0.5166,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.802898550724638e-05,
+      "loss": 0.4979,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.808695652173913e-05,
+      "loss": 0.4921,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8144927536231887e-05,
+      "loss": 0.5073,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.820289855072464e-05,
+      "loss": 0.4647,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8260869565217393e-05,
+      "loss": 0.4777,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8318840579710146e-05,
+      "loss": 0.4715,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.83768115942029e-05,
+      "loss": 0.4599,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8434782608695653e-05,
+      "loss": 0.492,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8492753623188406e-05,
+      "loss": 0.5111,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8550724637681162e-05,
+      "loss": 0.4547,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8608695652173912e-05,
+      "loss": 0.541,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.4972,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8724637681159422e-05,
+      "loss": 0.5237,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8782608695652175e-05,
+      "loss": 0.4739,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8840579710144928e-05,
+      "loss": 0.4893,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8898550724637685e-05,
+      "loss": 0.5082,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8956521739130434e-05,
+      "loss": 0.5275,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.901449275362319e-05,
+      "loss": 0.4738,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9072463768115944e-05,
+      "loss": 0.4831,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9130434782608697e-05,
+      "loss": 0.4719,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.918840579710145e-05,
+      "loss": 0.5296,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9246376811594204e-05,
+      "loss": 0.4939,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9304347826086957e-05,
+      "loss": 0.4734,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9362318840579713e-05,
+      "loss": 0.4886,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9420289855072467e-05,
+      "loss": 0.4818,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.947826086956522e-05,
+      "loss": 0.4735,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9536231884057973e-05,
+      "loss": 0.4673,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9594202898550726e-05,
+      "loss": 0.4793,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.965217391304348e-05,
+      "loss": 0.5311,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9710144927536236e-05,
+      "loss": 0.4646,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9768115942028986e-05,
+      "loss": 0.4786,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9826086956521742e-05,
+      "loss": 0.5056,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9884057971014495e-05,
+      "loss": 0.4746,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994202898550725e-05,
+      "loss": 0.4882,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4682,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999602636897e-05,
+      "loss": 0.4919,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998410547613e-05,
+      "loss": 0.4631,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999642373225e-05,
+      "loss": 0.4752,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999993642190964e-05,
+      "loss": 0.4988,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999006592397e-05,
+      "loss": 0.4811,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999985694931557e-05,
+      "loss": 0.4795,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998052921407e-05,
+      "loss": 0.4853,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999974568771928e-05,
+      "loss": 0.4655,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999967813605593e-05,
+      "loss": 0.4921,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999960263715606e-05,
+      "loss": 0.4646,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995191910257e-05,
+      "loss": 0.4813,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999942779767147e-05,
+      "loss": 0.4696,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993284571006e-05,
+      "loss": 0.4908,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999922116932104e-05,
+      "loss": 0.5165,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999910593434128e-05,
+      "loss": 0.4803,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999898275217047e-05,
+      "loss": 0.462,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999988516228184e-05,
+      "loss": 0.4942,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999987125462955e-05,
+      "loss": 0.447,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999856552261288e-05,
+      "loss": 0.5121,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999841055178216e-05,
+      "loss": 0.4673,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999824763381563e-05,
+      "loss": 0.4885,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999980767687263e-05,
+      "loss": 0.4802,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999789795652777e-05,
+      "loss": 0.4796,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999771119723413e-05,
+      "loss": 0.4746,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999751649086034e-05,
+      "loss": 0.4613,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999973138374218e-05,
+      "loss": 0.506,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999710323693466e-05,
+      "loss": 0.4384,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999688468941565e-05,
+      "loss": 0.4964,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999665819488214e-05,
+      "loss": 0.4753,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999964237533521e-05,
+      "loss": 0.5032,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999961813648442e-05,
+      "loss": 0.4484,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999593102937768e-05,
+      "loss": 0.4641,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999567274697244e-05,
+      "loss": 0.5128,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99995406517649e-05,
+      "loss": 0.482,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999513234142854e-05,
+      "loss": 0.5269,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999485021833283e-05,
+      "loss": 0.4456,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999456014838432e-05,
+      "loss": 0.4934,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999426213160603e-05,
+      "loss": 0.4288,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999395616802165e-05,
+      "loss": 0.4834,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999364225765548e-05,
+      "loss": 0.5068,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999933204005325e-05,
+      "loss": 0.5029,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999929905966783e-05,
+      "loss": 0.4996,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999265284611904e-05,
+      "loss": 0.4702,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999923071488816e-05,
+      "loss": 0.4575,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999919535049934e-05,
+      "loss": 0.4796,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999915919144826e-05,
+      "loss": 0.451,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999122237737798e-05,
+      "loss": 0.4757,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999084489370882e-05,
+      "loss": 0.4727,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999045946350514e-05,
+      "loss": 0.4137,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999900660867976e-05,
+      "loss": 0.5027,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998966476361744e-05,
+      "loss": 0.4399,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999892554939965e-05,
+      "loss": 0.4765,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998883827796744e-05,
+      "loss": 0.4903,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999884131155633e-05,
+      "loss": 0.4338,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998798000681794e-05,
+      "loss": 0.4936,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998753895176576e-05,
+      "loss": 0.4861,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998708995044176e-05,
+      "loss": 0.4256,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998663300288166e-05,
+      "loss": 0.4782,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998616810912185e-05,
+      "loss": 0.4771,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998569526919915e-05,
+      "loss": 0.4565,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999852144831512e-05,
+      "loss": 0.5183,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999847257510162e-05,
+      "loss": 0.4977,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99984229072833e-05,
+      "loss": 0.4459,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999837244486411e-05,
+      "loss": 0.4338,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998321187848054e-05,
+      "loss": 0.4489,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998269136239212e-05,
+      "loss": 0.4556,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999821629004171e-05,
+      "loss": 0.4747,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999816264925976e-05,
+      "loss": 0.4722,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999810821389762e-05,
+      "loss": 0.5119,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998052983959617e-05,
+      "loss": 0.5534,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999799695945014e-05,
+      "loss": 0.4972,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999794014037364e-05,
+      "loss": 0.4652,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997882526734632e-05,
+      "loss": 0.4792,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997824118537696e-05,
+      "loss": 0.4654,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997764915787474e-05,
+      "loss": 0.4694,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997704918488675e-05,
+      "loss": 0.4703,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999764412664606e-05,
+      "loss": 0.4339,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997582540264462e-05,
+      "loss": 0.4667,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999752015934878e-05,
+      "loss": 0.4575,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997456983903967e-05,
+      "loss": 0.4457,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997393013935044e-05,
+      "loss": 0.4676,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99973282494471e-05,
+      "loss": 0.4694,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997262690445275e-05,
+      "loss": 0.4775,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999719633693478e-05,
+      "loss": 0.4944,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997129188920895e-05,
+      "loss": 0.4836,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997061246408947e-05,
+      "loss": 0.4793,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996992509404344e-05,
+      "loss": 0.5092,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996922977912546e-05,
+      "loss": 0.454,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996852651939078e-05,
+      "loss": 0.4876,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996781531489528e-05,
+      "loss": 0.4635,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999670961656955e-05,
+      "loss": 0.4546,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996636907184856e-05,
+      "loss": 0.4753,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996563403341226e-05,
+      "loss": 0.4811,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996489105044507e-05,
+      "loss": 0.4717,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996414012300594e-05,
+      "loss": 0.4793,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996338125115465e-05,
+      "loss": 0.4679,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996261443495145e-05,
+      "loss": 0.4879,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996183967445725e-05,
+      "loss": 0.458,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999610569697337e-05,
+      "loss": 0.4422,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996026632084294e-05,
+      "loss": 0.4351,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995946772784785e-05,
+      "loss": 0.4202,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995866119081187e-05,
+      "loss": 0.5283,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995784670979906e-05,
+      "loss": 0.5123,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995702428487423e-05,
+      "loss": 0.4758,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999561939161027e-05,
+      "loss": 0.4509,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999553556035505e-05,
+      "loss": 0.4703,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995450934728418e-05,
+      "loss": 0.4554,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995365514737107e-05,
+      "loss": 0.4465,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995279300387896e-05,
+      "loss": 0.4868,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995192291687646e-05,
+      "loss": 0.453,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995104488643267e-05,
+      "loss": 0.4443,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999501589126174e-05,
+      "loss": 0.4856,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994926499550103e-05,
+      "loss": 0.4832,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999483631351546e-05,
+      "loss": 0.4233,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999474533316498e-05,
+      "loss": 0.4036,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994653558505893e-05,
+      "loss": 0.4703,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994560989545496e-05,
+      "loss": 0.5113,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999446762629114e-05,
+      "loss": 0.4671,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994373468750248e-05,
+      "loss": 0.4883,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994278516930302e-05,
+      "loss": 0.4609,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994182770838847e-05,
+      "loss": 0.4568,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994086230483493e-05,
+      "loss": 0.4486,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993988895871917e-05,
+      "loss": 0.4473,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993890767011845e-05,
+      "loss": 0.4423,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993791843911086e-05,
+      "loss": 0.4736,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993692126577493e-05,
+      "loss": 0.4947,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993591615018995e-05,
+      "loss": 0.4591,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999349030924358e-05,
+      "loss": 0.4997,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993388209259298e-05,
+      "loss": 0.5036,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999328531507426e-05,
+      "loss": 0.465,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999318162669665e-05,
+      "loss": 0.4631,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993077144134708e-05,
+      "loss": 0.4511,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992971867396728e-05,
+      "loss": 0.4779,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992865796491085e-05,
+      "loss": 0.5042,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999275893142621e-05,
+      "loss": 0.4813,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999265127221059e-05,
+      "loss": 0.5054,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992542818852785e-05,
+      "loss": 0.4382,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999243357136141e-05,
+      "loss": 0.4628,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992323529745157e-05,
+      "loss": 0.4822,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992212694012757e-05,
+      "loss": 0.4655,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999210106417303e-05,
+      "loss": 0.4341,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991988640234842e-05,
+      "loss": 0.4617,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991875422207126e-05,
+      "loss": 0.4814,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991761410098886e-05,
+      "loss": 0.4705,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991646603919176e-05,
+      "loss": 0.4519,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991531003677127e-05,
+      "loss": 0.4248,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999141460938192e-05,
+      "loss": 0.4473,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991297421042807e-05,
+      "loss": 0.47,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991179438669105e-05,
+      "loss": 0.4748,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991060662270185e-05,
+      "loss": 0.4796,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990941091855486e-05,
+      "loss": 0.4372,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990820727434517e-05,
+      "loss": 0.4634,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999069956901684e-05,
+      "loss": 0.4724,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990577616612083e-05,
+      "loss": 0.4027,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999045487022994e-05,
+      "loss": 0.4581,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999033132988016e-05,
+      "loss": 0.5132,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999020699557257e-05,
+      "loss": 0.4376,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990081867317043e-05,
+      "loss": 0.4708,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998995594512353e-05,
+      "loss": 0.444,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989829229002034e-05,
+      "loss": 0.4594,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989701718962625e-05,
+      "loss": 0.4341,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989573415015443e-05,
+      "loss": 0.4601,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998944431717068e-05,
+      "loss": 0.4364,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998931442543859e-05,
+      "loss": 0.4347,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989183739829508e-05,
+      "loss": 0.4468,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989052260353808e-05,
+      "loss": 0.4373,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998891998702195e-05,
+      "loss": 0.4398,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988786919844437e-05,
+      "loss": 0.4347,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998865305883185e-05,
+      "loss": 0.4544,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988518403994823e-05,
+      "loss": 0.4377,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998838295534406e-05,
+      "loss": 0.4293,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998824671289033e-05,
+      "loss": 0.4431,
+      "step": 517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998810967664445e-05,
+      "loss": 0.4384,
+      "step": 518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987971846617315e-05,
+      "loss": 0.5038,
+      "step": 519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987833222819883e-05,
+      "loss": 0.5084,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998769380526317e-05,
+      "loss": 0.4673,
+      "step": 521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998755359395825e-05,
+      "loss": 0.4396,
+      "step": 522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998741258891627e-05,
+      "loss": 0.4609,
+      "step": 523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987270790148433e-05,
+      "loss": 0.4683,
+      "step": 524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987128197666018e-05,
+      "loss": 0.4672,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998698481148034e-05,
+      "loss": 0.4559,
+      "step": 526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998684063160281e-05,
+      "loss": 0.4818,
+      "step": 527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986695658044883e-05,
+      "loss": 0.4915,
+      "step": 528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986549890818076e-05,
+      "loss": 0.5107,
+      "step": 529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986403329933974e-05,
+      "loss": 0.4659,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986255975404228e-05,
+      "loss": 0.472,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986107827240543e-05,
+      "loss": 0.4238,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99859588854547e-05,
+      "loss": 0.4725,
+      "step": 533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998580915005853e-05,
+      "loss": 0.4778,
+      "step": 534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985658621063936e-05,
+      "loss": 0.4388,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998550729848288e-05,
+      "loss": 0.4742,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998535518232739e-05,
+      "loss": 0.4668,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985202272609553e-05,
+      "loss": 0.4738,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985048569341517e-05,
+      "loss": 0.4725,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984894072535503e-05,
+      "loss": 0.4519,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998473878220379e-05,
+      "loss": 0.4793,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984582698358715e-05,
+      "loss": 0.4757,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984425821012684e-05,
+      "loss": 0.4388,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998426815017817e-05,
+      "loss": 0.4256,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984109685867692e-05,
+      "loss": 0.4683,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998395042809385e-05,
+      "loss": 0.493,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983790376869305e-05,
+      "loss": 0.4644,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998362953220677e-05,
+      "loss": 0.5091,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998346789411903e-05,
+      "loss": 0.4743,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998330546261893e-05,
+      "loss": 0.4712,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983142237719383e-05,
+      "loss": 0.48,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982978219433353e-05,
+      "loss": 0.4564,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982813407773882e-05,
+      "loss": 0.4696,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982647802754065e-05,
+      "loss": 0.4623,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982481404387063e-05,
+      "loss": 0.4884,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99823142126861e-05,
+      "loss": 0.4315,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982146227664467e-05,
+      "loss": 0.4283,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998197744933551e-05,
+      "loss": 0.42,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981807877712645e-05,
+      "loss": 0.4641,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998163751280934e-05,
+      "loss": 0.4479,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981466354639148e-05,
+      "loss": 0.4475,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998129440321566e-05,
+      "loss": 0.4547,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998112165855255e-05,
+      "loss": 0.449,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980948120663538e-05,
+      "loss": 0.4585,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980773789562426e-05,
+      "loss": 0.4265,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980598665263056e-05,
+      "loss": 0.4228,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980422747779354e-05,
+      "loss": 0.4836,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99802460371253e-05,
+      "loss": 0.4171,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980068533314937e-05,
+      "loss": 0.4866,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979890236362366e-05,
+      "loss": 0.4288,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979711146281765e-05,
+      "loss": 0.4253,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979531263087365e-05,
+      "loss": 0.4486,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997935058679346e-05,
+      "loss": 0.4771,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979169117414405e-05,
+      "loss": 0.4831,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997898685496463e-05,
+      "loss": 0.4445,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978803799458614e-05,
+      "loss": 0.4251,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978619950910904e-05,
+      "loss": 0.4756,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978435309336118e-05,
+      "loss": 0.4468,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978249874748924e-05,
+      "loss": 0.4514,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978063647164058e-05,
+      "loss": 0.4586,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977876626596328e-05,
+      "loss": 0.4416,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977688813060588e-05,
+      "loss": 0.4901,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977500206571766e-05,
+      "loss": 0.436,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977310807144855e-05,
+      "loss": 0.4378,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977120614794904e-05,
+      "loss": 0.4253,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997692962953703e-05,
+      "loss": 0.4595,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976737851386408e-05,
+      "loss": 0.4223,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997654528035828e-05,
+      "loss": 0.4387,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976351916467955e-05,
+      "loss": 0.4496,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976157759730792e-05,
+      "loss": 0.4548,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975962810162226e-05,
+      "loss": 0.4632,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975767067777753e-05,
+      "loss": 0.478,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997557053259292e-05,
+      "loss": 0.4598,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975373204623358e-05,
+      "loss": 0.4903,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997517508388474e-05,
+      "loss": 0.4811,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974976170392816e-05,
+      "loss": 0.4738,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974776464163387e-05,
+      "loss": 0.4633,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974575965212338e-05,
+      "loss": 0.4473,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974374673555587e-05,
+      "loss": 0.42,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997417258920914e-05,
+      "loss": 0.4197,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997396971218906e-05,
+      "loss": 0.4616,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973766042511468e-05,
+      "loss": 0.4625,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973561580192542e-05,
+      "loss": 0.451,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997335632524854e-05,
+      "loss": 0.4446,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997315027769577e-05,
+      "loss": 0.4884,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997294343755061e-05,
+      "loss": 0.4035,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972735804829502e-05,
+      "loss": 0.3913,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972527379548938e-05,
+      "loss": 0.4383,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972318161725482e-05,
+      "loss": 0.5105,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972108151375772e-05,
+      "loss": 0.5246,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971897348516486e-05,
+      "loss": 0.4403,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971685753164386e-05,
+      "loss": 0.4557,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971473365336284e-05,
+      "loss": 0.4705,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997126018504906e-05,
+      "loss": 0.444,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971046212319655e-05,
+      "loss": 0.4593,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970831447165075e-05,
+      "loss": 0.4828,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970615889602386e-05,
+      "loss": 0.452,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970399539648718e-05,
+      "loss": 0.4269,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997018239732127e-05,
+      "loss": 0.4406,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969964462637295e-05,
+      "loss": 0.4442,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969745735614118e-05,
+      "loss": 0.4563,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969526216269114e-05,
+      "loss": 0.4525,
+      "step": 622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996930590461973e-05,
+      "loss": 0.4313,
+      "step": 623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969084800683482e-05,
+      "loss": 0.4601,
+      "step": 624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968862904477936e-05,
+      "loss": 0.4919,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968640216020723e-05,
+      "loss": 0.4618,
+      "step": 626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996841673532955e-05,
+      "loss": 0.4287,
+      "step": 627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968192462422172e-05,
+      "loss": 0.4411,
+      "step": 628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967967397316412e-05,
+      "loss": 0.4288,
+      "step": 629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967741540030163e-05,
+      "loss": 0.4477,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967514890581364e-05,
+      "loss": 0.4621,
+      "step": 631
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967287448988034e-05,
+      "loss": 0.4435,
+      "step": 632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996705921526825e-05,
+      "loss": 0.4437,
+      "step": 633
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996683018944014e-05,
+      "loss": 0.4834,
+      "step": 634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996660037152192e-05,
+      "loss": 0.4219,
+      "step": 635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966369761531843e-05,
+      "loss": 0.4667,
+      "step": 636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966138359488246e-05,
+      "loss": 0.3972,
+      "step": 637
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965906165409506e-05,
+      "loss": 0.4697,
+      "step": 638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965673179314087e-05,
+      "loss": 0.4552,
+      "step": 639
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99654394012205e-05,
+      "loss": 0.4883,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996520483114733e-05,
+      "loss": 0.4745,
+      "step": 641
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996496946911321e-05,
+      "loss": 0.4547,
+      "step": 642
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964733315136852e-05,
+      "loss": 0.5001,
+      "step": 643
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964496369237017e-05,
+      "loss": 0.4675,
+      "step": 644
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996425863143254e-05,
+      "loss": 0.4535,
+      "step": 645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996402010174232e-05,
+      "loss": 0.4496,
+      "step": 646
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963780780185302e-05,
+      "loss": 0.4169,
+      "step": 647
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963540666780513e-05,
+      "loss": 0.4487,
+      "step": 648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963299761547034e-05,
+      "loss": 0.4406,
+      "step": 649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963058064504012e-05,
+      "loss": 0.4518,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996281557567065e-05,
+      "loss": 0.4926,
+      "step": 651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996257229506623e-05,
+      "loss": 0.4534,
+      "step": 652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996232822271007e-05,
+      "loss": 0.4425,
+      "step": 653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962083358621576e-05,
+      "loss": 0.4459,
+      "step": 654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996183770282021e-05,
+      "loss": 0.4702,
+      "step": 655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961591255325496e-05,
+      "loss": 0.4512,
+      "step": 656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961344016157013e-05,
+      "loss": 0.4329,
+      "step": 657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961095985334417e-05,
+      "loss": 0.451,
+      "step": 658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960847162877414e-05,
+      "loss": 0.4779,
+      "step": 659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996059754880578e-05,
+      "loss": 0.4461,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960347143139357e-05,
+      "loss": 0.4578,
+      "step": 661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996009594589804e-05,
+      "loss": 0.4306,
+      "step": 662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959843957101788e-05,
+      "loss": 0.4572,
+      "step": 663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959591176770636e-05,
+      "loss": 0.4306,
+      "step": 664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995933760492467e-05,
+      "loss": 0.4674,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959083241584046e-05,
+      "loss": 0.4867,
+      "step": 666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995882808676897e-05,
+      "loss": 0.4253,
+      "step": 667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995857214049973e-05,
+      "loss": 0.3979,
+      "step": 668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958315402796658e-05,
+      "loss": 0.465,
+      "step": 669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995805787368016e-05,
+      "loss": 0.479,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957799553170707e-05,
+      "loss": 0.4319,
+      "step": 671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957540441288823e-05,
+      "loss": 0.4527,
+      "step": 672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957280538055106e-05,
+      "loss": 0.4505,
+      "step": 673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957019843490202e-05,
+      "loss": 0.446,
+      "step": 674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995675835761484e-05,
+      "loss": 0.4795,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956496080449794e-05,
+      "loss": 0.4728,
+      "step": 676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956233012015905e-05,
+      "loss": 0.4465,
+      "step": 677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995596915233409e-05,
+      "loss": 0.4301,
+      "step": 678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955704501425314e-05,
+      "loss": 0.395,
+      "step": 679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955439059310604e-05,
+      "loss": 0.4922,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995517282601106e-05,
+      "loss": 0.4679,
+      "step": 681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954905801547842e-05,
+      "loss": 0.4686,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954637985942168e-05,
+      "loss": 0.4415,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954369379215324e-05,
+      "loss": 0.4162,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954099981388656e-05,
+      "loss": 0.4547,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953829792483576e-05,
+      "loss": 0.4804,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995355881252155e-05,
+      "loss": 0.453,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995328704152412e-05,
+      "loss": 0.4451,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953014479512882e-05,
+      "loss": 0.4395,
+      "step": 689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99527411265095e-05,
+      "loss": 0.4698,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995246698253569e-05,
+      "loss": 0.4316,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995219204761325e-05,
+      "loss": 0.4311,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951916321764023e-05,
+      "loss": 0.4376,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995163980500992e-05,
+      "loss": 0.4959,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995136249737292e-05,
+      "loss": 0.4017,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951084398875064e-05,
+      "loss": 0.4368,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995080550953845e-05,
+      "loss": 0.4275,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950525829385243e-05,
+      "loss": 0.4069,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950245358437662e-05,
+      "loss": 0.4603,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949964096718012e-05,
+      "loss": 0.4483,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994968204424863e-05,
+      "loss": 0.4327,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949399201051943e-05,
+      "loss": 0.4796,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994911556715043e-05,
+      "loss": 0.5014,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994883114256662e-05,
+      "loss": 0.435,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948545927323123e-05,
+      "loss": 0.4316,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994825992144261e-05,
+      "loss": 0.4395,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994797312494781e-05,
+      "loss": 0.4933,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994768553786151e-05,
+      "loss": 0.4469,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994739716020657e-05,
+      "loss": 0.4237,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994710799200591e-05,
+      "loss": 0.4608,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946818033282502e-05,
+      "loss": 0.4278,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946527284059397e-05,
+      "loss": 0.4325,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946235744359697e-05,
+      "loss": 0.4379,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945943414206578e-05,
+      "loss": 0.44,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994565029362327e-05,
+      "loss": 0.4469,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945356382633067e-05,
+      "loss": 0.4743,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945061681259323e-05,
+      "loss": 0.4529,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944766189525465e-05,
+      "loss": 0.4467,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994446990745497e-05,
+      "loss": 0.4474,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994417283507139e-05,
+      "loss": 0.4392,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943874972398333e-05,
+      "loss": 0.4401,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943576319459466e-05,
+      "loss": 0.418,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943276876278532e-05,
+      "loss": 0.4789,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942976642879324e-05,
+      "loss": 0.3775,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942675619285703e-05,
+      "loss": 0.4293,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994237380552159e-05,
+      "loss": 0.4402,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942071201610976e-05,
+      "loss": 0.4708,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941767807577904e-05,
+      "loss": 0.4359,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994146362344649e-05,
+      "loss": 0.4777,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99411586492409e-05,
+      "loss": 0.4484,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940852884985383e-05,
+      "loss": 0.4424,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940546330704233e-05,
+      "loss": 0.4008,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994023898642181e-05,
+      "loss": 0.3895,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993993085216255e-05,
+      "loss": 0.4247,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939621927950925e-05,
+      "loss": 0.4544,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99393122138115e-05,
+      "loss": 0.425,
+      "step": 736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993900170976888e-05,
+      "loss": 0.4267,
+      "step": 737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938690415847744e-05,
+      "loss": 0.4167,
+      "step": 738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938378332072835e-05,
+      "loss": 0.4215,
+      "step": 739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938065458468955e-05,
+      "loss": 0.4205,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993775179506096e-05,
+      "loss": 0.4148,
+      "step": 741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937437341873788e-05,
+      "loss": 0.4384,
+      "step": 742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937122098932428e-05,
+      "loss": 0.453,
+      "step": 743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936806066261926e-05,
+      "loss": 0.4308,
+      "step": 744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936489243887407e-05,
+      "loss": 0.5083,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936171631834044e-05,
+      "loss": 0.4435,
+      "step": 746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935853230127076e-05,
+      "loss": 0.4482,
+      "step": 747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935534038791814e-05,
+      "loss": 0.5035,
+      "step": 748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993521405785362e-05,
+      "loss": 0.4425,
+      "step": 749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934893287337925e-05,
+      "loss": 0.4681,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934571727270225e-05,
+      "loss": 0.4463,
+      "step": 751
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934249377676074e-05,
+      "loss": 0.4745,
+      "step": 752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933926238581084e-05,
+      "loss": 0.4354,
+      "step": 753
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933602310010942e-05,
+      "loss": 0.4312,
+      "step": 754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993327759199139e-05,
+      "loss": 0.4802,
+      "step": 755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932952084548233e-05,
+      "loss": 0.4775,
+      "step": 756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932625787707337e-05,
+      "loss": 0.4679,
+      "step": 757
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932298701494642e-05,
+      "loss": 0.4137,
+      "step": 758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931970825936138e-05,
+      "loss": 0.4192,
+      "step": 759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993164216105788e-05,
+      "loss": 0.4477,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993131270688599e-05,
+      "loss": 0.4307,
+      "step": 761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993098246344665e-05,
+      "loss": 0.4208,
+      "step": 762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930651430766105e-05,
+      "loss": 0.4462,
+      "step": 763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930319608870664e-05,
+      "loss": 0.4417,
+      "step": 764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929986997786698e-05,
+      "loss": 0.4357,
+      "step": 765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929653597540637e-05,
+      "loss": 0.4427,
+      "step": 766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929319408158985e-05,
+      "loss": 0.4374,
+      "step": 767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928984429668296e-05,
+      "loss": 0.4393,
+      "step": 768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928648662095184e-05,
+      "loss": 0.4699,
+      "step": 769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992831210546635e-05,
+      "loss": 0.4482,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927974759808525e-05,
+      "loss": 0.4765,
+      "step": 771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992763662514853e-05,
+      "loss": 0.4118,
+      "step": 772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927297701513232e-05,
+      "loss": 0.4113,
+      "step": 773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926957988929567e-05,
+      "loss": 0.4316,
+      "step": 774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992661748742453e-05,
+      "loss": 0.4111,
+      "step": 775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992627619702519e-05,
+      "loss": 0.482,
+      "step": 776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992593411775866e-05,
+      "loss": 0.4183,
+      "step": 777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925591249652133e-05,
+      "loss": 0.4684,
+      "step": 778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925247592732856e-05,
+      "loss": 0.4212,
+      "step": 779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992490314702814e-05,
+      "loss": 0.4062,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924557912565357e-05,
+      "loss": 0.4515,
+      "step": 781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924211889371948e-05,
+      "loss": 0.4212,
+      "step": 782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992386507747541e-05,
+      "loss": 0.467,
+      "step": 783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99235174769033e-05,
+      "loss": 0.4686,
+      "step": 784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992316908768325e-05,
+      "loss": 0.4542,
+      "step": 785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922819909842945e-05,
+      "loss": 0.4258,
+      "step": 786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922469943410138e-05,
+      "loss": 0.4406,
+      "step": 787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922119188412635e-05,
+      "loss": 0.4475,
+      "step": 788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992176764487832e-05,
+      "loss": 0.4296,
+      "step": 789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921415312835122e-05,
+      "loss": 0.4267,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921062192311046e-05,
+      "loss": 0.4428,
+      "step": 791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920708283334156e-05,
+      "loss": 0.4491,
+      "step": 792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992035358593258e-05,
+      "loss": 0.4323,
+      "step": 793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99199981001345e-05,
+      "loss": 0.4453,
+      "step": 794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919641825968177e-05,
+      "loss": 0.4435,
+      "step": 795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919284763461916e-05,
+      "loss": 0.4548,
+      "step": 796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99189269126441e-05,
+      "loss": 0.4752,
+      "step": 797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918568273543163e-05,
+      "loss": 0.4158,
+      "step": 798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918208846187612e-05,
+      "loss": 0.4287,
+      "step": 799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991784863060601e-05,
+      "loss": 0.4578,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991748762682698e-05,
+      "loss": 0.4252,
+      "step": 801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917125834879218e-05,
+      "loss": 0.4059,
+      "step": 802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916763254791472e-05,
+      "loss": 0.4319,
+      "step": 803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916399886592562e-05,
+      "loss": 0.4295,
+      "step": 804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916035730311366e-05,
+      "loss": 0.4695,
+      "step": 805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915670785976818e-05,
+      "loss": 0.4476,
+      "step": 806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991530505361792e-05,
+      "loss": 0.4035,
+      "step": 807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991493853326375e-05,
+      "loss": 0.4518,
+      "step": 808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914571224943428e-05,
+      "loss": 0.4722,
+      "step": 809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991420312868614e-05,
+      "loss": 0.3993,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991383424452115e-05,
+      "loss": 0.4318,
+      "step": 811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913464572477768e-05,
+      "loss": 0.3899,
+      "step": 812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913094112585374e-05,
+      "loss": 0.4343,
+      "step": 813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991272286487341e-05,
+      "loss": 0.4631,
+      "step": 814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912350829371385e-05,
+      "loss": 0.4404,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991197800610885e-05,
+      "loss": 0.4645,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991160439511545e-05,
+      "loss": 0.4222,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911229996420875e-05,
+      "loss": 0.4348,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991085481005487e-05,
+      "loss": 0.4356,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991047883604726e-05,
+      "loss": 0.4337,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991010207442792e-05,
+      "loss": 0.386,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99097245252268e-05,
+      "loss": 0.4517,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909346188473896e-05,
+      "loss": 0.4406,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908967064199282e-05,
+      "loss": 0.4546,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990858715243308e-05,
+      "loss": 0.455,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908206453205493e-05,
+      "loss": 0.4375,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990782496654677e-05,
+      "loss": 0.4901,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907442692487226e-05,
+      "loss": 0.4542,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907059631057248e-05,
+      "loss": 0.4168,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906675782287275e-05,
+      "loss": 0.4732,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906291146207812e-05,
+      "loss": 0.4368,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990590572284943e-05,
+      "loss": 0.4471,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905519512242754e-05,
+      "loss": 0.4091,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905132514418485e-05,
+      "loss": 0.433,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904744729407376e-05,
+      "loss": 0.4042,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990435615724024e-05,
+      "loss": 0.4139,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903966797947965e-05,
+      "loss": 0.4398,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990357665156149e-05,
+      "loss": 0.4066,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990318571811182e-05,
+      "loss": 0.4354,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990279399763003e-05,
+      "loss": 0.4807,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902401490147243e-05,
+      "loss": 0.4362,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990200819569466e-05,
+      "loss": 0.4492,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901614114303533e-05,
+      "loss": 0.4492,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990121924600518e-05,
+      "loss": 0.456,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900823590830984e-05,
+      "loss": 0.4418,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900427148812387e-05,
+      "loss": 0.4392,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900029919980897e-05,
+      "loss": 0.4362,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899631904368082e-05,
+      "loss": 0.4323,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899233102005573e-05,
+      "loss": 0.4628,
+      "step": 849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898833512925067e-05,
+      "loss": 0.4454,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898433137158317e-05,
+      "loss": 0.455,
+      "step": 851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898031974737143e-05,
+      "loss": 0.4324,
+      "step": 852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897630025693423e-05,
+      "loss": 0.4146,
+      "step": 853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897227290059108e-05,
+      "loss": 0.4475,
+      "step": 854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896823767866197e-05,
+      "loss": 0.4323,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989641945914677e-05,
+      "loss": 0.491,
+      "step": 856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896014363932943e-05,
+      "loss": 0.4534,
+      "step": 857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989560848225692e-05,
+      "loss": 0.4297,
+      "step": 858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895201814150957e-05,
+      "loss": 0.4142,
+      "step": 859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894794359647372e-05,
+      "loss": 0.4486,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894386118778545e-05,
+      "loss": 0.4221,
+      "step": 861
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893977091576923e-05,
+      "loss": 0.4254,
+      "step": 862
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893567278075006e-05,
+      "loss": 0.4203,
+      "step": 863
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989315667830537e-05,
+      "loss": 0.4234,
+      "step": 864
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892745292300645e-05,
+      "loss": 0.4177,
+      "step": 865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892333120093523e-05,
+      "loss": 0.4202,
+      "step": 866
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891920161716763e-05,
+      "loss": 0.4355,
+      "step": 867
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989150641720318e-05,
+      "loss": 0.4567,
+      "step": 868
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891091886585654e-05,
+      "loss": 0.414,
+      "step": 869
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890676569897137e-05,
+      "loss": 0.4346,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989026046717063e-05,
+      "loss": 0.4527,
+      "step": 871
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98898435784392e-05,
+      "loss": 0.4462,
+      "step": 872
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889425903735984e-05,
+      "loss": 0.4201,
+      "step": 873
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988900744309417e-05,
+      "loss": 0.4508,
+      "step": 874
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888588196547018e-05,
+      "loss": 0.4179,
+      "step": 875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888168164127842e-05,
+      "loss": 0.4102,
+      "step": 876
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988774734587003e-05,
+      "loss": 0.4383,
+      "step": 877
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887325741807017e-05,
+      "loss": 0.4381,
+      "step": 878
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886903351972316e-05,
+      "loss": 0.4592,
+      "step": 879
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886480176399496e-05,
+      "loss": 0.4148,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988605621512218e-05,
+      "loss": 0.4034,
+      "step": 881
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988563146817407e-05,
+      "loss": 0.4554,
+      "step": 882
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885205935588915e-05,
+      "loss": 0.4481,
+      "step": 883
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988477961740054e-05,
+      "loss": 0.4406,
+      "step": 884
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884352513642816e-05,
+      "loss": 0.4364,
+      "step": 885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883924624349697e-05,
+      "loss": 0.4123,
+      "step": 886
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883495949555183e-05,
+      "loss": 0.4363,
+      "step": 887
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883066489293342e-05,
+      "loss": 0.4341,
+      "step": 888
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98826362435983e-05,
+      "loss": 0.4055,
+      "step": 889
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882205212504257e-05,
+      "loss": 0.4253,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881773396045467e-05,
+      "loss": 0.4922,
+      "step": 891
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881340794256247e-05,
+      "loss": 0.4126,
+      "step": 892
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880907407170974e-05,
+      "loss": 0.4501,
+      "step": 893
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880473234824093e-05,
+      "loss": 0.4238,
+      "step": 894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988003827725011e-05,
+      "loss": 0.4299,
+      "step": 895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987960253448359e-05,
+      "loss": 0.4593,
+      "step": 896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987916600655916e-05,
+      "loss": 0.3888,
+      "step": 897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987872869351152e-05,
+      "loss": 0.461,
+      "step": 898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878290595375415e-05,
+      "loss": 0.4234,
+      "step": 899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987785171218567e-05,
+      "loss": 0.4786,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987741204397716e-05,
+      "loss": 0.3906,
+      "step": 901
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876971590784824e-05,
+      "loss": 0.4372,
+      "step": 902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876530352643672e-05,
+      "loss": 0.4557,
+      "step": 903
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876088329588768e-05,
+      "loss": 0.4668,
+      "step": 904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987564552165524e-05,
+      "loss": 0.4078,
+      "step": 905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875201928878277e-05,
+      "loss": 0.4036,
+      "step": 906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987475755129314e-05,
+      "loss": 0.4507,
+      "step": 907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874312388935134e-05,
+      "loss": 0.4008,
+      "step": 908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873866441839643e-05,
+      "loss": 0.4108,
+      "step": 909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987341971004211e-05,
+      "loss": 0.4469,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872972193578034e-05,
+      "loss": 0.4352,
+      "step": 911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872523892482983e-05,
+      "loss": 0.4329,
+      "step": 912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872074806792585e-05,
+      "loss": 0.4419,
+      "step": 913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871624936542527e-05,
+      "loss": 0.4225,
+      "step": 914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987117428176856e-05,
+      "loss": 0.4277,
+      "step": 915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870722842506504e-05,
+      "loss": 0.3982,
+      "step": 916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870270618792235e-05,
+      "loss": 0.4127,
+      "step": 917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986981761066169e-05,
+      "loss": 0.4284,
+      "step": 918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869363818150867e-05,
+      "loss": 0.4471,
+      "step": 919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986890924129584e-05,
+      "loss": 0.446,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986845388013273e-05,
+      "loss": 0.4161,
+      "step": 921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867997734697725e-05,
+      "loss": 0.4447,
+      "step": 922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867540805027078e-05,
+      "loss": 0.4716,
+      "step": 923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98670830911571e-05,
+      "loss": 0.4476,
+      "step": 924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986662459312417e-05,
+      "loss": 0.4316,
+      "step": 925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986616531096472e-05,
+      "loss": 0.4876,
+      "step": 926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865705244715258e-05,
+      "loss": 0.4687,
+      "step": 927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865244394412345e-05,
+      "loss": 0.4211,
+      "step": 928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98647827600926e-05,
+      "loss": 0.4534,
+      "step": 929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864320341792716e-05,
+      "loss": 0.4185,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986385713954944e-05,
+      "loss": 0.442,
+      "step": 931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986339315339959e-05,
+      "loss": 0.4585,
+      "step": 932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986292838338003e-05,
+      "loss": 0.44,
+      "step": 933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98624628295277e-05,
+      "loss": 0.4166,
+      "step": 934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861996491879605e-05,
+      "loss": 0.4446,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98615293704728e-05,
+      "loss": 0.4236,
+      "step": 936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861061465344413e-05,
+      "loss": 0.3988,
+      "step": 937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986059277653162e-05,
+      "loss": 0.4121,
+      "step": 938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860123304071678e-05,
+      "loss": 0.4289,
+      "step": 939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859653048001896e-05,
+      "loss": 0.4872,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859182008359645e-05,
+      "loss": 0.4393,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985871018518236e-05,
+      "loss": 0.4482,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985823757850754e-05,
+      "loss": 0.3777,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857764188372738e-05,
+      "loss": 0.3785,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857290014815582e-05,
+      "loss": 0.4355,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856815057873754e-05,
+      "loss": 0.4389,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856339317584997e-05,
+      "loss": 0.4223,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855862793987123e-05,
+      "loss": 0.5,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855385487118e-05,
+      "loss": 0.4229,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985490739701557e-05,
+      "loss": 0.4473,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854428523717812e-05,
+      "loss": 0.4453,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853948867262795e-05,
+      "loss": 0.4012,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853468427688637e-05,
+      "loss": 0.4526,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852987205033516e-05,
+      "loss": 0.4245,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985250519933568e-05,
+      "loss": 0.4542,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985202241063343e-05,
+      "loss": 0.424,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985153883896514e-05,
+      "loss": 0.4279,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985105448436924e-05,
+      "loss": 0.4223,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850569346884218e-05,
+      "loss": 0.4565,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850083426548633e-05,
+      "loss": 0.4429,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984959672340111e-05,
+      "loss": 0.4158,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984910923748031e-05,
+      "loss": 0.425,
+      "step": 962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848620968824992e-05,
+      "loss": 0.4201,
+      "step": 963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984813191747395e-05,
+      "loss": 0.4539,
+      "step": 964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847642083466055e-05,
+      "loss": 0.4497,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984715146684023e-05,
+      "loss": 0.4394,
+      "step": 966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846660067635477e-05,
+      "loss": 0.4374,
+      "step": 967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846167885890837e-05,
+      "loss": 0.4108,
+      "step": 968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984567492164543e-05,
+      "loss": 0.3819,
+      "step": 969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845181174938432e-05,
+      "loss": 0.4454,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844686645809085e-05,
+      "loss": 0.4438,
+      "step": 971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844191334296685e-05,
+      "loss": 0.4451,
+      "step": 972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843695240440603e-05,
+      "loss": 0.4185,
+      "step": 973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984319836428026e-05,
+      "loss": 0.4482,
+      "step": 974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984270070585514e-05,
+      "loss": 0.4057,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842202265204803e-05,
+      "loss": 0.3813,
+      "step": 976
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841703042368854e-05,
+      "loss": 0.4458,
+      "step": 977
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841203037386974e-05,
+      "loss": 0.4094,
+      "step": 978
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984070225029889e-05,
+      "loss": 0.4254,
+      "step": 979
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984020068114441e-05,
+      "loss": 0.4433,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839698329963393e-05,
+      "loss": 0.4065,
+      "step": 981
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839195196795763e-05,
+      "loss": 0.4314,
+      "step": 982
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98386912816815e-05,
+      "loss": 0.4339,
+      "step": 983
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838186584660654e-05,
+      "loss": 0.433,
+      "step": 984
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837681105773334e-05,
+      "loss": 0.413,
+      "step": 985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837174845059716e-05,
+      "loss": 0.4241,
+      "step": 986
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983666780256003e-05,
+      "loss": 0.416,
+      "step": 987
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983615997831457e-05,
+      "loss": 0.459,
+      "step": 988
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98356513723637e-05,
+      "loss": 0.446,
+      "step": 989
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835141984747838e-05,
+      "loss": 0.4576,
+      "step": 990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834631815507465e-05,
+      "loss": 0.4507,
+      "step": 991
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834120864683123e-05,
+      "loss": 0.4232,
+      "step": 992
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833609132315427e-05,
+      "loss": 0.4133,
+      "step": 993
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833096618445037e-05,
+      "loss": 0.4378,
+      "step": 994
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832583323112685e-05,
+      "loss": 0.4132,
+      "step": 995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832069246359168e-05,
+      "loss": 0.425,
+      "step": 996
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983155438822534e-05,
+      "loss": 0.4227,
+      "step": 997
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983103874875212e-05,
+      "loss": 0.437,
+      "step": 998
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983052232798048e-05,
+      "loss": 0.4467,
+      "step": 999
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830005125951467e-05,
+      "loss": 0.4615,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829487142706185e-05,
+      "loss": 0.4104,
+      "step": 1001
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828968378285797e-05,
+      "loss": 0.4014,
+      "step": 1002
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982844883273153e-05,
+      "loss": 0.403,
+      "step": 1003
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827928506084676e-05,
+      "loss": 0.403,
+      "step": 1004
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827407398386583e-05,
+      "loss": 0.3933,
+      "step": 1005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982688550967867e-05,
+      "loss": 0.4488,
+      "step": 1006
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982636284000241e-05,
+      "loss": 0.4319,
+      "step": 1007
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825839389399344e-05,
+      "loss": 0.4474,
+      "step": 1008
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825315157911064e-05,
+      "loss": 0.4141,
+      "step": 1009
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982479014557924e-05,
+      "loss": 0.4701,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824264352445596e-05,
+      "loss": 0.4041,
+      "step": 1011
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982373777855191e-05,
+      "loss": 0.4571,
+      "step": 1012
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982321042394004e-05,
+      "loss": 0.4235,
+      "step": 1013
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822682288651888e-05,
+      "loss": 0.4221,
+      "step": 1014
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822153372729435e-05,
+      "loss": 0.4306,
+      "step": 1015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821623676214706e-05,
+      "loss": 0.4781,
+      "step": 1016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821093199149806e-05,
+      "loss": 0.4182,
+      "step": 1017
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982056194157689e-05,
+      "loss": 0.458,
+      "step": 1018
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820029903538172e-05,
+      "loss": 0.4152,
+      "step": 1019
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819497085075942e-05,
+      "loss": 0.4311,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818963486232548e-05,
+      "loss": 0.4108,
+      "step": 1021
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818429107050385e-05,
+      "loss": 0.4207,
+      "step": 1022
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817893947571933e-05,
+      "loss": 0.4409,
+      "step": 1023
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817358007839713e-05,
+      "loss": 0.4314,
+      "step": 1024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816821287896325e-05,
+      "loss": 0.4245,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816283787784418e-05,
+      "loss": 0.411,
+      "step": 1026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815745507546713e-05,
+      "loss": 0.4506,
+      "step": 1027
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815206447225986e-05,
+      "loss": 0.437,
+      "step": 1028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814666606865075e-05,
+      "loss": 0.3961,
+      "step": 1029
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981412598650689e-05,
+      "loss": 0.4302,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813584586194388e-05,
+      "loss": 0.4527,
+      "step": 1031
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98130424059706e-05,
+      "loss": 0.4241,
+      "step": 1032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812499445878615e-05,
+      "loss": 0.4154,
+      "step": 1033
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811955705961578e-05,
+      "loss": 0.4158,
+      "step": 1034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811411186262702e-05,
+      "loss": 0.44,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810865886825272e-05,
+      "loss": 0.5291,
+      "step": 1036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981031980769261e-05,
+      "loss": 0.421,
+      "step": 1037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809772948908125e-05,
+      "loss": 0.4243,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809225310515274e-05,
+      "loss": 0.4617,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808676892557577e-05,
+      "loss": 0.4519,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980812769507862e-05,
+      "loss": 0.4257,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980757771812205e-05,
+      "loss": 0.4549,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807026961731572e-05,
+      "loss": 0.4296,
+      "step": 1043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980647542595096e-05,
+      "loss": 0.4624,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980592311082404e-05,
+      "loss": 0.4421,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805370016394715e-05,
+      "loss": 0.4414,
+      "step": 1046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804816142706936e-05,
+      "loss": 0.417,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804261489804716e-05,
+      "loss": 0.4101,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803706057732147e-05,
+      "loss": 0.4429,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803149846533355e-05,
+      "loss": 0.3993,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980259285625256e-05,
+      "loss": 0.3877,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980203508693401e-05,
+      "loss": 0.4601,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801476538622045e-05,
+      "loss": 0.4311,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800917211361055e-05,
+      "loss": 0.4512,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980035710519548e-05,
+      "loss": 0.4428,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799796220169845e-05,
+      "loss": 0.4379,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979923455632872e-05,
+      "loss": 0.4899,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798672113716742e-05,
+      "loss": 0.4426,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979810889237861e-05,
+      "loss": 0.4407,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979754489235908e-05,
+      "loss": 0.4194,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796980113702985e-05,
+      "loss": 0.4413,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796414556455202e-05,
+      "loss": 0.4294,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795848220660678e-05,
+      "loss": 0.4083,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979528110636442e-05,
+      "loss": 0.4443,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794713213611502e-05,
+      "loss": 0.4897,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794144542447053e-05,
+      "loss": 0.4022,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979357509291627e-05,
+      "loss": 0.4302,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793004865064404e-05,
+      "loss": 0.4597,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792433858936776e-05,
+      "loss": 0.4404,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791862074578763e-05,
+      "loss": 0.4192,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979128951203581e-05,
+      "loss": 0.4081,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790716171353414e-05,
+      "loss": 0.4341,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790142052577148e-05,
+      "loss": 0.4421,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978956715575263e-05,
+      "loss": 0.4653,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788991480925553e-05,
+      "loss": 0.4098,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978841502814167e-05,
+      "loss": 0.4516,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978783779744679e-05,
+      "loss": 0.4507,
+      "step": 1077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787259788886786e-05,
+      "loss": 0.4099,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786681002507596e-05,
+      "loss": 0.4468,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786101438355216e-05,
+      "loss": 0.4175,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785521096475704e-05,
+      "loss": 0.4131,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784939976915188e-05,
+      "loss": 0.4194,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784358079719842e-05,
+      "loss": 0.4184,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783775404935922e-05,
+      "loss": 0.3917,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783191952609725e-05,
+      "loss": 0.4688,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782607722787622e-05,
+      "loss": 0.4075,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782022715516044e-05,
+      "loss": 0.4079,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781436930841482e-05,
+      "loss": 0.409,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780850368810494e-05,
+      "loss": 0.4397,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780263029469692e-05,
+      "loss": 0.426,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779674912865753e-05,
+      "loss": 0.4525,
+      "step": 1091
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779086019045417e-05,
+      "loss": 0.4781,
+      "step": 1092
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778496348055487e-05,
+      "loss": 0.3988,
+      "step": 1093
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777905899942823e-05,
+      "loss": 0.4359,
+      "step": 1094
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977731467475435e-05,
+      "loss": 0.4146,
+      "step": 1095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776722672537052e-05,
+      "loss": 0.4259,
+      "step": 1096
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977612989333798e-05,
+      "loss": 0.4621,
+      "step": 1097
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775536337204246e-05,
+      "loss": 0.4311,
+      "step": 1098
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977494200418302e-05,
+      "loss": 0.4436,
+      "step": 1099
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774346894321527e-05,
+      "loss": 0.3945,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773751007667074e-05,
+      "loss": 0.4045,
+      "step": 1101
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977315434426701e-05,
+      "loss": 0.4738,
+      "step": 1102
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977255690416876e-05,
+      "loss": 0.4307,
+      "step": 1103
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771958687419795e-05,
+      "loss": 0.4334,
+      "step": 1104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771359694067663e-05,
+      "loss": 0.4251,
+      "step": 1105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770759924159968e-05,
+      "loss": 0.413,
+      "step": 1106
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770159377744372e-05,
+      "loss": 0.4424,
+      "step": 1107
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976955805486861e-05,
+      "loss": 0.4344,
+      "step": 1108
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768955955580457e-05,
+      "loss": 0.397,
+      "step": 1109
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976835307992777e-05,
+      "loss": 0.4522,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767749427958468e-05,
+      "loss": 0.4552,
+      "step": 1111
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767144999720512e-05,
+      "loss": 0.4189,
+      "step": 1112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976653979526195e-05,
+      "loss": 0.4202,
+      "step": 1113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765933814630868e-05,
+      "loss": 0.4906,
+      "step": 1114
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765327057875433e-05,
+      "loss": 0.4227,
+      "step": 1115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764719525043865e-05,
+      "loss": 0.4533,
+      "step": 1116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976411121618444e-05,
+      "loss": 0.4504,
+      "step": 1117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976350213134551e-05,
+      "loss": 0.4492,
+      "step": 1118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762892270575474e-05,
+      "loss": 0.4432,
+      "step": 1119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762281633922802e-05,
+      "loss": 0.4117,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761670221436024e-05,
+      "loss": 0.4691,
+      "step": 1121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761058033163727e-05,
+      "loss": 0.3895,
+      "step": 1122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976044506915457e-05,
+      "loss": 0.4472,
+      "step": 1123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759831329457254e-05,
+      "loss": 0.4447,
+      "step": 1124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759216814120567e-05,
+      "loss": 0.4172,
+      "step": 1125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758601523193346e-05,
+      "loss": 0.4321,
+      "step": 1126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975798545672448e-05,
+      "loss": 0.385,
+      "step": 1127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757368614762938e-05,
+      "loss": 0.3964,
+      "step": 1128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756750997357738e-05,
+      "loss": 0.4027,
+      "step": 1129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756132604557965e-05,
+      "loss": 0.429,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755513436412763e-05,
+      "loss": 0.4158,
+      "step": 1131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975489349297134e-05,
+      "loss": 0.4559,
+      "step": 1132
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754272774282966e-05,
+      "loss": 0.4174,
+      "step": 1133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753651280396972e-05,
+      "loss": 0.4292,
+      "step": 1134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753029011362742e-05,
+      "loss": 0.4569,
+      "step": 1135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752405967229736e-05,
+      "loss": 0.4475,
+      "step": 1136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975178214804747e-05,
+      "loss": 0.4105,
+      "step": 1137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751157553865517e-05,
+      "loss": 0.4408,
+      "step": 1138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975053218473352e-05,
+      "loss": 0.3863,
+      "step": 1139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974990604070117e-05,
+      "loss": 0.4562,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 0.449,
+      "step": 1141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974865142813454e-05,
+      "loss": 0.4812,
+      "step": 1142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748022959699962e-05,
+      "loss": 0.4252,
+      "step": 1143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747393716564453e-05,
+      "loss": 0.4436,
+      "step": 1144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746763698778015e-05,
+      "loss": 0.4339,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746132906390724e-05,
+      "loss": 0.4598,
+      "step": 1146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745501339452704e-05,
+      "loss": 0.4288,
+      "step": 1147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744868998014155e-05,
+      "loss": 0.4211,
+      "step": 1148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744235882125325e-05,
+      "loss": 0.4161,
+      "step": 1149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743601991836527e-05,
+      "loss": 0.436,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742967327198145e-05,
+      "loss": 0.4455,
+      "step": 1151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742331888260616e-05,
+      "loss": 0.3997,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974169567507443e-05,
+      "loss": 0.4235,
+      "step": 1153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741058687690168e-05,
+      "loss": 0.4507,
+      "step": 1154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740420926158435e-05,
+      "loss": 0.396,
+      "step": 1155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739782390529925e-05,
+      "loss": 0.4343,
+      "step": 1156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973914308085538e-05,
+      "loss": 0.385,
+      "step": 1157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973850299718561e-05,
+      "loss": 0.4354,
+      "step": 1158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737862139571482e-05,
+      "loss": 0.4565,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737220508063927e-05,
+      "loss": 0.4548,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973657810271394e-05,
+      "loss": 0.4429,
+      "step": 1161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973593492357257e-05,
+      "loss": 0.4487,
+      "step": 1162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735290970690938e-05,
+      "loss": 0.4113,
+      "step": 1163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734646244120216e-05,
+      "loss": 0.4093,
+      "step": 1164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734000743911644e-05,
+      "loss": 0.4159,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973335447011652e-05,
+      "loss": 0.4498,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732707422786207e-05,
+      "loss": 0.4033,
+      "step": 1167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732059601972128e-05,
+      "loss": 0.3854,
+      "step": 1168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731411007725763e-05,
+      "loss": 0.4033,
+      "step": 1169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730761640098662e-05,
+      "loss": 0.4186,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973011149914243e-05,
+      "loss": 0.3681,
+      "step": 1171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729460584908734e-05,
+      "loss": 0.44,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972880889744931e-05,
+      "loss": 0.3654,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728156436815936e-05,
+      "loss": 0.4328,
+      "step": 1174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972750320306048e-05,
+      "loss": 0.4582,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972684919623485e-05,
+      "loss": 0.412,
+      "step": 1176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726194416391017e-05,
+      "loss": 0.4457,
+      "step": 1177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725538863581027e-05,
+      "loss": 0.4516,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724882537856972e-05,
+      "loss": 0.4593,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724225439271014e-05,
+      "loss": 0.4361,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723567567875372e-05,
+      "loss": 0.4311,
+      "step": 1181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722908923722334e-05,
+      "loss": 0.441,
+      "step": 1182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972224950686424e-05,
+      "loss": 0.4372,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97215893173535e-05,
+      "loss": 0.4131,
+      "step": 1184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972092835524257e-05,
+      "loss": 0.4304,
+      "step": 1185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720266620583993e-05,
+      "loss": 0.3983,
+      "step": 1186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719604113430347e-05,
+      "loss": 0.4557,
+      "step": 1187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971894083383429e-05,
+      "loss": 0.4209,
+      "step": 1188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718276781848536e-05,
+      "loss": 0.4246,
+      "step": 1189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971761195752585e-05,
+      "loss": 0.4372,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716946360919075e-05,
+      "loss": 0.4339,
+      "step": 1191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716279992081107e-05,
+      "loss": 0.4476,
+      "step": 1192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97156128510649e-05,
+      "loss": 0.4199,
+      "step": 1193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971494493792348e-05,
+      "loss": 0.4102,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971427625270992e-05,
+      "loss": 0.3934,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713606795477366e-05,
+      "loss": 0.4306,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712936566279026e-05,
+      "loss": 0.4209,
+      "step": 1197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712265565168157e-05,
+      "loss": 0.4176,
+      "step": 1198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711593792198088e-05,
+      "loss": 0.4163,
+      "step": 1199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971092124742221e-05,
+      "loss": 0.4198,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710247930893968e-05,
+      "loss": 0.4298,
+      "step": 1201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709573842666873e-05,
+      "loss": 0.4068,
+      "step": 1202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97088989827945e-05,
+      "loss": 0.4367,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708223351330474e-05,
+      "loss": 0.4099,
+      "step": 1204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707546948328496e-05,
+      "loss": 0.4114,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970686977384232e-05,
+      "loss": 0.4241,
+      "step": 1206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970619182792576e-05,
+      "loss": 0.431,
+      "step": 1207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.97055131106327e-05,
+      "loss": 0.4438,
+      "step": 1208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970483362201707e-05,
+      "loss": 0.3999,
+      "step": 1209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704153362132885e-05,
+      "loss": 0.398,
+      "step": 1210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703472331034193e-05,
+      "loss": 0.4188,
+      "step": 1211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702790528775123e-05,
+      "loss": 0.4296,
+      "step": 1212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702107955409862e-05,
+      "loss": 0.3972,
+      "step": 1213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701424610992657e-05,
+      "loss": 0.411,
+      "step": 1214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700740495577807e-05,
+      "loss": 0.3807,
+      "step": 1215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700055609219687e-05,
+      "loss": 0.4173,
+      "step": 1216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969936995197272e-05,
+      "loss": 0.4324,
+      "step": 1217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698683523891408e-05,
+      "loss": 0.4323,
+      "step": 1218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697996325030292e-05,
+      "loss": 0.4244,
+      "step": 1219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697308355443997e-05,
+      "loss": 0.412,
+      "step": 1220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696619615187188e-05,
+      "loss": 0.4094,
+      "step": 1221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695930104314606e-05,
+      "loss": 0.4569,
+      "step": 1222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969523982288104e-05,
+      "loss": 0.4149,
+      "step": 1223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694548770941363e-05,
+      "loss": 0.4447,
+      "step": 1224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693856948550484e-05,
+      "loss": 0.4392,
+      "step": 1225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693164355763384e-05,
+      "loss": 0.4383,
+      "step": 1226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692470992635112e-05,
+      "loss": 0.4012,
+      "step": 1227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691776859220763e-05,
+      "loss": 0.397,
+      "step": 1228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969108195557551e-05,
+      "loss": 0.3649,
+      "step": 1229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690386281754572e-05,
+      "loss": 0.4141,
+      "step": 1230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968968983781324e-05,
+      "loss": 0.4287,
+      "step": 1231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688992623806862e-05,
+      "loss": 0.4366,
+      "step": 1232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688294639790842e-05,
+      "loss": 0.4138,
+      "step": 1233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687595885820657e-05,
+      "loss": 0.4314,
+      "step": 1234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968689636195184e-05,
+      "loss": 0.429,
+      "step": 1235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686196068239977e-05,
+      "loss": 0.4383,
+      "step": 1236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685495004740728e-05,
+      "loss": 0.4054,
+      "step": 1237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684793171509805e-05,
+      "loss": 0.4219,
+      "step": 1238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684090568602988e-05,
+      "loss": 0.4501,
+      "step": 1239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683387196076113e-05,
+      "loss": 0.4586,
+      "step": 1240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682683053985073e-05,
+      "loss": 0.4209,
+      "step": 1241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681978142385837e-05,
+      "loss": 0.4074,
+      "step": 1242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681272461334422e-05,
+      "loss": 0.4177,
+      "step": 1243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968056601088691e-05,
+      "loss": 0.4137,
+      "step": 1244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679858791099448e-05,
+      "loss": 0.4762,
+      "step": 1245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679150802028236e-05,
+      "loss": 0.4153,
+      "step": 1246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678442043729542e-05,
+      "loss": 0.4098,
+      "step": 1247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677732516259694e-05,
+      "loss": 0.4013,
+      "step": 1248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677022219675076e-05,
+      "loss": 0.4078,
+      "step": 1249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676311154032142e-05,
+      "loss": 0.4018,
+      "step": 1250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96755993193874e-05,
+      "loss": 0.4137,
+      "step": 1251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674886715797422e-05,
+      "loss": 0.4122,
+      "step": 1252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674173343318837e-05,
+      "loss": 0.414,
+      "step": 1253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673459202008344e-05,
+      "loss": 0.4584,
+      "step": 1254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967274429192269e-05,
+      "loss": 0.4442,
+      "step": 1255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96720286131187e-05,
+      "loss": 0.4164,
+      "step": 1256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671312165653246e-05,
+      "loss": 0.403,
+      "step": 1257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670594949583268e-05,
+      "loss": 0.434,
+      "step": 1258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669876964965762e-05,
+      "loss": 0.4119,
+      "step": 1259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669158211857793e-05,
+      "loss": 0.4122,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668438690316477e-05,
+      "loss": 0.4395,
+      "step": 1261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667718400398997e-05,
+      "loss": 0.3926,
+      "step": 1262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666997342162596e-05,
+      "loss": 0.4108,
+      "step": 1263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666275515664585e-05,
+      "loss": 0.4706,
+      "step": 1264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665552920962324e-05,
+      "loss": 0.4617,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664829558113236e-05,
+      "loss": 0.3884,
+      "step": 1266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664105427174816e-05,
+      "loss": 0.4397,
+      "step": 1267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663380528204607e-05,
+      "loss": 0.4334,
+      "step": 1268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966265486126022e-05,
+      "loss": 0.4239,
+      "step": 1269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966192842639933e-05,
+      "loss": 0.4329,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966120122367966e-05,
+      "loss": 0.4061,
+      "step": 1271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966047325315901e-05,
+      "loss": 0.3998,
+      "step": 1272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659744514895233e-05,
+      "loss": 0.4018,
+      "step": 1273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965901500894624e-05,
+      "loss": 0.3987,
+      "step": 1274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965828473537001e-05,
+      "loss": 0.4347,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965755369422458e-05,
+      "loss": 0.4219,
+      "step": 1276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656821885568046e-05,
+      "loss": 0.4498,
+      "step": 1277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656089309458567e-05,
+      "loss": 0.4162,
+      "step": 1278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655355965954362e-05,
+      "loss": 0.4325,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654621855113713e-05,
+      "loss": 0.3811,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965388697699496e-05,
+      "loss": 0.4364,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965315133165651e-05,
+      "loss": 0.3895,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652414919156824e-05,
+      "loss": 0.4449,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965167773955442e-05,
+      "loss": 0.4528,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96509397929079e-05,
+      "loss": 0.4275,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650201079275894e-05,
+      "loss": 0.4075,
+      "step": 1286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649461598717116e-05,
+      "loss": 0.386,
+      "step": 1287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648721351290338e-05,
+      "loss": 0.4185,
+      "step": 1288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647980337054384e-05,
+      "loss": 0.4173,
+      "step": 1289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647238556068146e-05,
+      "loss": 0.4225,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646496008390574e-05,
+      "loss": 0.3958,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645752694080684e-05,
+      "loss": 0.4623,
+      "step": 1292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645008613197546e-05,
+      "loss": 0.4203,
+      "step": 1293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644263765800295e-05,
+      "loss": 0.4202,
+      "step": 1294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643518151948128e-05,
+      "loss": 0.4174,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642771771700295e-05,
+      "loss": 0.4405,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642024625116117e-05,
+      "loss": 0.3996,
+      "step": 1297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641276712254974e-05,
+      "loss": 0.4384,
+      "step": 1298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96405280331763e-05,
+      "loss": 0.4314,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96397785879396e-05,
+      "loss": 0.4234,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639028376604424e-05,
+      "loss": 0.409,
+      "step": 1301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638277399230404e-05,
+      "loss": 0.414,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963752565587722e-05,
+      "loss": 0.4409,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963677314660461e-05,
+      "loss": 0.3806,
+      "step": 1304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636019871472384e-05,
+      "loss": 0.4365,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635265830540402e-05,
+      "loss": 0.4095,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634511023868592e-05,
+      "loss": 0.4341,
+      "step": 1307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963375545151694e-05,
+      "loss": 0.4211,
+      "step": 1308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963299911354549e-05,
+      "loss": 0.436,
+      "step": 1309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963224201001436e-05,
+      "loss": 0.4036,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631484140983704e-05,
+      "loss": 0.4446,
+      "step": 1311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630725506513766e-05,
+      "loss": 0.388,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962996610666483e-05,
+      "loss": 0.4319,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962920594149725e-05,
+      "loss": 0.3984,
+      "step": 1314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628445011071432e-05,
+      "loss": 0.3963,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627683315447855e-05,
+      "loss": 0.4419,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626920854687054e-05,
+      "loss": 0.4238,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962615762884962e-05,
+      "loss": 0.4513,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962539363799621e-05,
+      "loss": 0.4183,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962462888218754e-05,
+      "loss": 0.4158,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962386336148439e-05,
+      "loss": 0.3912,
+      "step": 1321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623097075947593e-05,
+      "loss": 0.4282,
+      "step": 1322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622330025638053e-05,
+      "loss": 0.4109,
+      "step": 1323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621562210616726e-05,
+      "loss": 0.4553,
+      "step": 1324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620793630944632e-05,
+      "loss": 0.443,
+      "step": 1325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620024286682852e-05,
+      "loss": 0.3927,
+      "step": 1326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961925417789253e-05,
+      "loss": 0.3897,
+      "step": 1327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618483304634866e-05,
+      "loss": 0.399,
+      "step": 1328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617711666971127e-05,
+      "loss": 0.4085,
+      "step": 1329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616939264962635e-05,
+      "loss": 0.4191,
+      "step": 1330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616166098670774e-05,
+      "loss": 0.4326,
+      "step": 1331
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961539216815699e-05,
+      "loss": 0.4455,
+      "step": 1332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614617473482788e-05,
+      "loss": 0.4425,
+      "step": 1333
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613842014709737e-05,
+      "loss": 0.4546,
+      "step": 1334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613065791899462e-05,
+      "loss": 0.4268,
+      "step": 1335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961228880511366e-05,
+      "loss": 0.4176,
+      "step": 1336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961151105441407e-05,
+      "loss": 0.3977,
+      "step": 1337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610732539862506e-05,
+      "loss": 0.4125,
+      "step": 1338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609953261520838e-05,
+      "loss": 0.382,
+      "step": 1339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609173219450998e-05,
+      "loss": 0.4086,
+      "step": 1340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608392413714976e-05,
+      "loss": 0.4102,
+      "step": 1341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607610844374828e-05,
+      "loss": 0.4208,
+      "step": 1342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606828511492666e-05,
+      "loss": 0.4428,
+      "step": 1343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606045415130663e-05,
+      "loss": 0.4079,
+      "step": 1344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605261555351057e-05,
+      "loss": 0.4093,
+      "step": 1345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604476932216136e-05,
+      "loss": 0.4044,
+      "step": 1346
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603691545788263e-05,
+      "loss": 0.4234,
+      "step": 1347
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960290539612985e-05,
+      "loss": 0.4084,
+      "step": 1348
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602118483303382e-05,
+      "loss": 0.411,
+      "step": 1349
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960133080737139e-05,
+      "loss": 0.4016,
+      "step": 1350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600542368396477e-05,
+      "loss": 0.4243,
+      "step": 1351
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599753166441296e-05,
+      "loss": 0.4345,
+      "step": 1352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959896320156857e-05,
+      "loss": 0.4014,
+      "step": 1353
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598172473841084e-05,
+      "loss": 0.398,
+      "step": 1354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597380983321675e-05,
+      "loss": 0.4071,
+      "step": 1355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596588730073244e-05,
+      "loss": 0.4013,
+      "step": 1356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595795714158756e-05,
+      "loss": 0.454,
+      "step": 1357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595001935641234e-05,
+      "loss": 0.4028,
+      "step": 1358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594207394583758e-05,
+      "loss": 0.4414,
+      "step": 1359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593412091049476e-05,
+      "loss": 0.4112,
+      "step": 1360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592616025101594e-05,
+      "loss": 0.4147,
+      "step": 1361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591819196803375e-05,
+      "loss": 0.4277,
+      "step": 1362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591021606218143e-05,
+      "loss": 0.441,
+      "step": 1363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590223253409287e-05,
+      "loss": 0.4088,
+      "step": 1364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589424138440252e-05,
+      "loss": 0.4259,
+      "step": 1365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588624261374553e-05,
+      "loss": 0.4223,
+      "step": 1366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958782362227575e-05,
+      "loss": 0.4317,
+      "step": 1367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587022221207476e-05,
+      "loss": 0.4498,
+      "step": 1368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958622005823342e-05,
+      "loss": 0.4107,
+      "step": 1369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958541713341733e-05,
+      "loss": 0.4628,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958461344682302e-05,
+      "loss": 0.4341,
+      "step": 1371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583808998514357e-05,
+      "loss": 0.4528,
+      "step": 1372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583003788555273e-05,
+      "loss": 0.3807,
+      "step": 1373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958219781700977e-05,
+      "loss": 0.4111,
+      "step": 1374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581391083941884e-05,
+      "loss": 0.3658,
+      "step": 1375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958058358941574e-05,
+      "loss": 0.4377,
+      "step": 1376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957977533349551e-05,
+      "loss": 0.3958,
+      "step": 1377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578966316245423e-05,
+      "loss": 0.4063,
+      "step": 1378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578156537729778e-05,
+      "loss": 0.4023,
+      "step": 1379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957734599801293e-05,
+      "loss": 0.4175,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576534697159298e-05,
+      "loss": 0.4625,
+      "step": 1381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957572263523335e-05,
+      "loss": 0.4321,
+      "step": 1382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957490981229963e-05,
+      "loss": 0.4478,
+      "step": 1383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957409622842273e-05,
+      "loss": 0.3584,
+      "step": 1384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957328188366731e-05,
+      "loss": 0.4168,
+      "step": 1385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572466778098092e-05,
+      "loss": 0.4152,
+      "step": 1386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957165091177985e-05,
+      "loss": 0.443,
+      "step": 1387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957083428477742e-05,
+      "loss": 0.4251,
+      "step": 1388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570016897155706e-05,
+      "loss": 0.4029,
+      "step": 1389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569198748979668e-05,
+      "loss": 0.4335,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568379840314326e-05,
+      "loss": 0.4563,
+      "step": 1391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956756017122476e-05,
+      "loss": 0.419,
+      "step": 1392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956673974177611e-05,
+      "loss": 0.4161,
+      "step": 1393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565918552033582e-05,
+      "loss": 0.41,
+      "step": 1394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565096602062436e-05,
+      "loss": 0.4075,
+      "step": 1395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564273891927992e-05,
+      "loss": 0.4264,
+      "step": 1396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563450421695634e-05,
+      "loss": 0.3815,
+      "step": 1397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562626191430807e-05,
+      "loss": 0.395,
+      "step": 1398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561801201199015e-05,
+      "loss": 0.4051,
+      "step": 1399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956097545106582e-05,
+      "loss": 0.4127,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560148941096848e-05,
+      "loss": 0.4043,
+      "step": 1401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955932167135778e-05,
+      "loss": 0.4235,
+      "step": 1402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558493641914368e-05,
+      "loss": 0.4348,
+      "step": 1403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557664852832415e-05,
+      "loss": 0.4387,
+      "step": 1404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955683530417779e-05,
+      "loss": 0.4402,
+      "step": 1405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556004996016408e-05,
+      "loss": 0.4204,
+      "step": 1406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955517392841427e-05,
+      "loss": 0.467,
+      "step": 1407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554342101437415e-05,
+      "loss": 0.4146,
+      "step": 1408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955350951515195e-05,
+      "loss": 0.3678,
+      "step": 1409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552676169624044e-05,
+      "loss": 0.4099,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551842064919932e-05,
+      "loss": 0.4049,
+      "step": 1411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551007201105895e-05,
+      "loss": 0.4323,
+      "step": 1412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955017157824828e-05,
+      "loss": 0.4322,
+      "step": 1413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95493351964135e-05,
+      "loss": 0.3823,
+      "step": 1414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548498055668027e-05,
+      "loss": 0.3982,
+      "step": 1415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547660156078387e-05,
+      "loss": 0.4746,
+      "step": 1416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954682149771117e-05,
+      "loss": 0.4537,
+      "step": 1417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954598208063303e-05,
+      "loss": 0.4285,
+      "step": 1418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954514190491067e-05,
+      "loss": 0.3865,
+      "step": 1419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544300970610872e-05,
+      "loss": 0.4891,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954345927780046e-05,
+      "loss": 0.4548,
+      "step": 1421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542616826546322e-05,
+      "loss": 0.4084,
+      "step": 1422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954177361691542e-05,
+      "loss": 0.4432,
+      "step": 1423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540929648974757e-05,
+      "loss": 0.4422,
+      "step": 1424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954008492279141e-05,
+      "loss": 0.4069,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539239438432515e-05,
+      "loss": 0.4422,
+      "step": 1426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538393195965253e-05,
+      "loss": 0.4243,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537546195456888e-05,
+      "loss": 0.4283,
+      "step": 1428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953669843697473e-05,
+      "loss": 0.4012,
+      "step": 1429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535849920586152e-05,
+      "loss": 0.3924,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953500064635859e-05,
+      "loss": 0.427,
+      "step": 1431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534150614359536e-05,
+      "loss": 0.3901,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533299824656545e-05,
+      "loss": 0.4173,
+      "step": 1433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953244827731723e-05,
+      "loss": 0.4049,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531595972409264e-05,
+      "loss": 0.4195,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9530742910000388e-05,
+      "loss": 0.4041,
+      "step": 1436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529889090158394e-05,
+      "loss": 0.3986,
+      "step": 1437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529034512951137e-05,
+      "loss": 0.4232,
+      "step": 1438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528179178446535e-05,
+      "loss": 0.405,
+      "step": 1439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527323086712557e-05,
+      "loss": 0.3778,
+      "step": 1440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526466237817243e-05,
+      "loss": 0.4307,
+      "step": 1441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525608631828693e-05,
+      "loss": 0.4248,
+      "step": 1442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524750268815057e-05,
+      "loss": 0.438,
+      "step": 1443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523891148844553e-05,
+      "loss": 0.4135,
+      "step": 1444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523031271985458e-05,
+      "loss": 0.3846,
+      "step": 1445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952217063830611e-05,
+      "loss": 0.37,
+      "step": 1446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521309247874905e-05,
+      "loss": 0.4535,
+      "step": 1447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95204471007603e-05,
+      "loss": 0.4041,
+      "step": 1448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519584197030812e-05,
+      "loss": 0.4101,
+      "step": 1449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951872053675502e-05,
+      "loss": 0.4034,
+      "step": 1450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517856120001558e-05,
+      "loss": 0.3945,
+      "step": 1451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951699094683912e-05,
+      "loss": 0.4188,
+      "step": 1452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516125017336473e-05,
+      "loss": 0.4056,
+      "step": 1453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951525833156243e-05,
+      "loss": 0.4153,
+      "step": 1454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514390889585867e-05,
+      "loss": 0.3901,
+      "step": 1455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513522691475725e-05,
+      "loss": 0.3926,
+      "step": 1456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512653737301003e-05,
+      "loss": 0.3924,
+      "step": 1457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511784027130755e-05,
+      "loss": 0.4011,
+      "step": 1458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95109135610341e-05,
+      "loss": 0.4112,
+      "step": 1459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510042339080217e-05,
+      "loss": 0.4087,
+      "step": 1460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509170361338344e-05,
+      "loss": 0.4438,
+      "step": 1461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950829762787778e-05,
+      "loss": 0.4072,
+      "step": 1462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507424138767884e-05,
+      "loss": 0.4096,
+      "step": 1463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506549894078076e-05,
+      "loss": 0.4424,
+      "step": 1464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950567489387783e-05,
+      "loss": 0.4437,
+      "step": 1465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950479913823669e-05,
+      "loss": 0.4494,
+      "step": 1466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503922627224247e-05,
+      "loss": 0.3931,
+      "step": 1467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503045360910164e-05,
+      "loss": 0.3947,
+      "step": 1468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950216733936416e-05,
+      "loss": 0.4063,
+      "step": 1469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501288562656014e-05,
+      "loss": 0.4353,
+      "step": 1470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500409030855566e-05,
+      "loss": 0.4233,
+      "step": 1471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949952874403271e-05,
+      "loss": 0.43,
+      "step": 1472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949864770225741e-05,
+      "loss": 0.3823,
+      "step": 1473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949776590559968e-05,
+      "loss": 0.4366,
+      "step": 1474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496883354129605e-05,
+      "loss": 0.4278,
+      "step": 1475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496000047917316e-05,
+      "loss": 0.4354,
+      "step": 1476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949511598703301e-05,
+      "loss": 0.4193,
+      "step": 1477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949423117154696e-05,
+      "loss": 0.435,
+      "step": 1478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949334560152947e-05,
+      "loss": 0.4462,
+      "step": 1479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492459277050922e-05,
+      "loss": 0.4012,
+      "step": 1480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491572198181764e-05,
+      "loss": 0.4012,
+      "step": 1481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949068436499248e-05,
+      "loss": 0.397,
+      "step": 1482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948979577755364e-05,
+      "loss": 0.3882,
+      "step": 1483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488906435935856e-05,
+      "loss": 0.419,
+      "step": 1484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948801634020981e-05,
+      "loss": 0.4054,
+      "step": 1485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487125490446236e-05,
+      "loss": 0.3787,
+      "step": 1486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486233886715933e-05,
+      "loss": 0.4057,
+      "step": 1487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485341529089762e-05,
+      "loss": 0.3731,
+      "step": 1488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948444841763864e-05,
+      "loss": 0.4238,
+      "step": 1489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483554552433545e-05,
+      "loss": 0.3696,
+      "step": 1490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482659933545515e-05,
+      "loss": 0.4027,
+      "step": 1491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481764561045648e-05,
+      "loss": 0.4463,
+      "step": 1492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480868435005098e-05,
+      "loss": 0.4252,
+      "step": 1493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479971555495087e-05,
+      "loss": 0.4443,
+      "step": 1494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479073922586888e-05,
+      "loss": 0.4174,
+      "step": 1495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947817553635184e-05,
+      "loss": 0.4049,
+      "step": 1496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477276396861345e-05,
+      "loss": 0.4219,
+      "step": 1497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947637650418685e-05,
+      "loss": 0.4165,
+      "step": 1498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475475858399884e-05,
+      "loss": 0.3945,
+      "step": 1499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474574459572014e-05,
+      "loss": 0.3571,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947367230777488e-05,
+      "loss": 0.418,
+      "step": 1501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472769403080177e-05,
+      "loss": 0.4462,
+      "step": 1502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471865745559664e-05,
+      "loss": 0.4067,
+      "step": 1503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947096133528516e-05,
+      "loss": 0.4094,
+      "step": 1504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947005617232853e-05,
+      "loss": 0.4054,
+      "step": 1505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469150256761714e-05,
+      "loss": 0.4096,
+      "step": 1506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468243588656716e-05,
+      "loss": 0.3975,
+      "step": 1507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467336168085578e-05,
+      "loss": 0.4212,
+      "step": 1508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466427995120427e-05,
+      "loss": 0.4183,
+      "step": 1509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465519069833432e-05,
+      "loss": 0.3832,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464609392296826e-05,
+      "loss": 0.453,
+      "step": 1511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463698962582908e-05,
+      "loss": 0.4332,
+      "step": 1512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946278778076403e-05,
+      "loss": 0.4467,
+      "step": 1513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461875846912607e-05,
+      "loss": 0.4416,
+      "step": 1514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460963161101107e-05,
+      "loss": 0.4597,
+      "step": 1515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460049723402072e-05,
+      "loss": 0.4482,
+      "step": 1516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459135533888093e-05,
+      "loss": 0.4341,
+      "step": 1517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458220592631817e-05,
+      "loss": 0.4198,
+      "step": 1518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457304899705967e-05,
+      "loss": 0.3971,
+      "step": 1519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456388455183308e-05,
+      "loss": 0.4304,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945547125913667e-05,
+      "loss": 0.4363,
+      "step": 1521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454553311638956e-05,
+      "loss": 0.4382,
+      "step": 1522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453634612763105e-05,
+      "loss": 0.4538,
+      "step": 1523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452715162582142e-05,
+      "loss": 0.4245,
+      "step": 1524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451794961169125e-05,
+      "loss": 0.4246,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945087400859719e-05,
+      "loss": 0.4163,
+      "step": 1526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944995230493953e-05,
+      "loss": 0.4037,
+      "step": 1527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944902985026939e-05,
+      "loss": 0.3872,
+      "step": 1528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448106644660084e-05,
+      "loss": 0.4523,
+      "step": 1529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447182688184982e-05,
+      "loss": 0.4145,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944625798091751e-05,
+      "loss": 0.4046,
+      "step": 1531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944533252293116e-05,
+      "loss": 0.4415,
+      "step": 1532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944440631429948e-05,
+      "loss": 0.3981,
+      "step": 1533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443479355096073e-05,
+      "loss": 0.425,
+      "step": 1534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442551645394615e-05,
+      "loss": 0.3849,
+      "step": 1535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441623185268826e-05,
+      "loss": 0.4056,
+      "step": 1536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440693974792505e-05,
+      "loss": 0.4118,
+      "step": 1537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439764014039484e-05,
+      "loss": 0.4651,
+      "step": 1538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438833303083677e-05,
+      "loss": 0.4079,
+      "step": 1539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437901841999053e-05,
+      "loss": 0.3979,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436969630859628e-05,
+      "loss": 0.4565,
+      "step": 1541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436036669739495e-05,
+      "loss": 0.3807,
+      "step": 1542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94351029587128e-05,
+      "loss": 0.368,
+      "step": 1543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434168497853742e-05,
+      "loss": 0.3937,
+      "step": 1544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433233287236586e-05,
+      "loss": 0.4193,
+      "step": 1545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943229732693566e-05,
+      "loss": 0.4061,
+      "step": 1546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431360617025344e-05,
+      "loss": 0.4796,
+      "step": 1547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943042315758008e-05,
+      "loss": 0.4059,
+      "step": 1548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429484948674374e-05,
+      "loss": 0.3985,
+      "step": 1549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428545990382782e-05,
+      "loss": 0.403,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942760628277993e-05,
+      "loss": 0.4091,
+      "step": 1551
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426665825940498e-05,
+      "loss": 0.373,
+      "step": 1552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425724619939225e-05,
+      "loss": 0.422,
+      "step": 1553
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424782664850917e-05,
+      "loss": 0.4324,
+      "step": 1554
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423839960750426e-05,
+      "loss": 0.3987,
+      "step": 1555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422896507712672e-05,
+      "loss": 0.4096,
+      "step": 1556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421952305812643e-05,
+      "loss": 0.4234,
+      "step": 1557
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421007355125368e-05,
+      "loss": 0.4406,
+      "step": 1558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420061655725942e-05,
+      "loss": 0.4163,
+      "step": 1559
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941911520768953e-05,
+      "loss": 0.3975,
+      "step": 1560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941816801109135e-05,
+      "loss": 0.4238,
+      "step": 1561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417220066006668e-05,
+      "loss": 0.4064,
+      "step": 1562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941627137251083e-05,
+      "loss": 0.4068,
+      "step": 1563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415321930679224e-05,
+      "loss": 0.4245,
+      "step": 1564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941437174058731e-05,
+      "loss": 0.4208,
+      "step": 1565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413420802310598e-05,
+      "loss": 0.4336,
+      "step": 1566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412469115924664e-05,
+      "loss": 0.4022,
+      "step": 1567
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941151668150514e-05,
+      "loss": 0.4425,
+      "step": 1568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941056349912772e-05,
+      "loss": 0.4512,
+      "step": 1569
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409609568868153e-05,
+      "loss": 0.3972,
+      "step": 1570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408654890802254e-05,
+      "loss": 0.3937,
+      "step": 1571
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940769946500589e-05,
+      "loss": 0.4312,
+      "step": 1572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940674329155499e-05,
+      "loss": 0.3903,
+      "step": 1573
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405786370525552e-05,
+      "loss": 0.4001,
+      "step": 1574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404828701993616e-05,
+      "loss": 0.406,
+      "step": 1575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403870286035295e-05,
+      "loss": 0.3787,
+      "step": 1576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402911122726756e-05,
+      "loss": 0.3766,
+      "step": 1577
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401951212144227e-05,
+      "loss": 0.4222,
+      "step": 1578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940099055436399e-05,
+      "loss": 0.4356,
+      "step": 1579
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94000291494624e-05,
+      "loss": 0.4427,
+      "step": 1580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399066997515854e-05,
+      "loss": 0.4165,
+      "step": 1581
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939810409860082e-05,
+      "loss": 0.391,
+      "step": 1582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397140452793822e-05,
+      "loss": 0.3812,
+      "step": 1583
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396176060171442e-05,
+      "loss": 0.373,
+      "step": 1584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395210920810324e-05,
+      "loss": 0.3846,
+      "step": 1585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394245034787173e-05,
+      "loss": 0.3682,
+      "step": 1586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393278402178747e-05,
+      "loss": 0.4569,
+      "step": 1587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939231102306187e-05,
+      "loss": 0.3905,
+      "step": 1588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391342897513414e-05,
+      "loss": 0.4566,
+      "step": 1589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390374025610325e-05,
+      "loss": 0.4276,
+      "step": 1590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389404407429602e-05,
+      "loss": 0.4316,
+      "step": 1591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388434043048303e-05,
+      "loss": 0.4048,
+      "step": 1592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387462932543545e-05,
+      "loss": 0.413,
+      "step": 1593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386491075992503e-05,
+      "loss": 0.405,
+      "step": 1594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385518473472412e-05,
+      "loss": 0.3788,
+      "step": 1595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384545125060573e-05,
+      "loss": 0.3655,
+      "step": 1596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938357103083434e-05,
+      "loss": 0.418,
+      "step": 1597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382596190871118e-05,
+      "loss": 0.4371,
+      "step": 1598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381620605248387e-05,
+      "loss": 0.4138,
+      "step": 1599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938064427404368e-05,
+      "loss": 0.4144,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379667197334587e-05,
+      "loss": 0.4477,
+      "step": 1601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378689375198754e-05,
+      "loss": 0.432,
+      "step": 1602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93777108077139e-05,
+      "loss": 0.384,
+      "step": 1603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937673149495779e-05,
+      "loss": 0.4095,
+      "step": 1604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375751437008253e-05,
+      "loss": 0.4195,
+      "step": 1605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374770633943175e-05,
+      "loss": 0.4045,
+      "step": 1606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373789085840505e-05,
+      "loss": 0.4023,
+      "step": 1607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372806792778246e-05,
+      "loss": 0.3856,
+      "step": 1608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937182375483447e-05,
+      "loss": 0.4015,
+      "step": 1609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370839972087297e-05,
+      "loss": 0.3979,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936985544461491e-05,
+      "loss": 0.4182,
+      "step": 1611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368870172495554e-05,
+      "loss": 0.4083,
+      "step": 1612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936788415580753e-05,
+      "loss": 0.4359,
+      "step": 1613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366897394629202e-05,
+      "loss": 0.3836,
+      "step": 1614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365909889038987e-05,
+      "loss": 0.3904,
+      "step": 1615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364921639115363e-05,
+      "loss": 0.391,
+      "step": 1616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363932644936877e-05,
+      "loss": 0.4165,
+      "step": 1617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362942906582117e-05,
+      "loss": 0.4118,
+      "step": 1618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936195242412975e-05,
+      "loss": 0.3769,
+      "step": 1619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360961197658482e-05,
+      "loss": 0.4176,
+      "step": 1620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359969227247098e-05,
+      "loss": 0.4224,
+      "step": 1621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358976512974425e-05,
+      "loss": 0.4042,
+      "step": 1622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935798305491936e-05,
+      "loss": 0.393,
+      "step": 1623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935698885316086e-05,
+      "loss": 0.4001,
+      "step": 1624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355993907777925e-05,
+      "loss": 0.3893,
+      "step": 1625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354998218849635e-05,
+      "loss": 0.4544,
+      "step": 1626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935400178645512e-05,
+      "loss": 0.3996,
+      "step": 1627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353004610673567e-05,
+      "loss": 0.4629,
+      "step": 1628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352006691584224e-05,
+      "loss": 0.4219,
+      "step": 1629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351008029266398e-05,
+      "loss": 0.4301,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350008623799455e-05,
+      "loss": 0.3792,
+      "step": 1631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349008475262823e-05,
+      "loss": 0.3953,
+      "step": 1632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348007583735985e-05,
+      "loss": 0.4285,
+      "step": 1633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347005949298484e-05,
+      "loss": 0.4331,
+      "step": 1634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934600357202992e-05,
+      "loss": 0.3946,
+      "step": 1635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934500045200996e-05,
+      "loss": 0.4248,
+      "step": 1636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934399658931832e-05,
+      "loss": 0.4529,
+      "step": 1637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342991984034783e-05,
+      "loss": 0.4317,
+      "step": 1638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341986636239187e-05,
+      "loss": 0.4207,
+      "step": 1639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340980546011426e-05,
+      "loss": 0.3723,
+      "step": 1640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339973713431458e-05,
+      "loss": 0.4465,
+      "step": 1641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338966138579306e-05,
+      "loss": 0.4098,
+      "step": 1642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337957821535036e-05,
+      "loss": 0.4657,
+      "step": 1643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933694876237878e-05,
+      "loss": 0.4319,
+      "step": 1644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335938961190738e-05,
+      "loss": 0.4253,
+      "step": 1645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334928418051156e-05,
+      "loss": 0.3792,
+      "step": 1646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933391713304035e-05,
+      "loss": 0.4231,
+      "step": 1647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332905106238683e-05,
+      "loss": 0.423,
+      "step": 1648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933189233772659e-05,
+      "loss": 0.4262,
+      "step": 1649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330878827584554e-05,
+      "loss": 0.4014,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932986457589312e-05,
+      "loss": 0.3893,
+      "step": 1651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93288495827329e-05,
+      "loss": 0.4205,
+      "step": 1652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327833848184548e-05,
+      "loss": 0.3878,
+      "step": 1653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326817372328798e-05,
+      "loss": 0.4321,
+      "step": 1654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325800155246425e-05,
+      "loss": 0.4224,
+      "step": 1655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324782197018267e-05,
+      "loss": 0.3976,
+      "step": 1656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323763497725235e-05,
+      "loss": 0.4131,
+      "step": 1657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322744057448276e-05,
+      "loss": 0.392,
+      "step": 1658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932172387626842e-05,
+      "loss": 0.3993,
+      "step": 1659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932070295426673e-05,
+      "loss": 0.4042,
+      "step": 1660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931968129152435e-05,
+      "loss": 0.379,
+      "step": 1661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318658888122472e-05,
+      "loss": 0.3828,
+      "step": 1662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317635744142346e-05,
+      "loss": 0.4205,
+      "step": 1663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931661185966529e-05,
+      "loss": 0.3868,
+      "step": 1664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315587234772674e-05,
+      "loss": 0.4109,
+      "step": 1665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931456186954592e-05,
+      "loss": 0.4031,
+      "step": 1666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313535764066524e-05,
+      "loss": 0.397,
+      "step": 1667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931250891841603e-05,
+      "loss": 0.4067,
+      "step": 1668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311481332676047e-05,
+      "loss": 0.3915,
+      "step": 1669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310453006928237e-05,
+      "loss": 0.3687,
+      "step": 1670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309423941254326e-05,
+      "loss": 0.4046,
+      "step": 1671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308394135736095e-05,
+      "loss": 0.4053,
+      "step": 1672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307363590455385e-05,
+      "loss": 0.3605,
+      "step": 1673
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306332305494097e-05,
+      "loss": 0.4769,
+      "step": 1674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305300280934187e-05,
+      "loss": 0.4217,
+      "step": 1675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930426751685768e-05,
+      "loss": 0.3754,
+      "step": 1676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303234013346642e-05,
+      "loss": 0.4122,
+      "step": 1677
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302199770483222e-05,
+      "loss": 0.3969,
+      "step": 1678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93011647883496e-05,
+      "loss": 0.3843,
+      "step": 1679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300129067028036e-05,
+      "loss": 0.4301,
+      "step": 1680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299092606600844e-05,
+      "loss": 0.4285,
+      "step": 1681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298055407150385e-05,
+      "loss": 0.4204,
+      "step": 1682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297017468759097e-05,
+      "loss": 0.406,
+      "step": 1683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295978791509466e-05,
+      "loss": 0.3916,
+      "step": 1684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294939375484034e-05,
+      "loss": 0.3965,
+      "step": 1685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929389922076541e-05,
+      "loss": 0.4003,
+      "step": 1686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292858327436254e-05,
+      "loss": 0.385,
+      "step": 1687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291816695579297e-05,
+      "loss": 0.4018,
+      "step": 1688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290774325277305e-05,
+      "loss": 0.4102,
+      "step": 1689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289731216613136e-05,
+      "loss": 0.4189,
+      "step": 1690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288687369669673e-05,
+      "loss": 0.4644,
+      "step": 1691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287642784529884e-05,
+      "loss": 0.4205,
+      "step": 1692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928659746127678e-05,
+      "loss": 0.4194,
+      "step": 1693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285551399993437e-05,
+      "loss": 0.4013,
+      "step": 1694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928450460076299e-05,
+      "loss": 0.4149,
+      "step": 1695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283457063668626e-05,
+      "loss": 0.4101,
+      "step": 1696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92824087887936e-05,
+      "loss": 0.4154,
+      "step": 1697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281359776221212e-05,
+      "loss": 0.424,
+      "step": 1698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280310026034845e-05,
+      "loss": 0.4745,
+      "step": 1699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279259538317917e-05,
+      "loss": 0.4101,
+      "step": 1700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927820831315391e-05,
+      "loss": 0.4117,
+      "step": 1701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277156350626373e-05,
+      "loss": 0.437,
+      "step": 1702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927610365081891e-05,
+      "loss": 0.4321,
+      "step": 1703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275050213815173e-05,
+      "loss": 0.433,
+      "step": 1704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273996039698887e-05,
+      "loss": 0.4093,
+      "step": 1705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927294112855383e-05,
+      "loss": 0.4132,
+      "step": 1706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927188548046384e-05,
+      "loss": 0.3814,
+      "step": 1707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270829095512805e-05,
+      "loss": 0.3795,
+      "step": 1708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926977197378469e-05,
+      "loss": 0.3652,
+      "step": 1709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268714115363498e-05,
+      "loss": 0.4407,
+      "step": 1710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267655520333304e-05,
+      "loss": 0.4518,
+      "step": 1711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266596188778238e-05,
+      "loss": 0.3984,
+      "step": 1712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265536120782486e-05,
+      "loss": 0.4181,
+      "step": 1713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926447531643029e-05,
+      "loss": 0.3885,
+      "step": 1714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263413775805962e-05,
+      "loss": 0.4294,
+      "step": 1715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926235149899386e-05,
+      "loss": 0.4207,
+      "step": 1716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261288486078414e-05,
+      "loss": 0.4076,
+      "step": 1717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260224737144098e-05,
+      "loss": 0.3862,
+      "step": 1718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925916025227545e-05,
+      "loss": 0.4205,
+      "step": 1719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925809503155707e-05,
+      "loss": 0.4227,
+      "step": 1720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257029075073612e-05,
+      "loss": 0.3838,
+      "step": 1721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255962382909795e-05,
+      "loss": 0.4161,
+      "step": 1722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254894955150385e-05,
+      "loss": 0.3969,
+      "step": 1723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925382679188022e-05,
+      "loss": 0.3993,
+      "step": 1724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252757893184183e-05,
+      "loss": 0.3826,
+      "step": 1725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251688259147228e-05,
+      "loss": 0.4102,
+      "step": 1726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925061788985436e-05,
+      "loss": 0.4193,
+      "step": 1727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924954678539064e-05,
+      "loss": 0.4287,
+      "step": 1728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248474945841197e-05,
+      "loss": 0.3926,
+      "step": 1729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247402371291214e-05,
+      "loss": 0.3949,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246329061825926e-05,
+      "loss": 0.4245,
+      "step": 1731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924525501753063e-05,
+      "loss": 0.4179,
+      "step": 1732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924418023849069e-05,
+      "loss": 0.3826,
+      "step": 1733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924310472479152e-05,
+      "loss": 0.4042,
+      "step": 1734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242028476518588e-05,
+      "loss": 0.4041,
+      "step": 1735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240951493757433e-05,
+      "loss": 0.4309,
+      "step": 1736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239873776593645e-05,
+      "loss": 0.4394,
+      "step": 1737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.3693,
+      "step": 1738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237716139400814e-05,
+      "loss": 0.4202,
+      "step": 1739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923663621954325e-05,
+      "loss": 0.3854,
+      "step": 1740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235555565625993e-05,
+      "loss": 0.3543,
+      "step": 1741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923447417773493e-05,
+      "loss": 0.4351,
+      "step": 1742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233392055956005e-05,
+      "loss": 0.3974,
+      "step": 1743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232309200375212e-05,
+      "loss": 0.4294,
+      "step": 1744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923122561107861e-05,
+      "loss": 0.4241,
+      "step": 1745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230141288152314e-05,
+      "loss": 0.3968,
+      "step": 1746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229056231682496e-05,
+      "loss": 0.3986,
+      "step": 1747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227970441755394e-05,
+      "loss": 0.4441,
+      "step": 1748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226883918457297e-05,
+      "loss": 0.4403,
+      "step": 1749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922579666187455e-05,
+      "loss": 0.3997,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224708672093563e-05,
+      "loss": 0.4537,
+      "step": 1751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92236199492008e-05,
+      "loss": 0.3782,
+      "step": 1752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222530493282786e-05,
+      "loss": 0.3994,
+      "step": 1753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221440304426102e-05,
+      "loss": 0.3878,
+      "step": 1754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922034938271739e-05,
+      "loss": 0.3856,
+      "step": 1755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219257728243346e-05,
+      "loss": 0.3809,
+      "step": 1756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218165341090728e-05,
+      "loss": 0.4027,
+      "step": 1757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217072221346348e-05,
+      "loss": 0.4177,
+      "step": 1758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215978369097087e-05,
+      "loss": 0.3726,
+      "step": 1759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214883784429865e-05,
+      "loss": 0.401,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921378846743168e-05,
+      "loss": 0.422,
+      "step": 1761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212692418189576e-05,
+      "loss": 0.411,
+      "step": 1762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921159563679066e-05,
+      "loss": 0.3865,
+      "step": 1763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210498123322098e-05,
+      "loss": 0.435,
+      "step": 1764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920939987787111e-05,
+      "loss": 0.4192,
+      "step": 1765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208300900524976e-05,
+      "loss": 0.4402,
+      "step": 1766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207201191371035e-05,
+      "loss": 0.3865,
+      "step": 1767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920610075049669e-05,
+      "loss": 0.38,
+      "step": 1768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920499957798938e-05,
+      "loss": 0.4036,
+      "step": 1769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203897673936634e-05,
+      "loss": 0.4382,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202795038426017e-05,
+      "loss": 0.407,
+      "step": 1771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201691671545157e-05,
+      "loss": 0.3839,
+      "step": 1772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200587573381747e-05,
+      "loss": 0.3986,
+      "step": 1773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199482744023528e-05,
+      "loss": 0.3791,
+      "step": 1774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.91983771835583e-05,
+      "loss": 0.4023,
+      "step": 1775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197270892073932e-05,
+      "loss": 0.403,
+      "step": 1776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196163869658342e-05,
+      "loss": 0.3847,
+      "step": 1777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195056116399506e-05,
+      "loss": 0.4206,
+      "step": 1778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193947632385462e-05,
+      "loss": 0.4193,
+      "step": 1779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9192838417704304e-05,
+      "loss": 0.41,
+      "step": 1780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919172847244418e-05,
+      "loss": 0.4148,
+      "step": 1781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190617796693307e-05,
+      "loss": 0.4112,
+      "step": 1782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918950639053995e-05,
+      "loss": 0.4017,
+      "step": 1783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918839425407243e-05,
+      "loss": 0.4085,
+      "step": 1784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187281387379144e-05,
+      "loss": 0.3859,
+      "step": 1785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186167790548524e-05,
+      "loss": 0.417,
+      "step": 1786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185053463669073e-05,
+      "loss": 0.3966,
+      "step": 1787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183938406829354e-05,
+      "loss": 0.4361,
+      "step": 1788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182822620117973e-05,
+      "loss": 0.3719,
+      "step": 1789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181706103623615e-05,
+      "loss": 0.4037,
+      "step": 1790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918058885743501e-05,
+      "loss": 0.4084,
+      "step": 1791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179470881640948e-05,
+      "loss": 0.3848,
+      "step": 1792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178352176330272e-05,
+      "loss": 0.4113,
+      "step": 1793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177232741591895e-05,
+      "loss": 0.4059,
+      "step": 1794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176112577514782e-05,
+      "loss": 0.3704,
+      "step": 1795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917499168418795e-05,
+      "loss": 0.382,
+      "step": 1796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173870061700483e-05,
+      "loss": 0.3834,
+      "step": 1797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917274771014152e-05,
+      "loss": 0.3895,
+      "step": 1798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917162462960025e-05,
+      "loss": 0.4112,
+      "step": 1799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917050082016594e-05,
+      "loss": 0.3972,
+      "step": 1800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916937628192789e-05,
+      "loss": 0.4544,
+      "step": 1801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168251014975475e-05,
+      "loss": 0.3896,
+      "step": 1802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167125019398125e-05,
+      "loss": 0.4227,
+      "step": 1803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165998295285325e-05,
+      "loss": 0.4294,
+      "step": 1804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164870842726613e-05,
+      "loss": 0.3784,
+      "step": 1805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163742661811598e-05,
+      "loss": 0.3898,
+      "step": 1806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162613752629933e-05,
+      "loss": 0.3792,
+      "step": 1807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161484115271344e-05,
+      "loss": 0.3858,
+      "step": 1808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160353749825598e-05,
+      "loss": 0.431,
+      "step": 1809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915922265638253e-05,
+      "loss": 0.4035,
+      "step": 1810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158090835032033e-05,
+      "loss": 0.3754,
+      "step": 1811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156958285864057e-05,
+      "loss": 0.3986,
+      "step": 1812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155825008968607e-05,
+      "loss": 0.4279,
+      "step": 1813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154691004435742e-05,
+      "loss": 0.3866,
+      "step": 1814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153556272355596e-05,
+      "loss": 0.3923,
+      "step": 1815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915242081281834e-05,
+      "loss": 0.389,
+      "step": 1816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151284625914214e-05,
+      "loss": 0.4482,
+      "step": 1817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915014771173351e-05,
+      "loss": 0.3773,
+      "step": 1818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149010070366594e-05,
+      "loss": 0.3971,
+      "step": 1819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147871701903864e-05,
+      "loss": 0.3725,
+      "step": 1820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146732606435795e-05,
+      "loss": 0.3692,
+      "step": 1821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914559278405291e-05,
+      "loss": 0.3976,
+      "step": 1822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144452234845802e-05,
+      "loss": 0.4004,
+      "step": 1823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143310958905106e-05,
+      "loss": 0.3961,
+      "step": 1824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142168956321526e-05,
+      "loss": 0.3863,
+      "step": 1825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141026227185813e-05,
+      "loss": 0.4345,
+      "step": 1826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139882771588794e-05,
+      "loss": 0.4133,
+      "step": 1827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138738589621336e-05,
+      "loss": 0.3895,
+      "step": 1828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913759368137437e-05,
+      "loss": 0.4324,
+      "step": 1829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136448046938883e-05,
+      "loss": 0.3774,
+      "step": 1830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135301686405924e-05,
+      "loss": 0.4356,
+      "step": 1831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91341545998666e-05,
+      "loss": 0.3787,
+      "step": 1832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913300678741207e-05,
+      "loss": 0.4231,
+      "step": 1833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131858249133553e-05,
+      "loss": 0.365,
+      "step": 1834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130708985122325e-05,
+      "loss": 0.4049,
+      "step": 1835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129558995469727e-05,
+      "loss": 0.4086,
+      "step": 1836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912840828026715e-05,
+      "loss": 0.4058,
+      "step": 1837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127256839606036e-05,
+      "loss": 0.3955,
+      "step": 1838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126104673577904e-05,
+      "loss": 0.4023,
+      "step": 1839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124951782274314e-05,
+      "loss": 0.4096,
+      "step": 1840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123798165786888e-05,
+      "loss": 0.4572,
+      "step": 1841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122643824207313e-05,
+      "loss": 0.3981,
+      "step": 1842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912148875762732e-05,
+      "loss": 0.3669,
+      "step": 1843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912033296613871e-05,
+      "loss": 0.3973,
+      "step": 1844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911917644983334e-05,
+      "loss": 0.3937,
+      "step": 1845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118019208803113e-05,
+      "loss": 0.3908,
+      "step": 1846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911686124314e-05,
+      "loss": 0.4711,
+      "step": 1847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115702552936033e-05,
+      "loss": 0.3714,
+      "step": 1848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911454313828329e-05,
+      "loss": 0.4193,
+      "step": 1849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911338299927392e-05,
+      "loss": 0.3801,
+      "step": 1850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112222136000115e-05,
+      "loss": 0.403,
+      "step": 1851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111060548554134e-05,
+      "loss": 0.3714,
+      "step": 1852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109898237028295e-05,
+      "loss": 0.3976,
+      "step": 1853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108735201514964e-05,
+      "loss": 0.4057,
+      "step": 1854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107571442106577e-05,
+      "loss": 0.4091,
+      "step": 1855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9106406958895615e-05,
+      "loss": 0.4151,
+      "step": 1856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105241751974624e-05,
+      "loss": 0.4073,
+      "step": 1857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910407582143621e-05,
+      "loss": 0.4063,
+      "step": 1858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102909167373024e-05,
+      "loss": 0.3916,
+      "step": 1859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101741789877796e-05,
+      "loss": 0.3649,
+      "step": 1860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100573689043285e-05,
+      "loss": 0.4193,
+      "step": 1861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099404864962337e-05,
+      "loss": 0.4236,
+      "step": 1862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098235317727837e-05,
+      "loss": 0.4365,
+      "step": 1863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097065047432732e-05,
+      "loss": 0.3945,
+      "step": 1864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909589405417002e-05,
+      "loss": 0.4029,
+      "step": 1865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909472233803277e-05,
+      "loss": 0.394,
+      "step": 1866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90935498991141e-05,
+      "loss": 0.4256,
+      "step": 1867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092376737507184e-05,
+      "loss": 0.4347,
+      "step": 1868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909120285330526e-05,
+      "loss": 0.3951,
+      "step": 1869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090028246601617e-05,
+      "loss": 0.4023,
+      "step": 1870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908885291748961e-05,
+      "loss": 0.4169,
+      "step": 1871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087676866062638e-05,
+      "loss": 0.4628,
+      "step": 1872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086500092414165e-05,
+      "loss": 0.4037,
+      "step": 1873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085322596637718e-05,
+      "loss": 0.4041,
+      "step": 1874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908414437882687e-05,
+      "loss": 0.4028,
+      "step": 1875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908296543907526e-05,
+      "loss": 0.4054,
+      "step": 1876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081785777476583e-05,
+      "loss": 0.4349,
+      "step": 1877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080605394124588e-05,
+      "loss": 0.4038,
+      "step": 1878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079424289113078e-05,
+      "loss": 0.4151,
+      "step": 1879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907824246253593e-05,
+      "loss": 0.4261,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077059914487058e-05,
+      "loss": 0.4054,
+      "step": 1881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075876645060444e-05,
+      "loss": 0.3907,
+      "step": 1882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074692654350126e-05,
+      "loss": 0.4254,
+      "step": 1883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073507942450198e-05,
+      "loss": 0.3931,
+      "step": 1884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072322509454814e-05,
+      "loss": 0.4364,
+      "step": 1885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071136355458186e-05,
+      "loss": 0.3998,
+      "step": 1886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069949480554574e-05,
+      "loss": 0.386,
+      "step": 1887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068761884838304e-05,
+      "loss": 0.4148,
+      "step": 1888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906757356840376e-05,
+      "loss": 0.4029,
+      "step": 1889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906638453134538e-05,
+      "loss": 0.4048,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906519477375766e-05,
+      "loss": 0.4151,
+      "step": 1891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906400429573515e-05,
+      "loss": 0.3895,
+      "step": 1892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062813097372467e-05,
+      "loss": 0.4335,
+      "step": 1893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906162117876427e-05,
+      "loss": 0.4363,
+      "step": 1894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060428540005287e-05,
+      "loss": 0.416,
+      "step": 1895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9059235181190306e-05,
+      "loss": 0.4001,
+      "step": 1896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905804110241416e-05,
+      "loss": 0.3749,
+      "step": 1897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905684630377175e-05,
+      "loss": 0.3975,
+      "step": 1898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055650785358023e-05,
+      "loss": 0.3769,
+      "step": 1899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054454547267997e-05,
+      "loss": 0.3893,
+      "step": 1900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053257589596736e-05,
+      "loss": 0.3934,
+      "step": 1901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052059912439366e-05,
+      "loss": 0.4129,
+      "step": 1902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050861515891072e-05,
+      "loss": 0.3987,
+      "step": 1903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049662400047092e-05,
+      "loss": 0.4178,
+      "step": 1904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048462565002724e-05,
+      "loss": 0.3893,
+      "step": 1905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047262010853317e-05,
+      "loss": 0.4038,
+      "step": 1906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046060737694286e-05,
+      "loss": 0.3687,
+      "step": 1907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90448587456211e-05,
+      "loss": 0.3679,
+      "step": 1908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904365603472929e-05,
+      "loss": 0.4471,
+      "step": 1909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042452605114427e-05,
+      "loss": 0.3748,
+      "step": 1910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041248456872158e-05,
+      "loss": 0.3929,
+      "step": 1911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040043590098176e-05,
+      "loss": 0.3641,
+      "step": 1912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903883800488824e-05,
+      "loss": 0.4072,
+      "step": 1913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037631701338154e-05,
+      "loss": 0.4241,
+      "step": 1914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903642467954379e-05,
+      "loss": 0.474,
+      "step": 1915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035216939601072e-05,
+      "loss": 0.3922,
+      "step": 1916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903400848160599e-05,
+      "loss": 0.4344,
+      "step": 1917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032799305654572e-05,
+      "loss": 0.3851,
+      "step": 1918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903158941184292e-05,
+      "loss": 0.3832,
+      "step": 1919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030378800267187e-05,
+      "loss": 0.386,
+      "step": 1920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902916747102358e-05,
+      "loss": 0.4333,
+      "step": 1921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027955424208376e-05,
+      "loss": 0.3916,
+      "step": 1922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902674265991789e-05,
+      "loss": 0.4338,
+      "step": 1923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025529178248504e-05,
+      "loss": 0.3851,
+      "step": 1924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902431497929666e-05,
+      "loss": 0.4064,
+      "step": 1925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023100063158857e-05,
+      "loss": 0.4163,
+      "step": 1926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021884429931638e-05,
+      "loss": 0.4113,
+      "step": 1927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020668079711623e-05,
+      "loss": 0.3894,
+      "step": 1928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901945101259547e-05,
+      "loss": 0.3943,
+      "step": 1929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018233228679906e-05,
+      "loss": 0.4235,
+      "step": 1930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017014728061714e-05,
+      "loss": 0.4077,
+      "step": 1931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901579551083773e-05,
+      "loss": 0.3695,
+      "step": 1932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014575577104842e-05,
+      "loss": 0.4308,
+      "step": 1933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013354926960007e-05,
+      "loss": 0.4347,
+      "step": 1934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012133560500235e-05,
+      "loss": 0.3771,
+      "step": 1935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901091147782259e-05,
+      "loss": 0.3872,
+      "step": 1936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.378,
+      "step": 1937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008465164202223e-05,
+      "loss": 0.3642,
+      "step": 1938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007240933453916e-05,
+      "loss": 0.4198,
+      "step": 1939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006015986876564e-05,
+      "loss": 0.4121,
+      "step": 1940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900479032456752e-05,
+      "loss": 0.4189,
+      "step": 1941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003563946624185e-05,
+      "loss": 0.378,
+      "step": 1942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002336853144027e-05,
+      "loss": 0.3667,
+      "step": 1943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001109044224567e-05,
+      "loss": 0.3966,
+      "step": 1944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8999880519963375e-05,
+      "loss": 0.4519,
+      "step": 1945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89986512804581e-05,
+      "loss": 0.4065,
+      "step": 1946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997421325806415e-05,
+      "loss": 0.4008,
+      "step": 1947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899619065610608e-05,
+      "loss": 0.3871,
+      "step": 1948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899495927145489e-05,
+      "loss": 0.3909,
+      "step": 1949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993727171950715e-05,
+      "loss": 0.4032,
+      "step": 1950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992494357691475e-05,
+      "loss": 0.4103,
+      "step": 1951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991260828775135e-05,
+      "loss": 0.371,
+      "step": 1952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899002658529973e-05,
+      "loss": 0.3816,
+      "step": 1953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898879162736335e-05,
+      "loss": 0.3867,
+      "step": 1954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987555955064144e-05,
+      "loss": 0.3824,
+      "step": 1955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986319568500315e-05,
+      "loss": 0.4344,
+      "step": 1956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898508246777011e-05,
+      "loss": 0.4228,
+      "step": 1957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983844652971854e-05,
+      "loss": 0.4149,
+      "step": 1958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898260612420392e-05,
+      "loss": 0.4012,
+      "step": 1959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981366881564733e-05,
+      "loss": 0.3847,
+      "step": 1960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980126925152783e-05,
+      "loss": 0.4375,
+      "step": 1961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897888625506661e-05,
+      "loss": 0.401,
+      "step": 1962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897764487140481e-05,
+      "loss": 0.4251,
+      "step": 1963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976402774266047e-05,
+      "loss": 0.4352,
+      "step": 1964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975159963749024e-05,
+      "loss": 0.3922,
+      "step": 1965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897391643995252e-05,
+      "loss": 0.4048,
+      "step": 1966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972672202975354e-05,
+      "loss": 0.3924,
+      "step": 1967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971427252916415e-05,
+      "loss": 0.4174,
+      "step": 1968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970181589874637e-05,
+      "loss": 0.3718,
+      "step": 1969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896893521394902e-05,
+      "loss": 0.4043,
+      "step": 1970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896768812523861e-05,
+      "loss": 0.4214,
+      "step": 1971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966440323842527e-05,
+      "loss": 0.4427,
+      "step": 1972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965191809859925e-05,
+      "loss": 0.3907,
+      "step": 1973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896394258339004e-05,
+      "loss": 0.3972,
+      "step": 1974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962692644532142e-05,
+      "loss": 0.4654,
+      "step": 1975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961441993385566e-05,
+      "loss": 0.3944,
+      "step": 1976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896019063004971e-05,
+      "loss": 0.3926,
+      "step": 1977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958938554624024e-05,
+      "loss": 0.4085,
+      "step": 1978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895768576720801e-05,
+      "loss": 0.3903,
+      "step": 1979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956432267901225e-05,
+      "loss": 0.3997,
+      "step": 1980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955178056803297e-05,
+      "loss": 0.4513,
+      "step": 1981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89539231340139e-05,
+      "loss": 0.4023,
+      "step": 1982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952667499632764e-05,
+      "loss": 0.4089,
+      "step": 1983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951411153759678e-05,
+      "loss": 0.3881,
+      "step": 1984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950154096494486e-05,
+      "loss": 0.4116,
+      "step": 1985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948896327937086e-05,
+      "loss": 0.3796,
+      "step": 1986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894763784818745e-05,
+      "loss": 0.4354,
+      "step": 1987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946378657345575e-05,
+      "loss": 0.3677,
+      "step": 1988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945118755511544e-05,
+      "loss": 0.3873,
+      "step": 1989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894385814278548e-05,
+      "loss": 0.3734,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942596819267573e-05,
+      "loss": 0.3875,
+      "step": 1991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941334785058054e-05,
+      "loss": 0.3948,
+      "step": 1992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940072040257228e-05,
+      "loss": 0.4379,
+      "step": 1993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938808584965444e-05,
+      "loss": 0.4154,
+      "step": 1994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937544419283115e-05,
+      "loss": 0.3451,
+      "step": 1995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936279543310706e-05,
+      "loss": 0.4312,
+      "step": 1996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893501395714874e-05,
+      "loss": 0.4037,
+      "step": 1997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933747660897803e-05,
+      "loss": 0.4073,
+      "step": 1998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932480654658518e-05,
+      "loss": 0.3926,
+      "step": 1999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931212938531586e-05,
+      "loss": 0.4403,
+      "step": 2000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929944512617753e-05,
+      "loss": 0.3852,
+      "step": 2001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892867537701783e-05,
+      "loss": 0.4174,
+      "step": 2002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892740553183267e-05,
+      "loss": 0.4269,
+      "step": 2003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926134977163196e-05,
+      "loss": 0.4071,
+      "step": 2004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892486371311038e-05,
+      "loss": 0.3886,
+      "step": 2005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923591739775254e-05,
+      "loss": 0.3657,
+      "step": 2006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89223190572589e-05,
+      "loss": 0.4378,
+      "step": 2007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921045665662472e-05,
+      "loss": 0.4389,
+      "step": 2008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891977156508716e-05,
+      "loss": 0.3737,
+      "step": 2009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918496755634227e-05,
+      "loss": 0.4327,
+      "step": 2010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891722123740498e-05,
+      "loss": 0.3966,
+      "step": 2011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915945010500792e-05,
+      "loss": 0.4242,
+      "step": 2012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914668075023087e-05,
+      "loss": 0.3774,
+      "step": 2013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913390431073346e-05,
+      "loss": 0.3955,
+      "step": 2014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912112078753103e-05,
+      "loss": 0.4066,
+      "step": 2015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8910833018163956e-05,
+      "loss": 0.4051,
+      "step": 2016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909553249407558e-05,
+      "loss": 0.4086,
+      "step": 2017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890827277258561e-05,
+      "loss": 0.4414,
+      "step": 2018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906991587799878e-05,
+      "loss": 0.4223,
+      "step": 2019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890570969515218e-05,
+      "loss": 0.3989,
+      "step": 2020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904427094744388e-05,
+      "loss": 0.4352,
+      "step": 2021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890314378667844e-05,
+      "loss": 0.4167,
+      "step": 2022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901859771056322e-05,
+      "loss": 0.4005,
+      "step": 2023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900575047980078e-05,
+      "loss": 0.4126,
+      "step": 2024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899289617551803e-05,
+      "loss": 0.3956,
+      "step": 2025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898003479873664e-05,
+      "loss": 0.4045,
+      "step": 2026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896716635047863e-05,
+      "loss": 0.3737,
+      "step": 2027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895429083176675e-05,
+      "loss": 0.4273,
+      "step": 2028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894140824362424e-05,
+      "loss": 0.3867,
+      "step": 2029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889285185870749e-05,
+      "loss": 0.4285,
+      "step": 2030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891562186314312e-05,
+      "loss": 0.4152,
+      "step": 2031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890271807285386e-05,
+      "loss": 0.4098,
+      "step": 2032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888980721723256e-05,
+      "loss": 0.3954,
+      "step": 2033
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888768892973053e-05,
+      "loss": 0.4139,
+      "step": 2034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886396431409874e-05,
+      "loss": 0.3981,
+      "step": 2035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885103226864e-05,
+      "loss": 0.3629,
+      "step": 2036
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888380931619569e-05,
+      "loss": 0.3743,
+      "step": 2037
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8882514699507766e-05,
+      "loss": 0.3754,
+      "step": 2038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881219376903122e-05,
+      "loss": 0.4024,
+      "step": 2039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879923348484694e-05,
+      "loss": 0.3896,
+      "step": 2040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878626614355487e-05,
+      "loss": 0.4749,
+      "step": 2041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887732917461855e-05,
+      "loss": 0.4128,
+      "step": 2042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876031029376998e-05,
+      "loss": 0.4069,
+      "step": 2043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874732178734e-05,
+      "loss": 0.3942,
+      "step": 2044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873432622792774e-05,
+      "loss": 0.437,
+      "step": 2045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88721323616566e-05,
+      "loss": 0.3889,
+      "step": 2046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870831395428814e-05,
+      "loss": 0.4048,
+      "step": 2047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886952972421281e-05,
+      "loss": 0.3795,
+      "step": 2048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868227348112033e-05,
+      "loss": 0.396,
+      "step": 2049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866924267229984e-05,
+      "loss": 0.4527,
+      "step": 2050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886562048167022e-05,
+      "loss": 0.3805,
+      "step": 2051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864315991536368e-05,
+      "loss": 0.3714,
+      "step": 2052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886301079693209e-05,
+      "loss": 0.4014,
+      "step": 2053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8861704897961112e-05,
+      "loss": 0.4469,
+      "step": 2054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886039829472722e-05,
+      "loss": 0.3945,
+      "step": 2055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859090987334258e-05,
+      "loss": 0.3977,
+      "step": 2056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885778297588611e-05,
+      "loss": 0.3811,
+      "step": 2057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8856474260486738e-05,
+      "loss": 0.4047,
+      "step": 2058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855164841240143e-05,
+      "loss": 0.4009,
+      "step": 2059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885385471825039e-05,
+      "loss": 0.3921,
+      "step": 2060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88525438916216e-05,
+      "loss": 0.4196,
+      "step": 2061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885123236145794e-05,
+      "loss": 0.3955,
+      "step": 2062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884992012786365e-05,
+      "loss": 0.4247,
+      "step": 2063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848607190943012e-05,
+      "loss": 0.377,
+      "step": 2064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884729355080037e-05,
+      "loss": 0.4195,
+      "step": 2065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845979207540123e-05,
+      "loss": 0.4165,
+      "step": 2066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8844664161266722e-05,
+      "loss": 0.384,
+      "step": 2067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843348412084677e-05,
+      "loss": 0.4404,
+      "step": 2068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884203196009856e-05,
+      "loss": 0.3989,
+      "step": 2069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840714805412986e-05,
+      "loss": 0.3975,
+      "step": 2070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883939694813264e-05,
+      "loss": 0.3922,
+      "step": 2071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838078388362243e-05,
+      "loss": 0.387,
+      "step": 2072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836759126206598e-05,
+      "loss": 0.3978,
+      "step": 2073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883543916177055e-05,
+      "loss": 0.4272,
+      "step": 2074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834118495158987e-05,
+      "loss": 0.387,
+      "step": 2075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832797126476875e-05,
+      "loss": 0.4058,
+      "step": 2076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831475055829228e-05,
+      "loss": 0.4322,
+      "step": 2077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830152283321113e-05,
+      "loss": 0.4042,
+      "step": 2078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828828809057653e-05,
+      "loss": 0.4115,
+      "step": 2079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827504633144026e-05,
+      "loss": 0.4026,
+      "step": 2080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882617975568547e-05,
+      "loss": 0.4014,
+      "step": 2081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824854176787276e-05,
+      "loss": 0.3996,
+      "step": 2082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823527896554794e-05,
+      "loss": 0.4159,
+      "step": 2083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882220091509342e-05,
+      "loss": 0.3873,
+      "step": 2084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820873232508624e-05,
+      "loss": 0.4042,
+      "step": 2085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881954484890591e-05,
+      "loss": 0.3889,
+      "step": 2086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881821576439085e-05,
+      "loss": 0.4397,
+      "step": 2087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881688597906907e-05,
+      "loss": 0.4221,
+      "step": 2088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881555549304626e-05,
+      "loss": 0.4318,
+      "step": 2089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814224306428147e-05,
+      "loss": 0.4039,
+      "step": 2090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812892419320525e-05,
+      "loss": 0.4017,
+      "step": 2091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811559831829247e-05,
+      "loss": 0.4231,
+      "step": 2092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810226544060215e-05,
+      "loss": 0.3924,
+      "step": 2093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880889255611939e-05,
+      "loss": 0.4123,
+      "step": 2094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807557868112783e-05,
+      "loss": 0.3856,
+      "step": 2095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880622248014647e-05,
+      "loss": 0.3979,
+      "step": 2096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804886392326578e-05,
+      "loss": 0.3914,
+      "step": 2097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803549604759286e-05,
+      "loss": 0.4163,
+      "step": 2098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802212117550833e-05,
+      "loss": 0.4213,
+      "step": 2099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800873930807515e-05,
+      "loss": 0.3941,
+      "step": 2100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879953504463568e-05,
+      "loss": 0.3784,
+      "step": 2101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879819545914173e-05,
+      "loss": 0.3522,
+      "step": 2102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879685517443213e-05,
+      "loss": 0.453,
+      "step": 2103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795514190613392e-05,
+      "loss": 0.4194,
+      "step": 2104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879417250779209e-05,
+      "loss": 0.4327,
+      "step": 2105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879283012607485e-05,
+      "loss": 0.3693,
+      "step": 2106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791487045568352e-05,
+      "loss": 0.4161,
+      "step": 2107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879014326637934e-05,
+      "loss": 0.4087,
+      "step": 2108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788798788614606e-05,
+      "loss": 0.3918,
+      "step": 2109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787453612380996e-05,
+      "loss": 0.3842,
+      "step": 2110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786107737785415e-05,
+      "loss": 0.3783,
+      "step": 2111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784761164934825e-05,
+      "loss": 0.3645,
+      "step": 2112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783413893936244e-05,
+      "loss": 0.4435,
+      "step": 2113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878206592489674e-05,
+      "loss": 0.4031,
+      "step": 2114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780717257923434e-05,
+      "loss": 0.4133,
+      "step": 2115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877936789312352e-05,
+      "loss": 0.4226,
+      "step": 2116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877801783060423e-05,
+      "loss": 0.358,
+      "step": 2117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776667070472852e-05,
+      "loss": 0.3907,
+      "step": 2118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775315612836747e-05,
+      "loss": 0.4255,
+      "step": 2119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773963457803302e-05,
+      "loss": 0.4498,
+      "step": 2120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772610605479992e-05,
+      "loss": 0.388,
+      "step": 2121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771257055974322e-05,
+      "loss": 0.4045,
+      "step": 2122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769902809393866e-05,
+      "loss": 0.424,
+      "step": 2123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876854786584625e-05,
+      "loss": 0.4214,
+      "step": 2124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767192225439154e-05,
+      "loss": 0.396,
+      "step": 2125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876583588828031e-05,
+      "loss": 0.4282,
+      "step": 2126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876447885447752e-05,
+      "loss": 0.3858,
+      "step": 2127
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876312112413862e-05,
+      "loss": 0.3933,
+      "step": 2128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761762697371516e-05,
+      "loss": 0.4054,
+      "step": 2129
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876040357428417e-05,
+      "loss": 0.4278,
+      "step": 2130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759043754984588e-05,
+      "loss": 0.446,
+      "step": 2131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757683239580845e-05,
+      "loss": 0.4209,
+      "step": 2132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756322028181063e-05,
+      "loss": 0.3647,
+      "step": 2133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754960120893418e-05,
+      "loss": 0.4515,
+      "step": 2134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8753597517826146e-05,
+      "loss": 0.4151,
+      "step": 2135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752234219087538e-05,
+      "loss": 0.4438,
+      "step": 2136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875087022478594e-05,
+      "loss": 0.435,
+      "step": 2137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749505535029744e-05,
+      "loss": 0.408,
+      "step": 2138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748140149927417e-05,
+      "loss": 0.401,
+      "step": 2139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746774069587464e-05,
+      "loss": 0.406,
+      "step": 2140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874540729411845e-05,
+      "loss": 0.4167,
+      "step": 2141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744039823628995e-05,
+      "loss": 0.396,
+      "step": 2142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742671658227785e-05,
+      "loss": 0.4019,
+      "step": 2143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741302798023538e-05,
+      "loss": 0.3978,
+      "step": 2144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873993324312505e-05,
+      "loss": 0.3737,
+      "step": 2145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873856299364116e-05,
+      "loss": 0.416,
+      "step": 2146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737192049680768e-05,
+      "loss": 0.4001,
+      "step": 2147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735820411352822e-05,
+      "loss": 0.3911,
+      "step": 2148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734448078766335e-05,
+      "loss": 0.37,
+      "step": 2149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733075052030364e-05,
+      "loss": 0.3663,
+      "step": 2150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873170133125403e-05,
+      "loss": 0.3768,
+      "step": 2151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730326916546512e-05,
+      "loss": 0.4023,
+      "step": 2152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872895180801703e-05,
+      "loss": 0.3883,
+      "step": 2153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727576005774865e-05,
+      "loss": 0.4256,
+      "step": 2154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726199509929366e-05,
+      "loss": 0.3971,
+      "step": 2155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872482232058992e-05,
+      "loss": 0.3995,
+      "step": 2156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723444437865975e-05,
+      "loss": 0.3941,
+      "step": 2157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872206586186704e-05,
+      "loss": 0.3715,
+      "step": 2158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720686592702672e-05,
+      "loss": 0.4167,
+      "step": 2159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719306630482485e-05,
+      "loss": 0.373,
+      "step": 2160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717925975316146e-05,
+      "loss": 0.4493,
+      "step": 2161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871654462731338e-05,
+      "loss": 0.4207,
+      "step": 2162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871516258658397e-05,
+      "loss": 0.433,
+      "step": 2163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713779853237744e-05,
+      "loss": 0.3849,
+      "step": 2164
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712396427384595e-05,
+      "loss": 0.3568,
+      "step": 2165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871101230913447e-05,
+      "loss": 0.3591,
+      "step": 2166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709627498597367e-05,
+      "loss": 0.4117,
+      "step": 2167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870824199588334e-05,
+      "loss": 0.3787,
+      "step": 2168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706855801102497e-05,
+      "loss": 0.4052,
+      "step": 2169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705468914365004e-05,
+      "loss": 0.401,
+      "step": 2170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870408133578108e-05,
+      "loss": 0.3839,
+      "step": 2171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702693065460997e-05,
+      "loss": 0.4104,
+      "step": 2172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870130410351509e-05,
+      "loss": 0.3635,
+      "step": 2173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699914450053743e-05,
+      "loss": 0.3471,
+      "step": 2174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869852410518739e-05,
+      "loss": 0.3653,
+      "step": 2175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869713306902653e-05,
+      "loss": 0.387,
+      "step": 2176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695741341681707e-05,
+      "loss": 0.4147,
+      "step": 2177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8694348923263534e-05,
+      "loss": 0.4062,
+      "step": 2178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692955813882662e-05,
+      "loss": 0.4018,
+      "step": 2179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691562013649806e-05,
+      "loss": 0.4041,
+      "step": 2180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690167522675743e-05,
+      "loss": 0.3624,
+      "step": 2181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688772341071287e-05,
+      "loss": 0.4018,
+      "step": 2182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868737646894732e-05,
+      "loss": 0.4094,
+      "step": 2183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868597990641478e-05,
+      "loss": 0.4128,
+      "step": 2184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684582653584647e-05,
+      "loss": 0.4547,
+      "step": 2185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868318471056797e-05,
+      "loss": 0.4004,
+      "step": 2186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681786077475854e-05,
+      "loss": 0.3988,
+      "step": 2187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868038675441944e-05,
+      "loss": 0.4086,
+      "step": 2188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678986741509935e-05,
+      "loss": 0.4261,
+      "step": 2189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867758603885861e-05,
+      "loss": 0.4091,
+      "step": 2190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867618464657678e-05,
+      "loss": 0.3865,
+      "step": 2191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674782564775816e-05,
+      "loss": 0.4128,
+      "step": 2192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867337979356715e-05,
+      "loss": 0.4169,
+      "step": 2193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671976333062253e-05,
+      "loss": 0.4049,
+      "step": 2194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670572183372672e-05,
+      "loss": 0.3666,
+      "step": 2195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669167344609995e-05,
+      "loss": 0.3855,
+      "step": 2196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8667761816885867e-05,
+      "loss": 0.3868,
+      "step": 2197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8666355600311994e-05,
+      "loss": 0.3545,
+      "step": 2198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664948695000124e-05,
+      "loss": 0.3881,
+      "step": 2199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663541101062073e-05,
+      "loss": 0.4094,
+      "step": 2200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662132818609703e-05,
+      "loss": 0.3927,
+      "step": 2201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8660723847754936e-05,
+      "loss": 0.4685,
+      "step": 2202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659314188609746e-05,
+      "loss": 0.4188,
+      "step": 2203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657903841286162e-05,
+      "loss": 0.4403,
+      "step": 2204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865649280589627e-05,
+      "loss": 0.3857,
+      "step": 2205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655081082552202e-05,
+      "loss": 0.3994,
+      "step": 2206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865366867136616e-05,
+      "loss": 0.3833,
+      "step": 2207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865225557245039e-05,
+      "loss": 0.3898,
+      "step": 2208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865084178591719e-05,
+      "loss": 0.4009,
+      "step": 2209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649427311878922e-05,
+      "loss": 0.4052,
+      "step": 2210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8648012150447994e-05,
+      "loss": 0.3941,
+      "step": 2211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646596301736878e-05,
+      "loss": 0.3984,
+      "step": 2212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645179765858092e-05,
+      "loss": 0.3664,
+      "step": 2213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643762542924208e-05,
+      "loss": 0.3955,
+      "step": 2214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642344633047864e-05,
+      "loss": 0.4319,
+      "step": 2215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864092603634174e-05,
+      "loss": 0.4117,
+      "step": 2216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639506752918578e-05,
+      "loss": 0.408,
+      "step": 2217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638086782891166e-05,
+      "loss": 0.3811,
+      "step": 2218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636666126372363e-05,
+      "loss": 0.392,
+      "step": 2219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8635244783475065e-05,
+      "loss": 0.3922,
+      "step": 2220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633822754312233e-05,
+      "loss": 0.3974,
+      "step": 2221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863240003899688e-05,
+      "loss": 0.3831,
+      "step": 2222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630976637642072e-05,
+      "loss": 0.4068,
+      "step": 2223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629552550360925e-05,
+      "loss": 0.418,
+      "step": 2224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8628127777266624e-05,
+      "loss": 0.4069,
+      "step": 2225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626702318472392e-05,
+      "loss": 0.3688,
+      "step": 2226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862527617409152e-05,
+      "loss": 0.4029,
+      "step": 2227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862384934423734e-05,
+      "loss": 0.4017,
+      "step": 2228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622421829023255e-05,
+      "loss": 0.3896,
+      "step": 2229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8620993628562707e-05,
+      "loss": 0.4211,
+      "step": 2230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86195647429692e-05,
+      "loss": 0.3687,
+      "step": 2231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861813517235629e-05,
+      "loss": 0.3927,
+      "step": 2232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616704916837595e-05,
+      "loss": 0.3524,
+      "step": 2233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615273976526772e-05,
+      "loss": 0.3918,
+      "step": 2234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861384235153755e-05,
+      "loss": 0.4021,
+      "step": 2235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86124100419837e-05,
+      "loss": 0.4762,
+      "step": 2236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610977047979052e-05,
+      "loss": 0.3984,
+      "step": 2237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609543369637487e-05,
+      "loss": 0.4168,
+      "step": 2238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860810900707295e-05,
+      "loss": 0.3646,
+      "step": 2239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8606673960399426e-05,
+      "loss": 0.3686,
+      "step": 2240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605238229730965e-05,
+      "loss": 0.3956,
+      "step": 2241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860380181518167e-05,
+      "loss": 0.3909,
+      "step": 2242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602364716865693e-05,
+      "loss": 0.3925,
+      "step": 2243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600926934897248e-05,
+      "loss": 0.3905,
+      "step": 2244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599488469390596e-05,
+      "loss": 0.4116,
+      "step": 2245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859804932046006e-05,
+      "loss": 0.383,
+      "step": 2246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8596609488220005e-05,
+      "loss": 0.4967,
+      "step": 2247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595168972784865e-05,
+      "loss": 0.4493,
+      "step": 2248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593727774269122e-05,
+      "loss": 0.3951,
+      "step": 2249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592285892787306e-05,
+      "loss": 0.423,
+      "step": 2250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590843328454013e-05,
+      "loss": 0.3644,
+      "step": 2251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589400081383886e-05,
+      "loss": 0.3917,
+      "step": 2252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587956151691622e-05,
+      "loss": 0.3886,
+      "step": 2253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8586511539491973e-05,
+      "loss": 0.4081,
+      "step": 2254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858506624489975e-05,
+      "loss": 0.4174,
+      "step": 2255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583620268029808e-05,
+      "loss": 0.3991,
+      "step": 2256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582173608997068e-05,
+      "loss": 0.402,
+      "step": 2257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85807262679165e-05,
+      "loss": 0.3934,
+      "step": 2258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579278244903124e-05,
+      "loss": 0.3873,
+      "step": 2259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577829540072022e-05,
+      "loss": 0.3625,
+      "step": 2260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576380153538324e-05,
+      "loss": 0.3545,
+      "step": 2261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857493008541722e-05,
+      "loss": 0.3767,
+      "step": 2262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573479335823946e-05,
+      "loss": 0.4274,
+      "step": 2263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572027904873796e-05,
+      "loss": 0.4173,
+      "step": 2264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570575792682126e-05,
+      "loss": 0.4015,
+      "step": 2265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569122999364337e-05,
+      "loss": 0.3697,
+      "step": 2266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856766952503588e-05,
+      "loss": 0.4294,
+      "step": 2267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566215369812274e-05,
+      "loss": 0.3821,
+      "step": 2268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564760533809077e-05,
+      "loss": 0.3999,
+      "step": 2269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563305017141913e-05,
+      "loss": 0.3658,
+      "step": 2270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561848819926464e-05,
+      "loss": 0.4219,
+      "step": 2271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560391942278443e-05,
+      "loss": 0.3677,
+      "step": 2272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855893438431364e-05,
+      "loss": 0.4002,
+      "step": 2273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855747614614789e-05,
+      "loss": 0.3764,
+      "step": 2274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556017227897077e-05,
+      "loss": 0.4192,
+      "step": 2275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855455762967716e-05,
+      "loss": 0.3563,
+      "step": 2276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855309735160412e-05,
+      "loss": 0.3911,
+      "step": 2277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551636393794022e-05,
+      "loss": 0.407,
+      "step": 2278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550174756362962e-05,
+      "loss": 0.3931,
+      "step": 2279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548712439427107e-05,
+      "loss": 0.3846,
+      "step": 2280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547249443102672e-05,
+      "loss": 0.4028,
+      "step": 2281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545785767505915e-05,
+      "loss": 0.3967,
+      "step": 2282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854432141275317e-05,
+      "loss": 0.3695,
+      "step": 2283
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854285637896081e-05,
+      "loss": 0.3779,
+      "step": 2284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854139066624526e-05,
+      "loss": 0.4113,
+      "step": 2285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853992427472301e-05,
+      "loss": 0.4169,
+      "step": 2286
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8538457204510597e-05,
+      "loss": 0.4019,
+      "step": 2287
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853698945572461e-05,
+      "loss": 0.3688,
+      "step": 2288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85355210284817e-05,
+      "loss": 0.3842,
+      "step": 2289
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534051922898556e-05,
+      "loss": 0.3953,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532582139091945e-05,
+      "loss": 0.389,
+      "step": 2291
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531111677178663e-05,
+      "loss": 0.4154,
+      "step": 2292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852964053727558e-05,
+      "loss": 0.3859,
+      "step": 2293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528168719499607e-05,
+      "loss": 0.4236,
+      "step": 2294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8526696223967714e-05,
+      "loss": 0.3883,
+      "step": 2295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525223050796928e-05,
+      "loss": 0.3758,
+      "step": 2296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523749200104318e-05,
+      "loss": 0.3825,
+      "step": 2297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522274672007022e-05,
+      "loss": 0.439,
+      "step": 2298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520799466622222e-05,
+      "loss": 0.4009,
+      "step": 2299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8519323584067154e-05,
+      "loss": 0.3743,
+      "step": 2300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517847024459116e-05,
+      "loss": 0.373,
+      "step": 2301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516369787915448e-05,
+      "loss": 0.4428,
+      "step": 2302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514891874553552e-05,
+      "loss": 0.3505,
+      "step": 2303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513413284490883e-05,
+      "loss": 0.3677,
+      "step": 2304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851193401784495e-05,
+      "loss": 0.3662,
+      "step": 2305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851045407473331e-05,
+      "loss": 0.4362,
+      "step": 2306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508973455273577e-05,
+      "loss": 0.4177,
+      "step": 2307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507492159583423e-05,
+      "loss": 0.3929,
+      "step": 2308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506010187780573e-05,
+      "loss": 0.4273,
+      "step": 2309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85045275399828e-05,
+      "loss": 0.421,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8503044216307924e-05,
+      "loss": 0.3678,
+      "step": 2311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850156021687385e-05,
+      "loss": 0.4232,
+      "step": 2312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85000755417985e-05,
+      "loss": 0.3876,
+      "step": 2313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849859019119987e-05,
+      "loss": 0.3767,
+      "step": 2314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8497104165196003e-05,
+      "loss": 0.3793,
+      "step": 2315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495617463905e-05,
+      "loss": 0.3693,
+      "step": 2316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8494130087445007e-05,
+      "loss": 0.3788,
+      "step": 2317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492642035934232e-05,
+      "loss": 0.4202,
+      "step": 2318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491153309490943e-05,
+      "loss": 0.3928,
+      "step": 2319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848966390823344e-05,
+      "loss": 0.4473,
+      "step": 2320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84881738322801e-05,
+      "loss": 0.392,
+      "step": 2321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486683081749336e-05,
+      "loss": 0.4072,
+      "step": 2322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848519165675963e-05,
+      "loss": 0.4294,
+      "step": 2323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483699557429497e-05,
+      "loss": 0.3841,
+      "step": 2324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8482206783877527e-05,
+      "loss": 0.4212,
+      "step": 2325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848071333622236e-05,
+      "loss": 0.421,
+      "step": 2326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8479219214582668e-05,
+      "loss": 0.402,
+      "step": 2327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8477724419077208e-05,
+      "loss": 0.3914,
+      "step": 2328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476228949824764e-05,
+      "loss": 0.3945,
+      "step": 2329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8474732806944195e-05,
+      "loss": 0.3757,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473235990554393e-05,
+      "loss": 0.4101,
+      "step": 2331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471738500774324e-05,
+      "loss": 0.3848,
+      "step": 2332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847024033772299e-05,
+      "loss": 0.4068,
+      "step": 2333
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8468741501519457e-05,
+      "loss": 0.37,
+      "step": 2334
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467241992282842e-05,
+      "loss": 0.3609,
+      "step": 2335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465741810132317e-05,
+      "loss": 0.415,
+      "step": 2336
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84642409551871e-05,
+      "loss": 0.4002,
+      "step": 2337
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846273942756647e-05,
+      "loss": 0.3949,
+      "step": 2338
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461237227389755e-05,
+      "loss": 0.4536,
+      "step": 2339
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459734354776343e-05,
+      "loss": 0.4187,
+      "step": 2340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458230809845674e-05,
+      "loss": 0.4054,
+      "step": 2341
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845672659271723e-05,
+      "loss": 0.4442,
+      "step": 2342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8455221703510563e-05,
+      "loss": 0.3878,
+      "step": 2343
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453716142345265e-05,
+      "loss": 0.3949,
+      "step": 2344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452209909340986e-05,
+      "loss": 0.3778,
+      "step": 2345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845070300461744e-05,
+      "loss": 0.4093,
+      "step": 2346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8449195428294372e-05,
+      "loss": 0.4383,
+      "step": 2347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447687180491602e-05,
+      "loss": 0.3804,
+      "step": 2348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446178261328986e-05,
+      "loss": 0.3774,
+      "step": 2349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444668670926455e-05,
+      "loss": 0.3747,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443158409403965e-05,
+      "loss": 0.3854,
+      "step": 2351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441647476881553e-05,
+      "loss": 0.3872,
+      "step": 2352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440135873479287e-05,
+      "loss": 0.3781,
+      "step": 2353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438623599317304e-05,
+      "loss": 0.3801,
+      "step": 2354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843711065451579e-05,
+      "loss": 0.3875,
+      "step": 2355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435597039194974e-05,
+      "loss": 0.4025,
+      "step": 2356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434082753475154e-05,
+      "loss": 0.4259,
+      "step": 2357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432567797476675e-05,
+      "loss": 0.3789,
+      "step": 2358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843105217131993e-05,
+      "loss": 0.3962,
+      "step": 2359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429535875125374e-05,
+      "loss": 0.365,
+      "step": 2360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842801890901351e-05,
+      "loss": 0.4365,
+      "step": 2361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842650127310489e-05,
+      "loss": 0.3989,
+      "step": 2362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8424982967520133e-05,
+      "loss": 0.3742,
+      "step": 2363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423463992379894e-05,
+      "loss": 0.425,
+      "step": 2364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84219443478049e-05,
+      "loss": 0.4225,
+      "step": 2365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842042403391591e-05,
+      "loss": 0.409,
+      "step": 2366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841890305083376e-05,
+      "loss": 0.3713,
+      "step": 2367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841738139867931e-05,
+      "loss": 0.4271,
+      "step": 2368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841585907757351e-05,
+      "loss": 0.382,
+      "step": 2369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414336087637325e-05,
+      "loss": 0.3938,
+      "step": 2370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84128124289918e-05,
+      "loss": 0.3747,
+      "step": 2371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841128810175802e-05,
+      "loss": 0.3988,
+      "step": 2372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840976310605713e-05,
+      "loss": 0.373,
+      "step": 2373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840823744201033e-05,
+      "loss": 0.3543,
+      "step": 2374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406711109738856e-05,
+      "loss": 0.4334,
+      "step": 2375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8405184109364023e-05,
+      "loss": 0.4035,
+      "step": 2376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403656441007176e-05,
+      "loss": 0.3874,
+      "step": 2377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8402128104789727e-05,
+      "loss": 0.3808,
+      "step": 2378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840059910083314e-05,
+      "loss": 0.4154,
+      "step": 2379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839906942925892e-05,
+      "loss": 0.4395,
+      "step": 2380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397539090188644e-05,
+      "loss": 0.4063,
+      "step": 2381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396008083743925e-05,
+      "loss": 0.3682,
+      "step": 2382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394476410046436e-05,
+      "loss": 0.3396,
+      "step": 2383
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392944069217908e-05,
+      "loss": 0.4065,
+      "step": 2384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839141106138012e-05,
+      "loss": 0.3669,
+      "step": 2385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83898773866549e-05,
+      "loss": 0.4052,
+      "step": 2386
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388343045164135e-05,
+      "loss": 0.4125,
+      "step": 2387
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386808037029763e-05,
+      "loss": 0.4334,
+      "step": 2388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8385272362373775e-05,
+      "loss": 0.4523,
+      "step": 2389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8383736021318217e-05,
+      "loss": 0.4579,
+      "step": 2390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382199013985184e-05,
+      "loss": 0.4439,
+      "step": 2391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380661340496825e-05,
+      "loss": 0.3785,
+      "step": 2392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379123000975346e-05,
+      "loss": 0.3739,
+      "step": 2393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8377583995543e-05,
+      "loss": 0.3867,
+      "step": 2394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83760443243221e-05,
+      "loss": 0.4322,
+      "step": 2395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374503987435e-05,
+      "loss": 0.416,
+      "step": 2396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372962985004122e-05,
+      "loss": 0.3849,
+      "step": 2397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371421317151933e-05,
+      "loss": 0.3753,
+      "step": 2398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369878984000948e-05,
+      "loss": 0.3835,
+      "step": 2399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368335985673746e-05,
+      "loss": 0.3596,
+      "step": 2400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8366792322292946e-05,
+      "loss": 0.4335,
+      "step": 2401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365247993981238e-05,
+      "loss": 0.3859,
+      "step": 2402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8363703000861344e-05,
+      "loss": 0.4092,
+      "step": 2403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362157343056056e-05,
+      "loss": 0.3825,
+      "step": 2404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8360611020688202e-05,
+      "loss": 0.4245,
+      "step": 2405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835906403388068e-05,
+      "loss": 0.3855,
+      "step": 2406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8357516382756434e-05,
+      "loss": 0.3992,
+      "step": 2407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355968067438453e-05,
+      "loss": 0.3886,
+      "step": 2408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835441908804979e-05,
+      "loss": 0.4107,
+      "step": 2409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352869444713545e-05,
+      "loss": 0.4102,
+      "step": 2410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351319137552876e-05,
+      "loss": 0.3849,
+      "step": 2411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834976816669099e-05,
+      "loss": 0.3943,
+      "step": 2412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834821653225114e-05,
+      "loss": 0.3936,
+      "step": 2413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834666423435664e-05,
+      "loss": 0.4561,
+      "step": 2414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834511127313086e-05,
+      "loss": 0.3925,
+      "step": 2415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343557648697215e-05,
+      "loss": 0.3902,
+      "step": 2416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834200336117918e-05,
+      "loss": 0.4043,
+      "step": 2417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8340448410700268e-05,
+      "loss": 0.3971,
+      "step": 2418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338892797384063e-05,
+      "loss": 0.4103,
+      "step": 2419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337336521354193e-05,
+      "loss": 0.3989,
+      "step": 2420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833577958273434e-05,
+      "loss": 0.3781,
+      "step": 2421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334221981648233e-05,
+      "loss": 0.42,
+      "step": 2422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332663718219667e-05,
+      "loss": 0.4241,
+      "step": 2423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833110479257247e-05,
+      "loss": 0.4236,
+      "step": 2424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329545204830545e-05,
+      "loss": 0.4206,
+      "step": 2425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832798495511783e-05,
+      "loss": 0.3742,
+      "step": 2426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8326424043558324e-05,
+      "loss": 0.4005,
+      "step": 2427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324862470276078e-05,
+      "loss": 0.3928,
+      "step": 2428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832330023539519e-05,
+      "loss": 0.3636,
+      "step": 2429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321737339039817e-05,
+      "loss": 0.3776,
+      "step": 2430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320173781334172e-05,
+      "loss": 0.3706,
+      "step": 2431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318609562402507e-05,
+      "loss": 0.3686,
+      "step": 2432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831704468236914e-05,
+      "loss": 0.4086,
+      "step": 2433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315479141358432e-05,
+      "loss": 0.3784,
+      "step": 2434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313912939494803e-05,
+      "loss": 0.3926,
+      "step": 2435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8312346076902724e-05,
+      "loss": 0.3869,
+      "step": 2436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310778553706712e-05,
+      "loss": 0.3809,
+      "step": 2437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309210370031356e-05,
+      "loss": 0.3724,
+      "step": 2438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307641526001263e-05,
+      "loss": 0.381,
+      "step": 2439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830607202174113e-05,
+      "loss": 0.4075,
+      "step": 2440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304501857375684e-05,
+      "loss": 0.3683,
+      "step": 2441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830293103302971e-05,
+      "loss": 0.3703,
+      "step": 2442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301359548828048e-05,
+      "loss": 0.3815,
+      "step": 2443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299787404895584e-05,
+      "loss": 0.4207,
+      "step": 2444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829821460135726e-05,
+      "loss": 0.4152,
+      "step": 2445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296641138338074e-05,
+      "loss": 0.3808,
+      "step": 2446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829506701596307e-05,
+      "loss": 0.4093,
+      "step": 2447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293492234357355e-05,
+      "loss": 0.3824,
+      "step": 2448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829191679364607e-05,
+      "loss": 0.4023,
+      "step": 2449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290340693954432e-05,
+      "loss": 0.4218,
+      "step": 2450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8288763935407685e-05,
+      "loss": 0.3867,
+      "step": 2451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287186518131147e-05,
+      "loss": 0.3621,
+      "step": 2452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8285608442250174e-05,
+      "loss": 0.4389,
+      "step": 2453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284029707890187e-05,
+      "loss": 0.3701,
+      "step": 2454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8282450315176643e-05,
+      "loss": 0.3802,
+      "step": 2455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280870264235066e-05,
+      "loss": 0.4229,
+      "step": 2456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279289555191028e-05,
+      "loss": 0.4152,
+      "step": 2457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277708188170146e-05,
+      "loss": 0.392,
+      "step": 2458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8276126163298103e-05,
+      "loss": 0.3585,
+      "step": 2459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827454348070062e-05,
+      "loss": 0.3892,
+      "step": 2460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827296014050348e-05,
+      "loss": 0.3957,
+      "step": 2461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827137614283252e-05,
+      "loss": 0.4051,
+      "step": 2462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269791487813614e-05,
+      "loss": 0.42,
+      "step": 2463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268206175572703e-05,
+      "loss": 0.3994,
+      "step": 2464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8266620206235782e-05,
+      "loss": 0.4028,
+      "step": 2465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265033579928886e-05,
+      "loss": 0.3781,
+      "step": 2466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826344629677811e-05,
+      "loss": 0.4195,
+      "step": 2467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261858356909598e-05,
+      "loss": 0.3796,
+      "step": 2468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8260269760449548e-05,
+      "loss": 0.3565,
+      "step": 2469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8258680507524212e-05,
+      "loss": 0.4065,
+      "step": 2470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825709059825989e-05,
+      "loss": 0.392,
+      "step": 2471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825550003278294e-05,
+      "loss": 0.3914,
+      "step": 2472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253908811219764e-05,
+      "loss": 0.3895,
+      "step": 2473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825231693369682e-05,
+      "loss": 0.417,
+      "step": 2474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8250724400340622e-05,
+      "loss": 0.3812,
+      "step": 2475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249131211277735e-05,
+      "loss": 0.3921,
+      "step": 2476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247537366634768e-05,
+      "loss": 0.3759,
+      "step": 2477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245942866538387e-05,
+      "loss": 0.347,
+      "step": 2478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824434771111532e-05,
+      "loss": 0.4081,
+      "step": 2479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824275190049233e-05,
+      "loss": 0.4032,
+      "step": 2480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241155434796246e-05,
+      "loss": 0.3581,
+      "step": 2481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239558314153938e-05,
+      "loss": 0.4029,
+      "step": 2482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237960538692335e-05,
+      "loss": 0.4068,
+      "step": 2483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236362108538415e-05,
+      "loss": 0.4268,
+      "step": 2484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234763023819216e-05,
+      "loss": 0.3717,
+      "step": 2485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233163284661816e-05,
+      "loss": 0.3855,
+      "step": 2486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823156289119335e-05,
+      "loss": 0.4037,
+      "step": 2487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822996184354101e-05,
+      "loss": 0.3739,
+      "step": 2488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822836014183203e-05,
+      "loss": 0.3952,
+      "step": 2489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226757786193703e-05,
+      "loss": 0.444,
+      "step": 2490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225154776753377e-05,
+      "loss": 0.3962,
+      "step": 2491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822355111363844e-05,
+      "loss": 0.4071,
+      "step": 2492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8221946796976344e-05,
+      "loss": 0.3938,
+      "step": 2493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220341826894588e-05,
+      "loss": 0.4082,
+      "step": 2494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821873620352072e-05,
+      "loss": 0.4322,
+      "step": 2495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821712992698235e-05,
+      "loss": 0.3817,
+      "step": 2496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215522997407125e-05,
+      "loss": 0.362,
+      "step": 2497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213915414922755e-05,
+      "loss": 0.4044,
+      "step": 2498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212307179657e-05,
+      "loss": 0.364,
+      "step": 2499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8210698291737672e-05,
+      "loss": 0.4183,
+      "step": 2500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820908875129263e-05,
+      "loss": 0.3567,
+      "step": 2501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207478558449786e-05,
+      "loss": 0.388,
+      "step": 2502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205867713337112e-05,
+      "loss": 0.4302,
+      "step": 2503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204256216082625e-05,
+      "loss": 0.4106,
+      "step": 2504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202644066814397e-05,
+      "loss": 0.4159,
+      "step": 2505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8201031265660543e-05,
+      "loss": 0.4408,
+      "step": 2506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8199417812749242e-05,
+      "loss": 0.3758,
+      "step": 2507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197803708208718e-05,
+      "loss": 0.3882,
+      "step": 2508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8196188952167243e-05,
+      "loss": 0.4161,
+      "step": 2509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194573544753156e-05,
+      "loss": 0.4199,
+      "step": 2510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819295748609483e-05,
+      "loss": 0.3894,
+      "step": 2511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8191340776320698e-05,
+      "loss": 0.4029,
+      "step": 2512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189723415559252e-05,
+      "loss": 0.419,
+      "step": 2513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8188105403939015e-05,
+      "loss": 0.4077,
+      "step": 2514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818648674158858e-05,
+      "loss": 0.3656,
+      "step": 2515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184867428636597e-05,
+      "loss": 0.3777,
+      "step": 2516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8183247465211743e-05,
+      "loss": 0.401,
+      "step": 2517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181626851442763e-05,
+      "loss": 0.3823,
+      "step": 2518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8180005587458456e-05,
+      "loss": 0.4228,
+      "step": 2519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178383673387668e-05,
+      "loss": 0.3885,
+      "step": 2520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817676110935929e-05,
+      "loss": 0.4073,
+      "step": 2521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817513789550228e-05,
+      "loss": 0.356,
+      "step": 2522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173514031945634e-05,
+      "loss": 0.3774,
+      "step": 2523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171889518818405e-05,
+      "loss": 0.3888,
+      "step": 2524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170264356249697e-05,
+      "loss": 0.4092,
+      "step": 2525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8168638544368673e-05,
+      "loss": 0.4039,
+      "step": 2526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167012083304532e-05,
+      "loss": 0.3915,
+      "step": 2527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165384973186535e-05,
+      "loss": 0.3884,
+      "step": 2528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163757214143993e-05,
+      "loss": 0.3978,
+      "step": 2529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162128806306268e-05,
+      "loss": 0.3859,
+      "step": 2530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160499749802778e-05,
+      "loss": 0.3673,
+      "step": 2531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158870044762982e-05,
+      "loss": 0.3764,
+      "step": 2532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157239691316402e-05,
+      "loss": 0.3922,
+      "step": 2533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8155608689592604e-05,
+      "loss": 0.3736,
+      "step": 2534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153977039721207e-05,
+      "loss": 0.4016,
+      "step": 2535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152344741831888e-05,
+      "loss": 0.3454,
+      "step": 2536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815071179605436e-05,
+      "loss": 0.3823,
+      "step": 2537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149078202518407e-05,
+      "loss": 0.409,
+      "step": 2538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814744396135385e-05,
+      "loss": 0.4196,
+      "step": 2539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145809072690568e-05,
+      "loss": 0.3853,
+      "step": 2540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814417353665849e-05,
+      "loss": 0.4113,
+      "step": 2541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142537353387596e-05,
+      "loss": 0.3694,
+      "step": 2542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814090052300792e-05,
+      "loss": 0.3716,
+      "step": 2543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813926304564954e-05,
+      "loss": 0.3788,
+      "step": 2544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137624921442592e-05,
+      "loss": 0.3849,
+      "step": 2545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813598615051727e-05,
+      "loss": 0.4242,
+      "step": 2546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81343467330038e-05,
+      "loss": 0.4019,
+      "step": 2547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132706669032477e-05,
+      "loss": 0.3779,
+      "step": 2548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813106595873364e-05,
+      "loss": 0.4189,
+      "step": 2549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8129424602237683e-05,
+      "loss": 0.4062,
+      "step": 2550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127782599675044e-05,
+      "loss": 0.3961,
+      "step": 2551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8126139951176216e-05,
+      "loss": 0.4139,
+      "step": 2552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812449665687175e-05,
+      "loss": 0.3628,
+      "step": 2553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122852716892248e-05,
+      "loss": 0.4405,
+      "step": 2554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121208131368346e-05,
+      "loss": 0.409,
+      "step": 2555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811956290043075e-05,
+      "loss": 0.3807,
+      "step": 2556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811791702421021e-05,
+      "loss": 0.4521,
+      "step": 2557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116270502837527e-05,
+      "loss": 0.4053,
+      "step": 2558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114623336443554e-05,
+      "loss": 0.4241,
+      "step": 2559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81129755251592e-05,
+      "loss": 0.4066,
+      "step": 2560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111327069115416e-05,
+      "loss": 0.4509,
+      "step": 2561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8109677968443212e-05,
+      "loss": 0.4096,
+      "step": 2562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108028223273643e-05,
+      "loss": 0.3911,
+      "step": 2563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106377833737824e-05,
+      "loss": 0.4093,
+      "step": 2564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810472679996691e-05,
+      "loss": 0.3684,
+      "step": 2565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8103075122092116e-05,
+      "loss": 0.3942,
+      "step": 2566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101422800244703e-05,
+      "loss": 0.3913,
+      "step": 2567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809976983455599e-05,
+      "loss": 0.3898,
+      "step": 2568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8098116225157342e-05,
+      "loss": 0.3667,
+      "step": 2569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096461972180166e-05,
+      "loss": 0.4079,
+      "step": 2570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809480707575594e-05,
+      "loss": 0.3949,
+      "step": 2571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093151536016185e-05,
+      "loss": 0.4046,
+      "step": 2572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809149535309246e-05,
+      "loss": 0.3827,
+      "step": 2573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089838527116394e-05,
+      "loss": 0.375,
+      "step": 2574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808818105821966e-05,
+      "loss": 0.3807,
+      "step": 2575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8086522946533976e-05,
+      "loss": 0.3695,
+      "step": 2576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808486419219112e-05,
+      "loss": 0.3829,
+      "step": 2577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083204795322918e-05,
+      "loss": 0.4049,
+      "step": 2578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8081544756061246e-05,
+      "loss": 0.3529,
+      "step": 2579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807988407453803e-05,
+      "loss": 0.3672,
+      "step": 2580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078222750885254e-05,
+      "loss": 0.3968,
+      "step": 2581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8076560785234938e-05,
+      "loss": 0.3765,
+      "step": 2582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807489817771917e-05,
+      "loss": 0.3953,
+      "step": 2583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807323492847008e-05,
+      "loss": 0.3814,
+      "step": 2584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8071571037619856e-05,
+      "loss": 0.403,
+      "step": 2585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069906505300724e-05,
+      "loss": 0.4163,
+      "step": 2586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068241331644973e-05,
+      "loss": 0.4071,
+      "step": 2587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806657551678493e-05,
+      "loss": 0.3722,
+      "step": 2588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064909060852997e-05,
+      "loss": 0.4322,
+      "step": 2589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8063241963981603e-05,
+      "loss": 0.3724,
+      "step": 2590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806157422630323e-05,
+      "loss": 0.3775,
+      "step": 2591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8059905847950433e-05,
+      "loss": 0.3876,
+      "step": 2592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058236829055792e-05,
+      "loss": 0.3812,
+      "step": 2593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056567169751947e-05,
+      "loss": 0.3769,
+      "step": 2594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054896870171598e-05,
+      "loss": 0.3365,
+      "step": 2595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805322593044748e-05,
+      "loss": 0.3944,
+      "step": 2596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051554350712392e-05,
+      "loss": 0.3809,
+      "step": 2597
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804988213109918e-05,
+      "loss": 0.3769,
+      "step": 2598
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048209271740735e-05,
+      "loss": 0.3853,
+      "step": 2599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046535772770008e-05,
+      "loss": 0.3767,
+      "step": 2600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804486163431999e-05,
+      "loss": 0.3979,
+      "step": 2601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043186856523737e-05,
+      "loss": 0.4612,
+      "step": 2602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041511439514338e-05,
+      "loss": 0.3921,
+      "step": 2603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8039835383424953e-05,
+      "loss": 0.4005,
+      "step": 2604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803815868838878e-05,
+      "loss": 0.3965,
+      "step": 2605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036481354539064e-05,
+      "loss": 0.3848,
+      "step": 2606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803480338200912e-05,
+      "loss": 0.3923,
+      "step": 2607
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033124770932286e-05,
+      "loss": 0.3854,
+      "step": 2608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803144552144197e-05,
+      "loss": 0.3893,
+      "step": 2609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802976563367164e-05,
+      "loss": 0.3759,
+      "step": 2610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802808510775478e-05,
+      "loss": 0.4063,
+      "step": 2611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802640394382496e-05,
+      "loss": 0.3735,
+      "step": 2612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024722142015784e-05,
+      "loss": 0.3872,
+      "step": 2613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80230397024609e-05,
+      "loss": 0.3954,
+      "step": 2614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802135662529403e-05,
+      "loss": 0.4425,
+      "step": 2615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8019672910648923e-05,
+      "loss": 0.3942,
+      "step": 2616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801798855865939e-05,
+      "loss": 0.4014,
+      "step": 2617
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016303569459295e-05,
+      "loss": 0.4072,
+      "step": 2618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014617943182546e-05,
+      "loss": 0.3818,
+      "step": 2619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80129316799631e-05,
+      "loss": 0.4036,
+      "step": 2620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011244779934977e-05,
+      "loss": 0.3985,
+      "step": 2621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800955724323223e-05,
+      "loss": 0.3813,
+      "step": 2622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800786906998898e-05,
+      "loss": 0.3699,
+      "step": 2623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8006180260339384e-05,
+      "loss": 0.4058,
+      "step": 2624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004490814417662e-05,
+      "loss": 0.3805,
+      "step": 2625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8002800732358073e-05,
+      "loss": 0.4269,
+      "step": 2626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001110014294938e-05,
+      "loss": 0.388,
+      "step": 2627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799941866036262e-05,
+      "loss": 0.4113,
+      "step": 2628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997726670695534e-05,
+      "loss": 0.4267,
+      "step": 2629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799603404542815e-05,
+      "loss": 0.3959,
+      "step": 2630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994340784694982e-05,
+      "loss": 0.421,
+      "step": 2631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79926468886306e-05,
+      "loss": 0.4056,
+      "step": 2632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799095235736962e-05,
+      "loss": 0.3716,
+      "step": 2633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7989257191046716e-05,
+      "loss": 0.3819,
+      "step": 2634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987561389796604e-05,
+      "loss": 0.3977,
+      "step": 2635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7985864953754052e-05,
+      "loss": 0.3969,
+      "step": 2636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984167883053883e-05,
+      "loss": 0.4238,
+      "step": 2637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7982470177830966e-05,
+      "loss": 0.3954,
+      "step": 2638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980771838220223e-05,
+      "loss": 0.4184,
+      "step": 2639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7979072864356623e-05,
+      "loss": 0.3711,
+      "step": 2640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977373256375194e-05,
+      "loss": 0.3674,
+      "step": 2641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975673014411003e-05,
+      "loss": 0.4327,
+      "step": 2642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7973972138599176e-05,
+      "loss": 0.3968,
+      "step": 2643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972270629074882e-05,
+      "loss": 0.362,
+      "step": 2644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7970568485973345e-05,
+      "loss": 0.3841,
+      "step": 2645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7968865709429844e-05,
+      "loss": 0.3868,
+      "step": 2646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967162299579696e-05,
+      "loss": 0.4091,
+      "step": 2647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796545825655828e-05,
+      "loss": 0.3646,
+      "step": 2648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796375358050102e-05,
+      "loss": 0.4081,
+      "step": 2649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962048271543394e-05,
+      "loss": 0.3625,
+      "step": 2650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7960342329820918e-05,
+      "loss": 0.3908,
+      "step": 2651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958635755469177e-05,
+      "loss": 0.3749,
+      "step": 2652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7956928548623796e-05,
+      "loss": 0.3776,
+      "step": 2653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7955220709420446e-05,
+      "loss": 0.3953,
+      "step": 2654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795351223799486e-05,
+      "loss": 0.4184,
+      "step": 2655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7951803134482806e-05,
+      "loss": 0.4017,
+      "step": 2656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795009339902012e-05,
+      "loss": 0.436,
+      "step": 2657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794838303174267e-05,
+      "loss": 0.3633,
+      "step": 2658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7946672032786395e-05,
+      "loss": 0.3739,
+      "step": 2659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944960402287262e-05,
+      "loss": 0.3422,
+      "step": 2660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943248140381304e-05,
+      "loss": 0.3722,
+      "step": 2661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941535247204597e-05,
+      "loss": 0.4034,
+      "step": 2662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793982172289327e-05,
+      "loss": 0.3872,
+      "step": 2663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7938107567583506e-05,
+      "loss": 0.3893,
+      "step": 2664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7936392781411524e-05,
+      "loss": 0.4021,
+      "step": 2665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934677364513605e-05,
+      "loss": 0.4146,
+      "step": 2666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932961317026085e-05,
+      "loss": 0.3796,
+      "step": 2667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7931244639085338e-05,
+      "loss": 0.4072,
+      "step": 2668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929527330827788e-05,
+      "loss": 0.3805,
+      "step": 2669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792780939238992e-05,
+      "loss": 0.3899,
+      "step": 2670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926090823908264e-05,
+      "loss": 0.3702,
+      "step": 2671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7924371625519396e-05,
+      "loss": 0.4048,
+      "step": 2672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7922651797359943e-05,
+      "loss": 0.3897,
+      "step": 2673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792093133956659e-05,
+      "loss": 0.4134,
+      "step": 2674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791921025227606e-05,
+      "loss": 0.3991,
+      "step": 2675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791748853562514e-05,
+      "loss": 0.4147,
+      "step": 2676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915766189750656e-05,
+      "loss": 0.3954,
+      "step": 2677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7914043214789482e-05,
+      "loss": 0.3699,
+      "step": 2678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912319610878555e-05,
+      "loss": 0.3946,
+      "step": 2679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910595378154848e-05,
+      "loss": 0.3969,
+      "step": 2680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908870516755396e-05,
+      "loss": 0.4017,
+      "step": 2681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790714502681728e-05,
+      "loss": 0.3777,
+      "step": 2682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905418908477617e-05,
+      "loss": 0.3678,
+      "step": 2683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7903692161873598e-05,
+      "loss": 0.4084,
+      "step": 2684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901964787142447e-05,
+      "loss": 0.4205,
+      "step": 2685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900236784421444e-05,
+      "loss": 0.4,
+      "step": 2686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789850815384792e-05,
+      "loss": 0.3792,
+      "step": 2687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789677889555925e-05,
+      "loss": 0.3754,
+      "step": 2688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895049009692864e-05,
+      "loss": 0.3962,
+      "step": 2689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893318496386243e-05,
+      "loss": 0.3575,
+      "step": 2690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7891587355776912e-05,
+      "loss": 0.3997,
+      "step": 2691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7889855588002452e-05,
+      "loss": 0.4086,
+      "step": 2692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788812319320049e-05,
+      "loss": 0.3866,
+      "step": 2693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7886390171508703e-05,
+      "loss": 0.3716,
+      "step": 2694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7884656523064822e-05,
+      "loss": 0.3971,
+      "step": 2695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7882922248006617e-05,
+      "loss": 0.356,
+      "step": 2696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881187346471924e-05,
+      "loss": 0.3895,
+      "step": 2697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879451818598617e-05,
+      "loss": 0.3841,
+      "step": 2698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7877715664524623e-05,
+      "loss": 0.385,
+      "step": 2699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787597888438792e-05,
+      "loss": 0.3354,
+      "step": 2700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787424147832653e-05,
+      "loss": 0.4321,
+      "step": 2701
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7872503446478534e-05,
+      "loss": 0.4126,
+      "step": 2702
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7870764788982052e-05,
+      "loss": 0.4111,
+      "step": 2703
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786902550597527e-05,
+      "loss": 0.4259,
+      "step": 2704
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78672855975964e-05,
+      "loss": 0.4043,
+      "step": 2705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786554506398373e-05,
+      "loss": 0.362,
+      "step": 2706
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7863803905275575e-05,
+      "loss": 0.4136,
+      "step": 2707
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786206212161032e-05,
+      "loss": 0.4382,
+      "step": 2708
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860319713126374e-05,
+      "loss": 0.3515,
+      "step": 2709
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7858576679962224e-05,
+      "loss": 0.4055,
+      "step": 2710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785683302225639e-05,
+      "loss": 0.388,
+      "step": 2711
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785508874014744e-05,
+      "loss": 0.4045,
+      "step": 2712
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7853343833774003e-05,
+      "loss": 0.3992,
+      "step": 2713
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7851598303274748e-05,
+      "loss": 0.352,
+      "step": 2714
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7849852148788397e-05,
+      "loss": 0.4144,
+      "step": 2715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7848105370453722e-05,
+      "loss": 0.4035,
+      "step": 2716
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7846357968409548e-05,
+      "loss": 0.391,
+      "step": 2717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844609942794738e-05,
+      "loss": 0.3972,
+      "step": 2718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7842861293748216e-05,
+      "loss": 0.3776,
+      "step": 2719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784111202140895e-05,
+      "loss": 0.3883,
+      "step": 2720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7839362125915967e-05,
+      "loss": 0.4219,
+      "step": 2721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837611607408326e-05,
+      "loss": 0.376,
+      "step": 2722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835860466025148e-05,
+      "loss": 0.3968,
+      "step": 2723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7834108701905603e-05,
+      "loss": 0.3882,
+      "step": 2724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832356315188907e-05,
+      "loss": 0.4125,
+      "step": 2725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7830603306014325e-05,
+      "loss": 0.3531,
+      "step": 2726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828849674521178e-05,
+      "loss": 0.3563,
+      "step": 2727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827095420848825e-05,
+      "loss": 0.3698,
+      "step": 2728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782534054513669e-05,
+      "loss": 0.3853,
+      "step": 2729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782358504752423e-05,
+      "loss": 0.3552,
+      "step": 2730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782182892815096e-05,
+      "loss": 0.3752,
+      "step": 2731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7820072187156445e-05,
+      "loss": 0.3743,
+      "step": 2732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 0.3943,
+      "step": 2733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781655684086218e-05,
+      "loss": 0.3707,
+      "step": 2734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7814798235841808e-05,
+      "loss": 0.3706,
+      "step": 2735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7813039009758938e-05,
+      "loss": 0.3804,
+      "step": 2736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811279162753375e-05,
+      "loss": 0.3776,
+      "step": 2737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780951869496499e-05,
+      "loss": 0.4301,
+      "step": 2738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780775760653368e-05,
+      "loss": 0.392,
+      "step": 2739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805995897599417e-05,
+      "loss": 0.3815,
+      "step": 2740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78042335683022e-05,
+      "loss": 0.4061,
+      "step": 2741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7802470618782083e-05,
+      "loss": 0.4187,
+      "step": 2742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800707049179176e-05,
+      "loss": 0.3975,
+      "step": 2743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779894285963364e-05,
+      "loss": 0.3736,
+      "step": 2744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797178050285672e-05,
+      "loss": 0.422,
+      "step": 2745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779541262127553e-05,
+      "loss": 0.4125,
+      "step": 2746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793646572743517e-05,
+      "loss": 0.3777,
+      "step": 2747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7791879904829984e-05,
+      "loss": 0.381,
+      "step": 2748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779011261767533e-05,
+      "loss": 0.3991,
+      "step": 2749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7788344711420012e-05,
+      "loss": 0.3922,
+      "step": 2750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786576186204524e-05,
+      "loss": 0.3695,
+      "step": 2751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7784807042169422e-05,
+      "loss": 0.448,
+      "step": 2752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77830372794553e-05,
+      "loss": 0.4097,
+      "step": 2753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778126689820281e-05,
+      "loss": 0.3664,
+      "step": 2754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777949589855264e-05,
+      "loss": 0.3888,
+      "step": 2755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7777724280645546e-05,
+      "loss": 0.3714,
+      "step": 2756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777595204462232e-05,
+      "loss": 0.4193,
+      "step": 2757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77741791906238e-05,
+      "loss": 0.3719,
+      "step": 2758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7772405718790885e-05,
+      "loss": 0.4054,
+      "step": 2759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777063162926452e-05,
+      "loss": 0.391,
+      "step": 2760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7768856922185693e-05,
+      "loss": 0.3976,
+      "step": 2761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767081597695446e-05,
+      "loss": 0.3859,
+      "step": 2762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7765305655934865e-05,
+      "loss": 0.3458,
+      "step": 2763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7763529097045093e-05,
+      "loss": 0.4183,
+      "step": 2764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7761751921167313e-05,
+      "loss": 0.3833,
+      "step": 2765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775997412844277e-05,
+      "loss": 0.3704,
+      "step": 2766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758195719012742e-05,
+      "loss": 0.3626,
+      "step": 2767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756416693018568e-05,
+      "loss": 0.3913,
+      "step": 2768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775463705060163e-05,
+      "loss": 0.3995,
+      "step": 2769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775285679190336e-05,
+      "loss": 0.3844,
+      "step": 2770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751075917065244e-05,
+      "loss": 0.3867,
+      "step": 2771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774929442622881e-05,
+      "loss": 0.3721,
+      "step": 2772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7747512319535636e-05,
+      "loss": 0.3976,
+      "step": 2773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7745729597127353e-05,
+      "loss": 0.3753,
+      "step": 2774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743946259145638e-05,
+      "loss": 0.397,
+      "step": 2775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774216230573222e-05,
+      "loss": 0.4047,
+      "step": 2776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740377737028867e-05,
+      "loss": 0.3777,
+      "step": 2777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738592553177416e-05,
+      "loss": 0.43,
+      "step": 2778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7736806754319725e-05,
+      "loss": 0.3743,
+      "step": 2779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773502034059773e-05,
+      "loss": 0.372,
+      "step": 2780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773323331215339e-05,
+      "loss": 0.3896,
+      "step": 2781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731445669128734e-05,
+      "loss": 0.3757,
+      "step": 2782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729657411665832e-05,
+      "loss": 0.3611,
+      "step": 2783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772786853990679e-05,
+      "loss": 0.394,
+      "step": 2784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7726079053993782e-05,
+      "loss": 0.4034,
+      "step": 2785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7724288954069027e-05,
+      "loss": 0.3916,
+      "step": 2786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772249824027478e-05,
+      "loss": 0.377,
+      "step": 2787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720706912753357e-05,
+      "loss": 0.4651,
+      "step": 2788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7718914971647122e-05,
+      "loss": 0.4051,
+      "step": 2789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771712241709849e-05,
+      "loss": 0.3726,
+      "step": 2790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7715329249249903e-05,
+      "loss": 0.4043,
+      "step": 2791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713535468243884e-05,
+      "loss": 0.4087,
+      "step": 2792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711741074222987e-05,
+      "loss": 0.3762,
+      "step": 2793
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7709946067329813e-05,
+      "loss": 0.3577,
+      "step": 2794
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7708150447707017e-05,
+      "loss": 0.3461,
+      "step": 2795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7706354215497306e-05,
+      "loss": 0.429,
+      "step": 2796
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7704557370843427e-05,
+      "loss": 0.3829,
+      "step": 2797
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702759913888177e-05,
+      "loss": 0.4046,
+      "step": 2798
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7700961844774417e-05,
+      "loss": 0.4026,
+      "step": 2799
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7699163163645028e-05,
+      "loss": 0.3368,
+      "step": 2800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7697363870642967e-05,
+      "loss": 0.3782,
+      "step": 2801
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7695563965911225e-05,
+      "loss": 0.3594,
+      "step": 2802
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693763449592842e-05,
+      "loss": 0.3986,
+      "step": 2803
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691962321830916e-05,
+      "loss": 0.3796,
+      "step": 2804
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690160582768584e-05,
+      "loss": 0.3786,
+      "step": 2805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688358232549037e-05,
+      "loss": 0.3518,
+      "step": 2806
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768655527131551e-05,
+      "loss": 0.3989,
+      "step": 2807
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7684751699211288e-05,
+      "loss": 0.3646,
+      "step": 2808
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682947516379706e-05,
+      "loss": 0.354,
+      "step": 2809
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7681142722964156e-05,
+      "loss": 0.388,
+      "step": 2810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679337319108055e-05,
+      "loss": 0.378,
+      "step": 2811
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7677531304954896e-05,
+      "loss": 0.3982,
+      "step": 2812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.76757246806482e-05,
+      "loss": 0.3884,
+      "step": 2813
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7673917446331543e-05,
+      "loss": 0.3889,
+      "step": 2814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7672109602148557e-05,
+      "loss": 0.3674,
+      "step": 2815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767030114824291e-05,
+      "loss": 0.3912,
+      "step": 2816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668492084758335e-05,
+      "loss": 0.3827,
+      "step": 2817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766668241183859e-05,
+      "loss": 0.3857,
+      "step": 2818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7664872129627504e-05,
+      "loss": 0.3545,
+      "step": 2819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766306123826894e-05,
+      "loss": 0.3754,
+      "step": 2820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7661249737906813e-05,
+      "loss": 0.4022,
+      "step": 2821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765943762868509e-05,
+      "loss": 0.4018,
+      "step": 2822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7657624910747783e-05,
+      "loss": 0.401,
+      "step": 2823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655811584238954e-05,
+      "loss": 0.3428,
+      "step": 2824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765399764930272e-05,
+      "loss": 0.3975,
+      "step": 2825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7652183106083227e-05,
+      "loss": 0.419,
+      "step": 2826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765036795472469e-05,
+      "loss": 0.3993,
+      "step": 2827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764855219537136e-05,
+      "loss": 0.3906,
+      "step": 2828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764673582816754e-05,
+      "loss": 0.3914,
+      "step": 2829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764491885325758e-05,
+      "loss": 0.3737,
+      "step": 2830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7643101270785883e-05,
+      "loss": 0.368,
+      "step": 2831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7641283080896898e-05,
+      "loss": 0.4002,
+      "step": 2832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763946428373512e-05,
+      "loss": 0.3889,
+      "step": 2833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7637644879445092e-05,
+      "loss": 0.3862,
+      "step": 2834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763582486817141e-05,
+      "loss": 0.4153,
+      "step": 2835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763400425005871e-05,
+      "loss": 0.4209,
+      "step": 2836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763218302525169e-05,
+      "loss": 0.3376,
+      "step": 2837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7630361193895074e-05,
+      "loss": 0.4148,
+      "step": 2838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7628538756133657e-05,
+      "loss": 0.3513,
+      "step": 2839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7626715712112277e-05,
+      "loss": 0.355,
+      "step": 2840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7624892061975808e-05,
+      "loss": 0.4518,
+      "step": 2841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762306780586918e-05,
+      "loss": 0.3956,
+      "step": 2842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7621242943937377e-05,
+      "loss": 0.3884,
+      "step": 2843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761941747632542e-05,
+      "loss": 0.3934,
+      "step": 2844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761759140317839e-05,
+      "loss": 0.423,
+      "step": 2845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7615764724641406e-05,
+      "loss": 0.3815,
+      "step": 2846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613937440859638e-05,
+      "loss": 0.3772,
+      "step": 2847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7612109551978305e-05,
+      "loss": 0.4058,
+      "step": 2848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7610281058142674e-05,
+      "loss": 0.4234,
+      "step": 2849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608451959498064e-05,
+      "loss": 0.3922,
+      "step": 2850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7606622256189837e-05,
+      "loss": 0.4078,
+      "step": 2851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7604791948363404e-05,
+      "loss": 0.3562,
+      "step": 2852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7602961036164218e-05,
+      "loss": 0.3731,
+      "step": 2853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7601129519737796e-05,
+      "loss": 0.3873,
+      "step": 2854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759929739922969e-05,
+      "loss": 0.3891,
+      "step": 2855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75974646747855e-05,
+      "loss": 0.3977,
+      "step": 2856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595631346550882e-05,
+      "loss": 0.4198,
+      "step": 2857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759379741467153e-05,
+      "loss": 0.37,
+      "step": 2858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75919628792932e-05,
+      "loss": 0.435,
+      "step": 2859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759012774056168e-05,
+      "loss": 0.3996,
+      "step": 2860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7588291998622816e-05,
+      "loss": 0.3893,
+      "step": 2861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75864556536225e-05,
+      "loss": 0.3962,
+      "step": 2862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7584618705706666e-05,
+      "loss": 0.4149,
+      "step": 2863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758278115502131e-05,
+      "loss": 0.3762,
+      "step": 2864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580943001712457e-05,
+      "loss": 0.3685,
+      "step": 2865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579104245926198e-05,
+      "loss": 0.3444,
+      "step": 2866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757726488780866e-05,
+      "loss": 0.4217,
+      "step": 2867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575424927506017e-05,
+      "loss": 0.3577,
+      "step": 2868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7573584365164506e-05,
+      "loss": 0.3822,
+      "step": 2869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75717432009304e-05,
+      "loss": 0.3935,
+      "step": 2870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756990143495001e-05,
+      "loss": 0.3799,
+      "step": 2871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568059067369714e-05,
+      "loss": 0.3667,
+      "step": 2872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756621609833593e-05,
+      "loss": 0.3447,
+      "step": 2873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564372527995122e-05,
+      "loss": 0.3643,
+      "step": 2874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7562528356493806e-05,
+      "loss": 0.3685,
+      "step": 2875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560683583978537e-05,
+      "loss": 0.3677,
+      "step": 2876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755883821059593e-05,
+      "loss": 0.3964,
+      "step": 2877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556992236492642e-05,
+      "loss": 0.386,
+      "step": 2878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755514566181537e-05,
+      "loss": 0.3551,
+      "step": 2879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553298486710876e-05,
+      "loss": 0.3785,
+      "step": 2880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755145071132595e-05,
+      "loss": 0.3811,
+      "step": 2881
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7549602335807446e-05,
+      "loss": 0.3873,
+      "step": 2882
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754775336030226e-05,
+      "loss": 0.3907,
+      "step": 2883
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754590378495733e-05,
+      "loss": 0.4019,
+      "step": 2884
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544053609919648e-05,
+      "loss": 0.4003,
+      "step": 2885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542202835336254e-05,
+      "loss": 0.3704,
+      "step": 2886
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7540351461354237e-05,
+      "loss": 0.4176,
+      "step": 2887
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7538499488120722e-05,
+      "loss": 0.4014,
+      "step": 2888
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75366469157829e-05,
+      "loss": 0.3626,
+      "step": 2889
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534793744487992e-05,
+      "loss": 0.4021,
+      "step": 2890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753293997438328e-05,
+      "loss": 0.3593,
+      "step": 2891
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753108560561608e-05,
+      "loss": 0.3934,
+      "step": 2892
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752923063833377e-05,
+      "loss": 0.4161,
+      "step": 2893
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752737507268377e-05,
+      "loss": 0.3732,
+      "step": 2894
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525518908813544e-05,
+      "loss": 0.3532,
+      "step": 2895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752366214687061e-05,
+      "loss": 0.4037,
+      "step": 2896
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752180478700252e-05,
+      "loss": 0.4368,
+      "step": 2897
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751994682935689e-05,
+      "loss": 0.4038,
+      "step": 2898
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751808827408138e-05,
+      "loss": 0.3738,
+      "step": 2899
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7516229121323693e-05,
+      "loss": 0.4047,
+      "step": 2900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7514369371231574e-05,
+      "loss": 0.3761,
+      "step": 2901
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7512509023952828e-05,
+      "loss": 0.3974,
+      "step": 2902
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7510648079635303e-05,
+      "loss": 0.3538,
+      "step": 2903
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7508786538426887e-05,
+      "loss": 0.4047,
+      "step": 2904
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506924400475526e-05,
+      "loss": 0.3473,
+      "step": 2905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505061665929205e-05,
+      "loss": 0.4332,
+      "step": 2906
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750319833493597e-05,
+      "loss": 0.3681,
+      "step": 2907
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7501334407643894e-05,
+      "loss": 0.4279,
+      "step": 2908
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749946988420111e-05,
+      "loss": 0.4013,
+      "step": 2909
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497604764755807e-05,
+      "loss": 0.4034,
+      "step": 2910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7495739049456196e-05,
+      "loss": 0.3721,
+      "step": 2911
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749387273845056e-05,
+      "loss": 0.4046,
+      "step": 2912
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492005831887217e-05,
+      "loss": 0.3904,
+      "step": 2913
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749013832991453e-05,
+      "loss": 0.434,
+      "step": 2914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488270232680924e-05,
+      "loss": 0.3771,
+      "step": 2915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7486401540334855e-05,
+      "loss": 0.3941,
+      "step": 2916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484532253024834e-05,
+      "loss": 0.3864,
+      "step": 2917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748266237089942e-05,
+      "loss": 0.4011,
+      "step": 2918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7480791894107216e-05,
+      "loss": 0.3918,
+      "step": 2919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7478920822796876e-05,
+      "loss": 0.409,
+      "step": 2920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477049157117093e-05,
+      "loss": 0.3936,
+      "step": 2921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747517689721662e-05,
+      "loss": 0.3838,
+      "step": 2922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473304043244242e-05,
+      "loss": 0.3933,
+      "step": 2923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7471430595348812e-05,
+      "loss": 0.4027,
+      "step": 2924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7469556553679203e-05,
+      "loss": 0.3698,
+      "step": 2925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467681918384364e-05,
+      "loss": 0.3761,
+      "step": 2926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7465806689613268e-05,
+      "loss": 0.4017,
+      "step": 2927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746393086751495e-05,
+      "loss": 0.4188,
+      "step": 2928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746205445223848e-05,
+      "loss": 0.3625,
+      "step": 2929
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.746017744393299e-05,
+      "loss": 0.3723,
+      "step": 2930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745829984274764e-05,
+      "loss": 0.3966,
+      "step": 2931
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7456421648831658e-05,
+      "loss": 0.4005,
+      "step": 2932
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454542862334305e-05,
+      "loss": 0.3948,
+      "step": 2933
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7452663483404893e-05,
+      "loss": 0.3611,
+      "step": 2934
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745078351219278e-05,
+      "loss": 0.386,
+      "step": 2935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7448902948847375e-05,
+      "loss": 0.3446,
+      "step": 2936
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744702179351813e-05,
+      "loss": 0.3566,
+      "step": 2937
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744514004635454e-05,
+      "loss": 0.3717,
+      "step": 2938
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744325770750617e-05,
+      "loss": 0.4002,
+      "step": 2939
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744137477712259e-05,
+      "loss": 0.3978,
+      "step": 2940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7439491255353457e-05,
+      "loss": 0.3961,
+      "step": 2941
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437607142348457e-05,
+      "loss": 0.3862,
+      "step": 2942
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7435722438257324e-05,
+      "loss": 0.4019,
+      "step": 2943
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743383714322984e-05,
+      "loss": 0.3499,
+      "step": 2944
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7431951257415835e-05,
+      "loss": 0.4172,
+      "step": 2945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430064780965183e-05,
+      "loss": 0.3808,
+      "step": 2946
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428177714027813e-05,
+      "loss": 0.335,
+      "step": 2947
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742629005675369e-05,
+      "loss": 0.388,
+      "step": 2948
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7424401809292833e-05,
+      "loss": 0.4253,
+      "step": 2949
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7422512971795304e-05,
+      "loss": 0.3995,
+      "step": 2950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7420623544411216e-05,
+      "loss": 0.356,
+      "step": 2951
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741873352729073e-05,
+      "loss": 0.3889,
+      "step": 2952
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7416842920584044e-05,
+      "loss": 0.416,
+      "step": 2953
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414951724441413e-05,
+      "loss": 0.3536,
+      "step": 2954
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741305993901313e-05,
+      "loss": 0.3972,
+      "step": 2955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411167564449552e-05,
+      "loss": 0.4122,
+      "step": 2956
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740927460090106e-05,
+      "loss": 0.386,
+      "step": 2957
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7407381048518098e-05,
+      "loss": 0.3681,
+      "step": 2958
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740548690745115e-05,
+      "loss": 0.387,
+      "step": 2959
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7403592177850748e-05,
+      "loss": 0.3378,
+      "step": 2960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740169685986747e-05,
+      "loss": 0.3644,
+      "step": 2961
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399800953651945e-05,
+      "loss": 0.3741,
+      "step": 2962
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7397904459354843e-05,
+      "loss": 0.393,
+      "step": 2963
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7396007377126888e-05,
+      "loss": 0.3714,
+      "step": 2964
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7394109707118837e-05,
+      "loss": 0.3633,
+      "step": 2965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392211449481512e-05,
+      "loss": 0.3926,
+      "step": 2966
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7390312604365768e-05,
+      "loss": 0.3669,
+      "step": 2967
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738841317192251e-05,
+      "loss": 0.3978,
+      "step": 2968
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386513152302696e-05,
+      "loss": 0.3759,
+      "step": 2969
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384612545657323e-05,
+      "loss": 0.372,
+      "step": 2970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7382711352137436e-05,
+      "loss": 0.3787,
+      "step": 2971
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7380809571894128e-05,
+      "loss": 0.3523,
+      "step": 2972
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7378907205078538e-05,
+      "loss": 0.3672,
+      "step": 2973
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7377004251841853e-05,
+      "loss": 0.3507,
+      "step": 2974
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7375100712335307e-05,
+      "loss": 0.381,
+      "step": 2975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737319658671018e-05,
+      "loss": 0.3902,
+      "step": 2976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737129187511779e-05,
+      "loss": 0.4044,
+      "step": 2977
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736938657770952e-05,
+      "loss": 0.3702,
+      "step": 2978
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7367480694636783e-05,
+      "loss": 0.3969,
+      "step": 2979
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7365574226051048e-05,
+      "loss": 0.3611,
+      "step": 2980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736366717210382e-05,
+      "loss": 0.4094,
+      "step": 2981
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7361759532946667e-05,
+      "loss": 0.3863,
+      "step": 2982
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7359851308731186e-05,
+      "loss": 0.3789,
+      "step": 2983
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7357942499609035e-05,
+      "loss": 0.3865,
+      "step": 2984
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7356033105731906e-05,
+      "loss": 0.4108,
+      "step": 2985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735412312725155e-05,
+      "loss": 0.4155,
+      "step": 2986
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352212564319752e-05,
+      "loss": 0.4074,
+      "step": 2987
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350301417088356e-05,
+      "loss": 0.4079,
+      "step": 2988
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734838968570924e-05,
+      "loss": 0.3935,
+      "step": 2989
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346477370334332e-05,
+      "loss": 0.423,
+      "step": 2990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734456447111562e-05,
+      "loss": 0.3854,
+      "step": 2991
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7342650988205115e-05,
+      "loss": 0.391,
+      "step": 2992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340736921754894e-05,
+      "loss": 0.3911,
+      "step": 2993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733882227191707e-05,
+      "loss": 0.3934,
+      "step": 2994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336907038843805e-05,
+      "loss": 0.3708,
+      "step": 2995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733499122268731e-05,
+      "loss": 0.3905,
+      "step": 2996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733307482359984e-05,
+      "loss": 0.411,
+      "step": 2997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733115784173369e-05,
+      "loss": 0.4089,
+      "step": 2998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329240277241215e-05,
+      "loss": 0.3745,
+      "step": 2999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7327322130274805e-05,
+      "loss": 0.3806,
+      "step": 3000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325403400986898e-05,
+      "loss": 0.3604,
+      "step": 3001
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7323484089529987e-05,
+      "loss": 0.42,
+      "step": 3002
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7321564196056603e-05,
+      "loss": 0.3617,
+      "step": 3003
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319643720719322e-05,
+      "loss": 0.3877,
+      "step": 3004
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7317722663670767e-05,
+      "loss": 0.3998,
+      "step": 3005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315801025063617e-05,
+      "loss": 0.4088,
+      "step": 3006
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313878805050586e-05,
+      "loss": 0.4002,
+      "step": 3007
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7311956003784437e-05,
+      "loss": 0.386,
+      "step": 3008
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731003262141798e-05,
+      "loss": 0.3722,
+      "step": 3009
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730810865810407e-05,
+      "loss": 0.3674,
+      "step": 3010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730618411399561e-05,
+      "loss": 0.3771,
+      "step": 3011
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7304258989245552e-05,
+      "loss": 0.3836,
+      "step": 3012
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730233328400689e-05,
+      "loss": 0.3315,
+      "step": 3013
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730040699843266e-05,
+      "loss": 0.3894,
+      "step": 3014
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7298480132675953e-05,
+      "loss": 0.3551,
+      "step": 3015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7296552686889905e-05,
+      "loss": 0.4057,
+      "step": 3016
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729462466122769e-05,
+      "loss": 0.4136,
+      "step": 3017
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729269605584253e-05,
+      "loss": 0.3511,
+      "step": 3018
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7290766870887707e-05,
+      "loss": 0.3685,
+      "step": 3019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728883710651653e-05,
+      "loss": 0.4106,
+      "step": 3020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7286906762882364e-05,
+      "loss": 0.3781,
+      "step": 3021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7284975840138617e-05,
+      "loss": 0.3884,
+      "step": 3022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283044338438754e-05,
+      "loss": 0.3925,
+      "step": 3023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7281112257936263e-05,
+      "loss": 0.3671,
+      "step": 3024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72791795987847e-05,
+      "loss": 0.3942,
+      "step": 3025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7277246361137658e-05,
+      "loss": 0.3939,
+      "step": 3026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727531254514877e-05,
+      "loss": 0.3823,
+      "step": 3027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727337815097173e-05,
+      "loss": 0.3765,
+      "step": 3028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7271443178760265e-05,
+      "loss": 0.4067,
+      "step": 3029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7269507628668152e-05,
+      "loss": 0.4119,
+      "step": 3030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7267571500849217e-05,
+      "loss": 0.3529,
+      "step": 3031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7265634795457327e-05,
+      "loss": 0.3722,
+      "step": 3032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7263697512646397e-05,
+      "loss": 0.4039,
+      "step": 3033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7261759652570384e-05,
+      "loss": 0.3641,
+      "step": 3034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7259821215383302e-05,
+      "loss": 0.3605,
+      "step": 3035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257882201239206e-05,
+      "loss": 0.4244,
+      "step": 3036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7255942610292182e-05,
+      "loss": 0.3887,
+      "step": 3037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7254002442696386e-05,
+      "loss": 0.3436,
+      "step": 3038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7252061698606002e-05,
+      "loss": 0.3775,
+      "step": 3039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7250120378175272e-05,
+      "loss": 0.3957,
+      "step": 3040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7248178481558466e-05,
+      "loss": 0.348,
+      "step": 3041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7246236008909925e-05,
+      "loss": 0.3795,
+      "step": 3042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7244292960384013e-05,
+      "loss": 0.4067,
+      "step": 3043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7242349336135155e-05,
+      "loss": 0.3792,
+      "step": 3044
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.724040513631782e-05,
+      "loss": 0.3809,
+      "step": 3045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72384603610865e-05,
+      "loss": 0.425,
+      "step": 3046
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236515010595773e-05,
+      "loss": 0.387,
+      "step": 3047
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723456908500023e-05,
+      "loss": 0.3779,
+      "step": 3048
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7232622584454518e-05,
+      "loss": 0.3849,
+      "step": 3049
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230675509113336e-05,
+      "loss": 0.4191,
+      "step": 3050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722872785913142e-05,
+      "loss": 0.3954,
+      "step": 3051
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226779634663554e-05,
+      "loss": 0.3651,
+      "step": 3052
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722483083586457e-05,
+      "loss": 0.4135,
+      "step": 3053
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7222881462889345e-05,
+      "loss": 0.4193,
+      "step": 3054
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72209315158928e-05,
+      "loss": 0.3957,
+      "step": 3055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7218980995029902e-05,
+      "loss": 0.3801,
+      "step": 3056
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721702990045566e-05,
+      "loss": 0.3817,
+      "step": 3057
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7215078232325142e-05,
+      "loss": 0.3905,
+      "step": 3058
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213125990793447e-05,
+      "loss": 0.4238,
+      "step": 3059
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721117317601572e-05,
+      "loss": 0.376,
+      "step": 3060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7209219788147166e-05,
+      "loss": 0.4321,
+      "step": 3061
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7207265827343017e-05,
+      "loss": 0.359,
+      "step": 3062
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7205311293758568e-05,
+      "loss": 0.3873,
+      "step": 3063
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720335618754914e-05,
+      "loss": 0.3993,
+      "step": 3064
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720140050887012e-05,
+      "loss": 0.3398,
+      "step": 3065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199444257876928e-05,
+      "loss": 0.3662,
+      "step": 3066
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719748743472503e-05,
+      "loss": 0.3998,
+      "step": 3067
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7195530039569944e-05,
+      "loss": 0.4067,
+      "step": 3068
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193572072567224e-05,
+      "loss": 0.3928,
+      "step": 3069
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719161353387248e-05,
+      "loss": 0.3651,
+      "step": 3070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718965442364136e-05,
+      "loss": 0.3786,
+      "step": 3071
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7187694742029558e-05,
+      "loss": 0.3719,
+      "step": 3072
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7185734489192817e-05,
+      "loss": 0.3484,
+      "step": 3073
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183773665286923e-05,
+      "loss": 0.3709,
+      "step": 3074
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7181812270467708e-05,
+      "loss": 0.3907,
+      "step": 3075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7179850304891052e-05,
+      "loss": 0.3941,
+      "step": 3076
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717788776871287e-05,
+      "loss": 0.3808,
+      "step": 3077
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175924662089136e-05,
+      "loss": 0.407,
+      "step": 3078
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173960985175865e-05,
+      "loss": 0.3624,
+      "step": 3079
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171996738129108e-05,
+      "loss": 0.3867,
+      "step": 3080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7170031921104974e-05,
+      "loss": 0.3775,
+      "step": 3081
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7168066534259614e-05,
+      "loss": 0.3543,
+      "step": 3082
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7166100577749215e-05,
+      "loss": 0.4088,
+      "step": 3083
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7164134051730027e-05,
+      "loss": 0.3872,
+      "step": 3084
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7162166956358328e-05,
+      "loss": 0.3817,
+      "step": 3085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160199291790446e-05,
+      "loss": 0.3521,
+      "step": 3086
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7158231058182764e-05,
+      "loss": 0.3907,
+      "step": 3087
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156262255691698e-05,
+      "loss": 0.3842,
+      "step": 3088
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154292884473712e-05,
+      "loss": 0.4114,
+      "step": 3089
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7152322944685324e-05,
+      "loss": 0.4083,
+      "step": 3090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150352436483087e-05,
+      "loss": 0.4091,
+      "step": 3091
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7148381360023597e-05,
+      "loss": 0.3791,
+      "step": 3092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7146409715463506e-05,
+      "loss": 0.3661,
+      "step": 3093
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7144437502959507e-05,
+      "loss": 0.3911,
+      "step": 3094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142464722668333e-05,
+      "loss": 0.3976,
+      "step": 3095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140491374746768e-05,
+      "loss": 0.3712,
+      "step": 3096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713851745935164e-05,
+      "loss": 0.4055,
+      "step": 3097
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713654297663982e-05,
+      "loss": 0.3774,
+      "step": 3098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7134567926768224e-05,
+      "loss": 0.3933,
+      "step": 3099
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132592309893818e-05,
+      "loss": 0.3716,
+      "step": 3100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7130616126173606e-05,
+      "loss": 0.3889,
+      "step": 3101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7128639375764643e-05,
+      "loss": 0.3757,
+      "step": 3102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7126662058824025e-05,
+      "loss": 0.421,
+      "step": 3103
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7124684175508894e-05,
+      "loss": 0.3931,
+      "step": 3104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712270572597644e-05,
+      "loss": 0.3986,
+      "step": 3105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712072671038389e-05,
+      "loss": 0.3728,
+      "step": 3106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711874712888853e-05,
+      "loss": 0.3892,
+      "step": 3107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116766981647678e-05,
+      "loss": 0.3463,
+      "step": 3108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71147862688187e-05,
+      "loss": 0.3703,
+      "step": 3109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711280499055901e-05,
+      "loss": 0.4089,
+      "step": 3110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7110823147026067e-05,
+      "loss": 0.4106,
+      "step": 3111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710884073837737e-05,
+      "loss": 0.3646,
+      "step": 3112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710685776477047e-05,
+      "loss": 0.3766,
+      "step": 3113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7104874226362957e-05,
+      "loss": 0.356,
+      "step": 3114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102890123312467e-05,
+      "loss": 0.3429,
+      "step": 3115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100905455776683e-05,
+      "loss": 0.3945,
+      "step": 3116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709892022391333e-05,
+      "loss": 0.4026,
+      "step": 3117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7096934427880183e-05,
+      "loss": 0.4277,
+      "step": 3118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094948067835056e-05,
+      "loss": 0.36,
+      "step": 3119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709296114393581e-05,
+      "loss": 0.413,
+      "step": 3120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090973656340354e-05,
+      "loss": 0.4182,
+      "step": 3121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088985605206637e-05,
+      "loss": 0.3735,
+      "step": 3122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086996990692656e-05,
+      "loss": 0.3918,
+      "step": 3123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7085007812956448e-05,
+      "loss": 0.4079,
+      "step": 3124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70830180721561e-05,
+      "loss": 0.3909,
+      "step": 3125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7081027768449742e-05,
+      "loss": 0.3765,
+      "step": 3126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707903690199555e-05,
+      "loss": 0.4076,
+      "step": 3127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7077045472951738e-05,
+      "loss": 0.3504,
+      "step": 3128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075053481476577e-05,
+      "loss": 0.3879,
+      "step": 3129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707306092772837e-05,
+      "loss": 0.3611,
+      "step": 3130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.3542,
+      "step": 3131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7069074134046292e-05,
+      "loss": 0.3938,
+      "step": 3132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7067079894429256e-05,
+      "loss": 0.3865,
+      "step": 3133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706508509317286e-05,
+      "loss": 0.3722,
+      "step": 3134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063089730435636e-05,
+      "loss": 0.3845,
+      "step": 3135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061093806376158e-05,
+      "loss": 0.3713,
+      "step": 3136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705909732115305e-05,
+      "loss": 0.3855,
+      "step": 3137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7057100274924973e-05,
+      "loss": 0.3636,
+      "step": 3138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7055102667850646e-05,
+      "loss": 0.4019,
+      "step": 3139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053104500088817e-05,
+      "loss": 0.3642,
+      "step": 3140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705110577179829e-05,
+      "loss": 0.3918,
+      "step": 3141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7049106483137902e-05,
+      "loss": 0.3942,
+      "step": 3142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704710663426655e-05,
+      "loss": 0.3756,
+      "step": 3143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7045106225343165e-05,
+      "loss": 0.3242,
+      "step": 3144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7043105256526723e-05,
+      "loss": 0.4218,
+      "step": 3145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041103727976246e-05,
+      "loss": 0.3953,
+      "step": 3146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039101639850803e-05,
+      "loss": 0.3686,
+      "step": 3147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037098992309504e-05,
+      "loss": 0.4174,
+      "step": 3148
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7035095785511506e-05,
+      "loss": 0.4029,
+      "step": 3149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033092019616005e-05,
+      "loss": 0.3765,
+      "step": 3150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703108769478225e-05,
+      "loss": 0.3451,
+      "step": 3151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7029082811169525e-05,
+      "loss": 0.389,
+      "step": 3152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702707736893717e-05,
+      "loss": 0.3767,
+      "step": 3153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7025071368244556e-05,
+      "loss": 0.3939,
+      "step": 3154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702306480925111e-05,
+      "loss": 0.4127,
+      "step": 3155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70210576921163e-05,
+      "loss": 0.3633,
+      "step": 3156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701905001699963e-05,
+      "loss": 0.3557,
+      "step": 3157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701704178406066e-05,
+      "loss": 0.3697,
+      "step": 3158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7015032993458987e-05,
+      "loss": 0.3511,
+      "step": 3159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701302364535426e-05,
+      "loss": 0.365,
+      "step": 3160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701101373990616e-05,
+      "loss": 0.3838,
+      "step": 3161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009003277274423e-05,
+      "loss": 0.3632,
+      "step": 3162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700699225761883e-05,
+      "loss": 0.3637,
+      "step": 3163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7004980681099192e-05,
+      "loss": 0.4069,
+      "step": 3164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002968547875383e-05,
+      "loss": 0.3844,
+      "step": 3165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700095585810731e-05,
+      "loss": 0.359,
+      "step": 3166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6998942611954923e-05,
+      "loss": 0.3992,
+      "step": 3167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996928809578228e-05,
+      "loss": 0.3851,
+      "step": 3168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6994914451137256e-05,
+      "loss": 0.4151,
+      "step": 3169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6992899536792106e-05,
+      "loss": 0.3461,
+      "step": 3170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6990884066702896e-05,
+      "loss": 0.3505,
+      "step": 3171
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698886804102981e-05,
+      "loss": 0.4006,
+      "step": 3172
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6986851459933067e-05,
+      "loss": 0.3861,
+      "step": 3173
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984834323572922e-05,
+      "loss": 0.3708,
+      "step": 3174
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698281663210969e-05,
+      "loss": 0.4319,
+      "step": 3175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6980798385703716e-05,
+      "loss": 0.4201,
+      "step": 3176
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69787795845154e-05,
+      "loss": 0.3457,
+      "step": 3177
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697676022870518e-05,
+      "loss": 0.3769,
+      "step": 3178
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6974740318433537e-05,
+      "loss": 0.4281,
+      "step": 3179
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6972719853861005e-05,
+      "loss": 0.4021,
+      "step": 3180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6970698835148148e-05,
+      "loss": 0.3796,
+      "step": 3181
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6968677262455587e-05,
+      "loss": 0.3841,
+      "step": 3182
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696665513594398e-05,
+      "loss": 0.3669,
+      "step": 3183
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6964632455774027e-05,
+      "loss": 0.3875,
+      "step": 3184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696260922210648e-05,
+      "loss": 0.4004,
+      "step": 3185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960585435102133e-05,
+      "loss": 0.3666,
+      "step": 3186
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958561094921816e-05,
+      "loss": 0.3719,
+      "step": 3187
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695653620172641e-05,
+      "loss": 0.3937,
+      "step": 3188
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695451075567684e-05,
+      "loss": 0.3829,
+      "step": 3189
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6952484756934077e-05,
+      "loss": 0.3963,
+      "step": 3190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6950458205659127e-05,
+      "loss": 0.3934,
+      "step": 3191
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948431102013047e-05,
+      "loss": 0.4229,
+      "step": 3192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6946403446156935e-05,
+      "loss": 0.3765,
+      "step": 3193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944375238251935e-05,
+      "loss": 0.3526,
+      "step": 3194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6942346478459237e-05,
+      "loss": 0.4322,
+      "step": 3195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940317166940068e-05,
+      "loss": 0.3663,
+      "step": 3196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6938287303855703e-05,
+      "loss": 0.3708,
+      "step": 3197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693625688936746e-05,
+      "loss": 0.3868,
+      "step": 3198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6934225923636703e-05,
+      "loss": 0.3523,
+      "step": 3199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693219440682484e-05,
+      "loss": 0.3624,
+      "step": 3200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693016233909332e-05,
+      "loss": 0.3681,
+      "step": 3201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6928129720603635e-05,
+      "loss": 0.4185,
+      "step": 3202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692609655151732e-05,
+      "loss": 0.3762,
+      "step": 3203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6924062831995963e-05,
+      "loss": 0.3628,
+      "step": 3204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6922028562201183e-05,
+      "loss": 0.3914,
+      "step": 3205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919993742294655e-05,
+      "loss": 0.3705,
+      "step": 3206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6917958372438084e-05,
+      "loss": 0.4085,
+      "step": 3207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691592245279323e-05,
+      "loss": 0.4065,
+      "step": 3208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6913885983521895e-05,
+      "loss": 0.3817,
+      "step": 3209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6911848964785917e-05,
+      "loss": 0.3762,
+      "step": 3210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690981139674719e-05,
+      "loss": 0.4154,
+      "step": 3211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690777327956764e-05,
+      "loss": 0.3622,
+      "step": 3212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6905734613409242e-05,
+      "loss": 0.3746,
+      "step": 3213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6903695398434017e-05,
+      "loss": 0.3587,
+      "step": 3214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6901655634804022e-05,
+      "loss": 0.4056,
+      "step": 3215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6899615322681366e-05,
+      "loss": 0.3927,
+      "step": 3216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897574462228198e-05,
+      "loss": 0.3835,
+      "step": 3217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689553305360671e-05,
+      "loss": 0.3776,
+      "step": 3218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689349109697914e-05,
+      "loss": 0.3682,
+      "step": 3219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6891448592507763e-05,
+      "loss": 0.3979,
+      "step": 3220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6889405540354906e-05,
+      "loss": 0.348,
+      "step": 3221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6887361940682933e-05,
+      "loss": 0.3967,
+      "step": 3222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6885317793654255e-05,
+      "loss": 0.4091,
+      "step": 3223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688327309943132e-05,
+      "loss": 0.39,
+      "step": 3224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6881227858176642e-05,
+      "loss": 0.3979,
+      "step": 3225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6879182070052745e-05,
+      "loss": 0.3885,
+      "step": 3226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877135735222224e-05,
+      "loss": 0.3622,
+      "step": 3227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6875088853847698e-05,
+      "loss": 0.4048,
+      "step": 3228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6873041426091845e-05,
+      "loss": 0.3928,
+      "step": 3229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6870993452117374e-05,
+      "loss": 0.3575,
+      "step": 3230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6868944932087047e-05,
+      "loss": 0.369,
+      "step": 3231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6866895866163663e-05,
+      "loss": 0.3555,
+      "step": 3232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864846254510064e-05,
+      "loss": 0.354,
+      "step": 3233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6862796097289143e-05,
+      "loss": 0.373,
+      "step": 3234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686074539466383e-05,
+      "loss": 0.3727,
+      "step": 3235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68586941467971e-05,
+      "loss": 0.4178,
+      "step": 3236
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685664235385197e-05,
+      "loss": 0.3737,
+      "step": 3237
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68545900159915e-05,
+      "loss": 0.3764,
+      "step": 3238
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6852537133378795e-05,
+      "loss": 0.3847,
+      "step": 3239
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6850483706177007e-05,
+      "loss": 0.3948,
+      "step": 3240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6848429734549325e-05,
+      "loss": 0.3975,
+      "step": 3241
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684637521865898e-05,
+      "loss": 0.361,
+      "step": 3242
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844320158669256e-05,
+      "loss": 0.3871,
+      "step": 3243
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684226455474347e-05,
+      "loss": 0.3531,
+      "step": 3244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840208407044983e-05,
+      "loss": 0.3704,
+      "step": 3245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838151715737213e-05,
+      "loss": 0.3723,
+      "step": 3246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68360944809836e-05,
+      "loss": 0.3642,
+      "step": 3247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683403670294764e-05,
+      "loss": 0.3975,
+      "step": 3248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6831978381792878e-05,
+      "loss": 0.3884,
+      "step": 3249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6829919517682883e-05,
+      "loss": 0.3478,
+      "step": 3250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827860110781283e-05,
+      "loss": 0.3975,
+      "step": 3251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682580016125175e-05,
+      "loss": 0.4094,
+      "step": 3252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823739669257984e-05,
+      "loss": 0.379,
+      "step": 3253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6821678634963743e-05,
+      "loss": 0.3868,
+      "step": 3254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6819617058532823e-05,
+      "loss": 0.381,
+      "step": 3255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817554940129062e-05,
+      "loss": 0.4383,
+      "step": 3256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681549227991634e-05,
+      "loss": 0.3877,
+      "step": 3257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6813429078058586e-05,
+      "loss": 0.3414,
+      "step": 3258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6811365334719764e-05,
+      "loss": 0.3862,
+      "step": 3259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6809301050063887e-05,
+      "loss": 0.3878,
+      "step": 3260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680723622425501e-05,
+      "loss": 0.385,
+      "step": 3261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680517085745723e-05,
+      "loss": 0.3699,
+      "step": 3262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803104949834686e-05,
+      "loss": 0.4182,
+      "step": 3263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680103850155156e-05,
+      "loss": 0.3866,
+      "step": 3264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6798971512772076e-05,
+      "loss": 0.4079,
+      "step": 3265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796903983660514e-05,
+      "loss": 0.3271,
+      "step": 3266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794835914381177e-05,
+      "loss": 0.3557,
+      "step": 3267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6792767305098417e-05,
+      "loss": 0.3936,
+      "step": 3268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790698155976638e-05,
+      "loss": 0.3561,
+      "step": 3269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678862846718028e-05,
+      "loss": 0.3744,
+      "step": 3270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786558238873826e-05,
+      "loss": 0.3488,
+      "step": 3271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.67844874712218e-05,
+      "loss": 0.3859,
+      "step": 3272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782416164388774e-05,
+      "loss": 0.3542,
+      "step": 3273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678034431853936e-05,
+      "loss": 0.4174,
+      "step": 3274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677827193383821e-05,
+      "loss": 0.3643,
+      "step": 3275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776199010450026e-05,
+      "loss": 0.3793,
+      "step": 3276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6774125548539547e-05,
+      "loss": 0.3517,
+      "step": 3277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6772051548271556e-05,
+      "loss": 0.3887,
+      "step": 3278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6769977009810882e-05,
+      "loss": 0.4094,
+      "step": 3279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6767901933322388e-05,
+      "loss": 0.3836,
+      "step": 3280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6765826318970994e-05,
+      "loss": 0.4012,
+      "step": 3281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6763750166921642e-05,
+      "loss": 0.3701,
+      "step": 3282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6761673477339344e-05,
+      "loss": 0.3747,
+      "step": 3283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6759596250389133e-05,
+      "loss": 0.3608,
+      "step": 3284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6757518486236088e-05,
+      "loss": 0.4179,
+      "step": 3285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675544018504534e-05,
+      "loss": 0.3944,
+      "step": 3286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6753361346982054e-05,
+      "loss": 0.3734,
+      "step": 3287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6751281972211445e-05,
+      "loss": 0.3856,
+      "step": 3288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749202060898763e-05,
+      "loss": 0.3603,
+      "step": 3289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67471216132093e-05,
+      "loss": 0.3749,
+      "step": 3290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745040629308406e-05,
+      "loss": 0.3735,
+      "step": 3291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6742959109361447e-05,
+      "loss": 0.4069,
+      "step": 3292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674087705353386e-05,
+      "loss": 0.42,
+      "step": 3293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6738794461991102e-05,
+      "loss": 0.3523,
+      "step": 3294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736711334898693e-05,
+      "loss": 0.3571,
+      "step": 3295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673462767242217e-05,
+      "loss": 0.4298,
+      "step": 3296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732543474727144e-05,
+      "loss": 0.3922,
+      "step": 3297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6730458741979236e-05,
+      "loss": 0.3804,
+      "step": 3298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6728373474344136e-05,
+      "loss": 0.3614,
+      "step": 3299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672628767198756e-05,
+      "loss": 0.387,
+      "step": 3300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672420133507527e-05,
+      "loss": 0.3687,
+      "step": 3301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722114463773085e-05,
+      "loss": 0.3599,
+      "step": 3302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672002705824684e-05,
+      "loss": 0.3911,
+      "step": 3303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6717939118662433e-05,
+      "loss": 0.3783,
+      "step": 3304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6715850645185796e-05,
+      "loss": 0.4112,
+      "step": 3305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713761637982908e-05,
+      "loss": 0.3869,
+      "step": 3306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711672097219787e-05,
+      "loss": 0.3598,
+      "step": 3307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6709582023062496e-05,
+      "loss": 0.4312,
+      "step": 3308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670749141567713e-05,
+      "loss": 0.3949,
+      "step": 3309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6705400275229845e-05,
+      "loss": 0.3658,
+      "step": 3310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703308601886828e-05,
+      "loss": 0.3686,
+      "step": 3311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6701216395814308e-05,
+      "loss": 0.3456,
+      "step": 3312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699123657178553e-05,
+      "loss": 0.3531,
+      "step": 3313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697030386145886e-05,
+      "loss": 0.3682,
+      "step": 3314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6694936582882663e-05,
+      "loss": 0.3929,
+      "step": 3315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669284224755528e-05,
+      "loss": 0.3312,
+      "step": 3316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6690747380330185e-05,
+      "loss": 0.3457,
+      "step": 3317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668865198137386e-05,
+      "loss": 0.4331,
+      "step": 3318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6686556050852833e-05,
+      "loss": 0.3637,
+      "step": 3319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668445958893367e-05,
+      "loss": 0.3933,
+      "step": 3320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6682362595782983e-05,
+      "loss": 0.4503,
+      "step": 3321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6680265071567428e-05,
+      "loss": 0.378,
+      "step": 3322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66781670164537e-05,
+      "loss": 0.3911,
+      "step": 3323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6676068430608535e-05,
+      "loss": 0.3824,
+      "step": 3324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667396931419872e-05,
+      "loss": 0.3717,
+      "step": 3325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671869667391063e-05,
+      "loss": 0.406,
+      "step": 3326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666976949035244e-05,
+      "loss": 0.3485,
+      "step": 3327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666766878324976e-05,
+      "loss": 0.3468,
+      "step": 3328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666556754624996e-05,
+      "loss": 0.3773,
+      "step": 3329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6663465779520042e-05,
+      "loss": 0.35,
+      "step": 3330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666136348322703e-05,
+      "loss": 0.361,
+      "step": 3331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659260657538008e-05,
+      "loss": 0.3945,
+      "step": 3332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657157302620084e-05,
+      "loss": 0.375,
+      "step": 3333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6655053418640424e-05,
+      "loss": 0.3489,
+      "step": 3334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6652949005766226e-05,
+      "loss": 0.3536,
+      "step": 3335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6650844064164735e-05,
+      "loss": 0.3682,
+      "step": 3336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6648738594003233e-05,
+      "loss": 0.3815,
+      "step": 3337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664663259544905e-05,
+      "loss": 0.3514,
+      "step": 3338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644526068669555e-05,
+      "loss": 0.3983,
+      "step": 3339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6642419013832157e-05,
+      "loss": 0.3671,
+      "step": 3340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640311431104314e-05,
+      "loss": 0.4039,
+      "step": 3341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6638203320653515e-05,
+      "loss": 0.3621,
+      "step": 3342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6636094682647303e-05,
+      "loss": 0.3796,
+      "step": 3343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6633985517253252e-05,
+      "loss": 0.4015,
+      "step": 3344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631875824638986e-05,
+      "loss": 0.3425,
+      "step": 3345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6629765604972163e-05,
+      "loss": 0.3912,
+      "step": 3346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6627654858420496e-05,
+      "loss": 0.4356,
+      "step": 3347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625543585151725e-05,
+      "loss": 0.3583,
+      "step": 3348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623431785333643e-05,
+      "loss": 0.3777,
+      "step": 3349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6621319459134072e-05,
+      "loss": 0.3645,
+      "step": 3350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6619206606720896e-05,
+      "loss": 0.3737,
+      "step": 3351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661709322826202e-05,
+      "loss": 0.4315,
+      "step": 3352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6614979323925404e-05,
+      "loss": 0.4011,
+      "step": 3353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6612864893879043e-05,
+      "loss": 0.365,
+      "step": 3354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6610749938290977e-05,
+      "loss": 0.3437,
+      "step": 3355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6608634457329285e-05,
+      "loss": 0.3736,
+      "step": 3356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606518451162094e-05,
+      "loss": 0.3713,
+      "step": 3357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660440191995757e-05,
+      "loss": 0.3948,
+      "step": 3358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660228486388391e-05,
+      "loss": 0.358,
+      "step": 3359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6600167283109372e-05,
+      "loss": 0.4128,
+      "step": 3360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659804917780224e-05,
+      "loss": 0.3507,
+      "step": 3361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659593054813085e-05,
+      "loss": 0.3329,
+      "step": 3362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6593811394263568e-05,
+      "loss": 0.3899,
+      "step": 3363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6591691716368812e-05,
+      "loss": 0.3794,
+      "step": 3364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6589571514615045e-05,
+      "loss": 0.3916,
+      "step": 3365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6587450789170754e-05,
+      "loss": 0.3654,
+      "step": 3366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6585329540204486e-05,
+      "loss": 0.3456,
+      "step": 3367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658320776788482e-05,
+      "loss": 0.3742,
+      "step": 3368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658108547238038e-05,
+      "loss": 0.3461,
+      "step": 3369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6578962653859827e-05,
+      "loss": 0.3853,
+      "step": 3370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6576839312491868e-05,
+      "loss": 0.3777,
+      "step": 3371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574715448445253e-05,
+      "loss": 0.3749,
+      "step": 3372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657259106188877e-05,
+      "loss": 0.3484,
+      "step": 3373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657046615299125e-05,
+      "loss": 0.387,
+      "step": 3374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6568340721921564e-05,
+      "loss": 0.3978,
+      "step": 3375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6566214768848624e-05,
+      "loss": 0.3385,
+      "step": 3376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656408829394139e-05,
+      "loss": 0.3441,
+      "step": 3377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6561961297368855e-05,
+      "loss": 0.391,
+      "step": 3378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559833779300053e-05,
+      "loss": 0.3678,
+      "step": 3379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6557705739904072e-05,
+      "loss": 0.3945,
+      "step": 3380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555577179350026e-05,
+      "loss": 0.3932,
+      "step": 3381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6553448097807082e-05,
+      "loss": 0.3632,
+      "step": 3382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6551318495444443e-05,
+      "loss": 0.3704,
+      "step": 3383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6549188372431352e-05,
+      "loss": 0.3945,
+      "step": 3384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6547057728937095e-05,
+      "loss": 0.3785,
+      "step": 3385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6544926565131003e-05,
+      "loss": 0.4081,
+      "step": 3386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542794881182443e-05,
+      "loss": 0.3842,
+      "step": 3387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6540662677260824e-05,
+      "loss": 0.4104,
+      "step": 3388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65385299535356e-05,
+      "loss": 0.383,
+      "step": 3389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6536396710176264e-05,
+      "loss": 0.3763,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653426294735235e-05,
+      "loss": 0.3653,
+      "step": 3391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653212866523344e-05,
+      "loss": 0.3954,
+      "step": 3392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6529993863989134e-05,
+      "loss": 0.3554,
+      "step": 3393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652785854378911e-05,
+      "loss": 0.3795,
+      "step": 3394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6525722704803054e-05,
+      "loss": 0.3652,
+      "step": 3395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6523586347200713e-05,
+      "loss": 0.3912,
+      "step": 3396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6521449471151867e-05,
+      "loss": 0.3676,
+      "step": 3397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651931207682634e-05,
+      "loss": 0.3807,
+      "step": 3398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517174164393998e-05,
+      "loss": 0.3609,
+      "step": 3399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651503573402474e-05,
+      "loss": 0.3517,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651289678588852e-05,
+      "loss": 0.3667,
+      "step": 3401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651075732015532e-05,
+      "loss": 0.3873,
+      "step": 3402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650861733699517e-05,
+      "loss": 0.3759,
+      "step": 3403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506476836578148e-05,
+      "loss": 0.349,
+      "step": 3404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6504335819074358e-05,
+      "loss": 0.4268,
+      "step": 3405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6502194284653953e-05,
+      "loss": 0.3651,
+      "step": 3406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6500052233487122e-05,
+      "loss": 0.3649,
+      "step": 3407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497909665744106e-05,
+      "loss": 0.3764,
+      "step": 3408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649576658159518e-05,
+      "loss": 0.3545,
+      "step": 3409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493622981210658e-05,
+      "loss": 0.3633,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6491478864760902e-05,
+      "loss": 0.3547,
+      "step": 3411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6489334232416304e-05,
+      "loss": 0.3476,
+      "step": 3412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6487189084347307e-05,
+      "loss": 0.3687,
+      "step": 3413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648504342072439e-05,
+      "loss": 0.3995,
+      "step": 3414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482897241718076e-05,
+      "loss": 0.3591,
+      "step": 3415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648075054749893e-05,
+      "loss": 0.4328,
+      "step": 3416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6478603338237552e-05,
+      "loss": 0.4057,
+      "step": 3417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6476455614104586e-05,
+      "loss": 0.3693,
+      "step": 3418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647430737527072e-05,
+      "loss": 0.4151,
+      "step": 3419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647215862190667e-05,
+      "loss": 0.3461,
+      "step": 3420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647000935418322e-05,
+      "loss": 0.3615,
+      "step": 3421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6467859572271166e-05,
+      "loss": 0.3692,
+      "step": 3422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6465709276341364e-05,
+      "loss": 0.3674,
+      "step": 3423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64635584665647e-05,
+      "loss": 0.3867,
+      "step": 3424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64614071431121e-05,
+      "loss": 0.3531,
+      "step": 3425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645925530615454e-05,
+      "loss": 0.365,
+      "step": 3426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457102955863034e-05,
+      "loss": 0.3691,
+      "step": 3427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6454950092408632e-05,
+      "loss": 0.3825,
+      "step": 3428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6452796715962427e-05,
+      "loss": 0.3768,
+      "step": 3429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6450642826695557e-05,
+      "loss": 0.3616,
+      "step": 3430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6448488424779194e-05,
+      "loss": 0.3619,
+      "step": 3431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446333510384557e-05,
+      "loss": 0.3831,
+      "step": 3432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6444178083682895e-05,
+      "loss": 0.359,
+      "step": 3433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6442022144845517e-05,
+      "loss": 0.3408,
+      "step": 3434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439865694043753e-05,
+      "loss": 0.4017,
+      "step": 3435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6437708731448988e-05,
+      "loss": 0.354,
+      "step": 3436
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6435551257232632e-05,
+      "loss": 0.3519,
+      "step": 3437
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643339327156615e-05,
+      "loss": 0.3455,
+      "step": 3438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643123477462105e-05,
+      "loss": 0.37,
+      "step": 3439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642907576656886e-05,
+      "loss": 0.4139,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6426916247581172e-05,
+      "loss": 0.4383,
+      "step": 3441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642475621782961e-05,
+      "loss": 0.3834,
+      "step": 3442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422595677485825e-05,
+      "loss": 0.3754,
+      "step": 3443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642043462672153e-05,
+      "loss": 0.3629,
+      "step": 3444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6418273065708474e-05,
+      "loss": 0.4041,
+      "step": 3445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641611099461843e-05,
+      "loss": 0.454,
+      "step": 3446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641394841362323e-05,
+      "loss": 0.3754,
+      "step": 3447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641178532289474e-05,
+      "loss": 0.388,
+      "step": 3448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6409621722604865e-05,
+      "loss": 0.3663,
+      "step": 3449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407457612925554e-05,
+      "loss": 0.3827,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6405292994028795e-05,
+      "loss": 0.3419,
+      "step": 3451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640312786608661e-05,
+      "loss": 0.3803,
+      "step": 3452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6400962229271074e-05,
+      "loss": 0.4051,
+      "step": 3453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6398796083754295e-05,
+      "loss": 0.3587,
+      "step": 3454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6396629429708414e-05,
+      "loss": 0.371,
+      "step": 3455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6394462267305634e-05,
+      "loss": 0.4192,
+      "step": 3456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6392294596718176e-05,
+      "loss": 0.4149,
+      "step": 3457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6390126418118314e-05,
+      "loss": 0.3615,
+      "step": 3458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6387957731678358e-05,
+      "loss": 0.3801,
+      "step": 3459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6385788537570655e-05,
+      "loss": 0.3492,
+      "step": 3460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6383618835967606e-05,
+      "loss": 0.3687,
+      "step": 3461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638144862704163e-05,
+      "loss": 0.352,
+      "step": 3462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6379277910965213e-05,
+      "loss": 0.4139,
+      "step": 3463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6377106687910858e-05,
+      "loss": 0.3673,
+      "step": 3464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6374934958051124e-05,
+      "loss": 0.396,
+      "step": 3465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63727627215586e-05,
+      "loss": 0.3881,
+      "step": 3466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6370589978605914e-05,
+      "loss": 0.3659,
+      "step": 3467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6368416729365754e-05,
+      "loss": 0.3546,
+      "step": 3468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636624297401082e-05,
+      "loss": 0.4032,
+      "step": 3469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6364068712713875e-05,
+      "loss": 0.4083,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636189394564771e-05,
+      "loss": 0.3658,
+      "step": 3471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6359718672985154e-05,
+      "loss": 0.4029,
+      "step": 3472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635754289489909e-05,
+      "loss": 0.3791,
+      "step": 3473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6355366611562433e-05,
+      "loss": 0.3791,
+      "step": 3474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635318982314813e-05,
+      "loss": 0.3529,
+      "step": 3475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6351012529829184e-05,
+      "loss": 0.3762,
+      "step": 3476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348834731778626e-05,
+      "loss": 0.4031,
+      "step": 3477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634665642916953e-05,
+      "loss": 0.3755,
+      "step": 3478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6344477622175017e-05,
+      "loss": 0.349,
+      "step": 3479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6342298310968237e-05,
+      "loss": 0.3833,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634011849572239e-05,
+      "loss": 0.398,
+      "step": 3481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6337938176610705e-05,
+      "loss": 0.388,
+      "step": 3482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6335757353806468e-05,
+      "loss": 0.3647,
+      "step": 3483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333576027482983e-05,
+      "loss": 0.3585,
+      "step": 3484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331394197813612e-05,
+      "loss": 0.4021,
+      "step": 3485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6329211864971753e-05,
+      "loss": 0.4164,
+      "step": 3486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632702902913084e-05,
+      "loss": 0.3674,
+      "step": 3487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324845690464345e-05,
+      "loss": 0.3826,
+      "step": 3488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632266184914579e-05,
+      "loss": 0.3638,
+      "step": 3489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6320477505348717e-05,
+      "loss": 0.3678,
+      "step": 3490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631829265924674e-05,
+      "loss": 0.3882,
+      "step": 3491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6316107311013483e-05,
+      "loss": 0.3758,
+      "step": 3492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6313921460822628e-05,
+      "loss": 0.3795,
+      "step": 3493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311735108847878e-05,
+      "loss": 0.3861,
+      "step": 3494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6309548255263003e-05,
+      "loss": 0.3808,
+      "step": 3495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630736090024179e-05,
+      "loss": 0.3464,
+      "step": 3496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6305173043958076e-05,
+      "loss": 0.3873,
+      "step": 3497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6302984686585733e-05,
+      "loss": 0.4103,
+      "step": 3498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630079582829868e-05,
+      "loss": 0.3603,
+      "step": 3499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629860646927087e-05,
+      "loss": 0.3605,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629641660967629e-05,
+      "loss": 0.3509,
+      "step": 3501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6294226249688984e-05,
+      "loss": 0.3555,
+      "step": 3502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629203538948302e-05,
+      "loss": 0.3818,
+      "step": 3503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6289844029232515e-05,
+      "loss": 0.382,
+      "step": 3504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6287652169111617e-05,
+      "loss": 0.409,
+      "step": 3505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6285459809294522e-05,
+      "loss": 0.3985,
+      "step": 3506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6283266949955464e-05,
+      "loss": 0.3645,
+      "step": 3507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6281073591268715e-05,
+      "loss": 0.338,
+      "step": 3508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6278879733408587e-05,
+      "loss": 0.3583,
+      "step": 3509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6276685376549427e-05,
+      "loss": 0.3676,
+      "step": 3510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6274490520865632e-05,
+      "loss": 0.3783,
+      "step": 3511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6272295166531627e-05,
+      "loss": 0.3506,
+      "step": 3512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6270099313721887e-05,
+      "loss": 0.3767,
+      "step": 3513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626790296261092e-05,
+      "loss": 0.3863,
+      "step": 3514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6265706113373282e-05,
+      "loss": 0.4064,
+      "step": 3515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6263508766183554e-05,
+      "loss": 0.3581,
+      "step": 3516
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6261310921216368e-05,
+      "loss": 0.3972,
+      "step": 3517
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6259112578646395e-05,
+      "loss": 0.3995,
+      "step": 3518
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256913738648338e-05,
+      "loss": 0.4033,
+      "step": 3519
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6254714401396947e-05,
+      "loss": 0.3391,
+      "step": 3520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6252514567067012e-05,
+      "loss": 0.3908,
+      "step": 3521
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250314235833357e-05,
+      "loss": 0.3822,
+      "step": 3522
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6248113407870847e-05,
+      "loss": 0.3553,
+      "step": 3523
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624591208335439e-05,
+      "loss": 0.3613,
+      "step": 3524
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6243710262458928e-05,
+      "loss": 0.3671,
+      "step": 3525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624150794535945e-05,
+      "loss": 0.3465,
+      "step": 3526
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6239305132230977e-05,
+      "loss": 0.3518,
+      "step": 3527
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623710182324857e-05,
+      "loss": 0.3859,
+      "step": 3528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.4093,
+      "step": 3529
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232693718422418e-05,
+      "loss": 0.345,
+      "step": 3530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623048892292899e-05,
+      "loss": 0.3751,
+      "step": 3531
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622828363228228e-05,
+      "loss": 0.3601,
+      "step": 3532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6226077846657546e-05,
+      "loss": 0.3967,
+      "step": 3533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622387156623009e-05,
+      "loss": 0.4104,
+      "step": 3534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221664791175243e-05,
+      "loss": 0.3456,
+      "step": 3535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6219457521668394e-05,
+      "loss": 0.374,
+      "step": 3536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217249757884954e-05,
+      "loss": 0.3726,
+      "step": 3537
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215041500000383e-05,
+      "loss": 0.3509,
+      "step": 3538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621283274819017e-05,
+      "loss": 0.4153,
+      "step": 3539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210623502629856e-05,
+      "loss": 0.3881,
+      "step": 3540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620841376349502e-05,
+      "loss": 0.3916,
+      "step": 3541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6206203530961264e-05,
+      "loss": 0.3607,
+      "step": 3542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6203992805204252e-05,
+      "loss": 0.383,
+      "step": 3543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620178158639967e-05,
+      "loss": 0.4048,
+      "step": 3544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6199569874723252e-05,
+      "loss": 0.37,
+      "step": 3545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6197357670350762e-05,
+      "loss": 0.3681,
+      "step": 3546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6195144973458023e-05,
+      "loss": 0.4111,
+      "step": 3547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6192931784220874e-05,
+      "loss": 0.3757,
+      "step": 3548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6190718102815207e-05,
+      "loss": 0.3366,
+      "step": 3549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618850392941694e-05,
+      "loss": 0.3809,
+      "step": 3550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6186289264202052e-05,
+      "loss": 0.385,
+      "step": 3551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6184074107346544e-05,
+      "loss": 0.3723,
+      "step": 3552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6181858459026456e-05,
+      "loss": 0.3986,
+      "step": 3553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6179642319417874e-05,
+      "loss": 0.4308,
+      "step": 3554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6177425688696923e-05,
+      "loss": 0.374,
+      "step": 3555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175208567039758e-05,
+      "loss": 0.3558,
+      "step": 3556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6172990954622586e-05,
+      "loss": 0.374,
+      "step": 3557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6170772851621644e-05,
+      "loss": 0.3465,
+      "step": 3558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616855425821321e-05,
+      "loss": 0.377,
+      "step": 3559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6166335174573603e-05,
+      "loss": 0.3862,
+      "step": 3560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6164115600879177e-05,
+      "loss": 0.3373,
+      "step": 3561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6161895537306332e-05,
+      "loss": 0.352,
+      "step": 3562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159674984031496e-05,
+      "loss": 0.3751,
+      "step": 3563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6157453941231146e-05,
+      "loss": 0.3673,
+      "step": 3564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6155232409081794e-05,
+      "loss": 0.3564,
+      "step": 3565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153010387759987e-05,
+      "loss": 0.3683,
+      "step": 3566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6150787877442324e-05,
+      "loss": 0.4377,
+      "step": 3567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148564878305425e-05,
+      "loss": 0.3779,
+      "step": 3568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146341390525957e-05,
+      "loss": 0.3773,
+      "step": 3569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614411741428063e-05,
+      "loss": 0.3496,
+      "step": 3570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141892949746196e-05,
+      "loss": 0.3482,
+      "step": 3571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6139667997099424e-05,
+      "loss": 0.3896,
+      "step": 3572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613744255651715e-05,
+      "loss": 0.3649,
+      "step": 3573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6135216628176227e-05,
+      "loss": 0.3622,
+      "step": 3574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613299021225356e-05,
+      "loss": 0.3862,
+      "step": 3575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613076330892609e-05,
+      "loss": 0.3551,
+      "step": 3576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6128535918370786e-05,
+      "loss": 0.3505,
+      "step": 3577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6126308040764674e-05,
+      "loss": 0.3322,
+      "step": 3578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6124079676284806e-05,
+      "loss": 0.3712,
+      "step": 3579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6121850825108274e-05,
+      "loss": 0.3848,
+      "step": 3580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6119621487412212e-05,
+      "loss": 0.3898,
+      "step": 3581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6117391663373793e-05,
+      "loss": 0.3936,
+      "step": 3582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6115161353170222e-05,
+      "loss": 0.3681,
+      "step": 3583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6112930556978757e-05,
+      "loss": 0.3537,
+      "step": 3584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611069927497667e-05,
+      "loss": 0.3642,
+      "step": 3585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6108467507341307e-05,
+      "loss": 0.4069,
+      "step": 3586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6106235254250015e-05,
+      "loss": 0.3868,
+      "step": 3587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6104002515880205e-05,
+      "loss": 0.3698,
+      "step": 3588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101769292409317e-05,
+      "loss": 0.3962,
+      "step": 3589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6099535584014834e-05,
+      "loss": 0.3444,
+      "step": 3590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609730139087427e-05,
+      "loss": 0.3945,
+      "step": 3591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6095066713165182e-05,
+      "loss": 0.3575,
+      "step": 3592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609283155106517e-05,
+      "loss": 0.4053,
+      "step": 3593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090595904751864e-05,
+      "loss": 0.3783,
+      "step": 3594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6088359774402936e-05,
+      "loss": 0.4153,
+      "step": 3595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6086123160196107e-05,
+      "loss": 0.3561,
+      "step": 3596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6083886062309114e-05,
+      "loss": 0.372,
+      "step": 3597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6081648480919754e-05,
+      "loss": 0.3842,
+      "step": 3598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607941041620585e-05,
+      "loss": 0.3551,
+      "step": 3599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607717186834526e-05,
+      "loss": 0.3832,
+      "step": 3600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60749328375159e-05,
+      "loss": 0.3546,
+      "step": 3601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6072693323895706e-05,
+      "loss": 0.4092,
+      "step": 3602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6070453327662657e-05,
+      "loss": 0.3744,
+      "step": 3603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606821284899477e-05,
+      "loss": 0.3803,
+      "step": 3604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6065971888070104e-05,
+      "loss": 0.3952,
+      "step": 3605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606373044506676e-05,
+      "loss": 0.3888,
+      "step": 3606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6061488520162854e-05,
+      "loss": 0.3605,
+      "step": 3607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059246113536574e-05,
+      "loss": 0.3778,
+      "step": 3608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057003225366127e-05,
+      "loss": 0.3745,
+      "step": 3609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605475985582976e-05,
+      "loss": 0.3887,
+      "step": 3610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6052516005105755e-05,
+      "loss": 0.4007,
+      "step": 3611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050271673372438e-05,
+      "loss": 0.4048,
+      "step": 3612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6048026860808177e-05,
+      "loss": 0.3676,
+      "step": 3613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604578156759137e-05,
+      "loss": 0.3607,
+      "step": 3614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6043535793900457e-05,
+      "loss": 0.3457,
+      "step": 3615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6041289539913913e-05,
+      "loss": 0.3382,
+      "step": 3616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6039042805810258e-05,
+      "loss": 0.3643,
+      "step": 3617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603679559176804e-05,
+      "loss": 0.3722,
+      "step": 3618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6034547897965857e-05,
+      "loss": 0.3664,
+      "step": 3619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6032299724582337e-05,
+      "loss": 0.3941,
+      "step": 3620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6030051071796146e-05,
+      "loss": 0.4225,
+      "step": 3621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6027801939785998e-05,
+      "loss": 0.4078,
+      "step": 3622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025552328730624e-05,
+      "loss": 0.3755,
+      "step": 3623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602330223880882e-05,
+      "loss": 0.3792,
+      "step": 3624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6021051670199394e-05,
+      "loss": 0.4146,
+      "step": 3625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6018800623081214e-05,
+      "loss": 0.3837,
+      "step": 3626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6016549097633177e-05,
+      "loss": 0.3993,
+      "step": 3627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601429709403421e-05,
+      "loss": 0.3693,
+      "step": 3628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6012044612463292e-05,
+      "loss": 0.3534,
+      "step": 3629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6009791653099434e-05,
+      "loss": 0.3458,
+      "step": 3630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600753821612168e-05,
+      "loss": 0.3858,
+      "step": 3631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600528430170912e-05,
+      "loss": 0.3745,
+      "step": 3632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6003029910040877e-05,
+      "loss": 0.3587,
+      "step": 3633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600077504129611e-05,
+      "loss": 0.3706,
+      "step": 3634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599851969565403e-05,
+      "loss": 0.3697,
+      "step": 3635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5996263873293862e-05,
+      "loss": 0.3528,
+      "step": 3636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599400757439489e-05,
+      "loss": 0.334,
+      "step": 3637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599175079913643e-05,
+      "loss": 0.39,
+      "step": 3638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5989493547697827e-05,
+      "loss": 0.3488,
+      "step": 3639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5987235820258477e-05,
+      "loss": 0.3835,
+      "step": 3640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59849776169978e-05,
+      "loss": 0.3581,
+      "step": 3641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5982718938095265e-05,
+      "loss": 0.3728,
+      "step": 3642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598045978373038e-05,
+      "loss": 0.3764,
+      "step": 3643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597820015408268e-05,
+      "loss": 0.4189,
+      "step": 3644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5975940049331748e-05,
+      "loss": 0.3356,
+      "step": 3645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5973679469657196e-05,
+      "loss": 0.3879,
+      "step": 3646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5971418415238678e-05,
+      "loss": 0.3294,
+      "step": 3647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596915688625589e-05,
+      "loss": 0.38,
+      "step": 3648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596689488288856e-05,
+      "loss": 0.3895,
+      "step": 3649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5964632405316457e-05,
+      "loss": 0.3952,
+      "step": 3650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596236945371938e-05,
+      "loss": 0.4231,
+      "step": 3651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5960106028277178e-05,
+      "loss": 0.3579,
+      "step": 3652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595784212916973e-05,
+      "loss": 0.3793,
+      "step": 3653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595557775657695e-05,
+      "loss": 0.3655,
+      "step": 3654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59533129106788e-05,
+      "loss": 0.3793,
+      "step": 3655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595104759165527e-05,
+      "loss": 0.3895,
+      "step": 3656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5948781799686387e-05,
+      "loss": 0.3835,
+      "step": 3657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5946515534952228e-05,
+      "loss": 0.3797,
+      "step": 3658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594424879763289e-05,
+      "loss": 0.4221,
+      "step": 3659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5941981587908525e-05,
+      "loss": 0.3484,
+      "step": 3660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5939713905959306e-05,
+      "loss": 0.3606,
+      "step": 3661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5937445751965456e-05,
+      "loss": 0.3608,
+      "step": 3662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935177126107234e-05,
+      "loss": 0.4103,
+      "step": 3663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5932908028564926e-05,
+      "loss": 0.374,
+      "step": 3664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593063845951887e-05,
+      "loss": 0.3842,
+      "step": 3665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592836841914943e-05,
+      "loss": 0.3484,
+      "step": 3666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5926097907637018e-05,
+      "loss": 0.4021,
+      "step": 3667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5923826925162073e-05,
+      "loss": 0.3734,
+      "step": 3668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592155547190507e-05,
+      "loss": 0.4534,
+      "step": 3669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591928354804654e-05,
+      "loss": 0.4037,
+      "step": 3670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591701115376703e-05,
+      "loss": 0.3622,
+      "step": 3671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591473828924714e-05,
+      "loss": 0.3778,
+      "step": 3672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912464954667493e-05,
+      "loss": 0.3646,
+      "step": 3673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591019115020876e-05,
+      "loss": 0.3995,
+      "step": 3674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5907916876051647e-05,
+      "loss": 0.3666,
+      "step": 3675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905642132376895e-05,
+      "loss": 0.3661,
+      "step": 3676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5903366919365283e-05,
+      "loss": 0.4015,
+      "step": 3677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901091237197636e-05,
+      "loss": 0.4093,
+      "step": 3678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58988150860548e-05,
+      "loss": 0.4004,
+      "step": 3679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5896538466117664e-05,
+      "loss": 0.4162,
+      "step": 3680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5894261377567164e-05,
+      "loss": 0.3863,
+      "step": 3681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5891983820584265e-05,
+      "loss": 0.3723,
+      "step": 3682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588970579534997e-05,
+      "loss": 0.4009,
+      "step": 3683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588742730204532e-05,
+      "loss": 0.3805,
+      "step": 3684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885148340851387e-05,
+      "loss": 0.3617,
+      "step": 3685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5882868911949297e-05,
+      "loss": 0.3733,
+      "step": 3686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880589015520193e-05,
+      "loss": 0.3772,
+      "step": 3687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5878308651745268e-05,
+      "loss": 0.3934,
+      "step": 3688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587602782080575e-05,
+      "loss": 0.3654,
+      "step": 3689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5873746522882897e-05,
+      "loss": 0.3872,
+      "step": 3690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587146475815802e-05,
+      "loss": 0.3897,
+      "step": 3691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5869182526812444e-05,
+      "loss": 0.3508,
+      "step": 3692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586689982902755e-05,
+      "loss": 0.3933,
+      "step": 3693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586461666498475e-05,
+      "loss": 0.3467,
+      "step": 3694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5862333034865495e-05,
+      "loss": 0.41,
+      "step": 3695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586004893885127e-05,
+      "loss": 0.3783,
+      "step": 3696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5857764377123596e-05,
+      "loss": 0.394,
+      "step": 3697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5855479349864035e-05,
+      "loss": 0.3776,
+      "step": 3698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5853193857254182e-05,
+      "loss": 0.3786,
+      "step": 3699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5850907899475674e-05,
+      "loss": 0.3729,
+      "step": 3700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584862147671018e-05,
+      "loss": 0.3654,
+      "step": 3701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5846334589139408e-05,
+      "loss": 0.3669,
+      "step": 3702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5844047236945105e-05,
+      "loss": 0.4149,
+      "step": 3703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584175942030905e-05,
+      "loss": 0.3739,
+      "step": 3704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839471139413065e-05,
+      "loss": 0.3897,
+      "step": 3705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5837182394439e-05,
+      "loss": 0.3696,
+      "step": 3706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834893185568755e-05,
+      "loss": 0.4012,
+      "step": 3707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5832603512984257e-05,
+      "loss": 0.3599,
+      "step": 3708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830313376867468e-05,
+      "loss": 0.3653,
+      "step": 3709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5828022777400397e-05,
+      "loss": 0.3669,
+      "step": 3710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5825731714765078e-05,
+      "loss": 0.3533,
+      "step": 3711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823440189143593e-05,
+      "loss": 0.3938,
+      "step": 3712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5821148200718055e-05,
+      "loss": 0.3881,
+      "step": 3713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581885574967061e-05,
+      "loss": 0.3422,
+      "step": 3714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581656283618345e-05,
+      "loss": 0.3933,
+      "step": 3715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5814269460438796e-05,
+      "loss": 0.3506,
+      "step": 3716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581197562261891e-05,
+      "loss": 0.4052,
+      "step": 3717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5809681322906087e-05,
+      "loss": 0.3918,
+      "step": 3718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5807386561482663e-05,
+      "loss": 0.3837,
+      "step": 3719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580509133853101e-05,
+      "loss": 0.33,
+      "step": 3720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802795654233534e-05,
+      "loss": 0.4036,
+      "step": 3721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800499508772674e-05,
+      "loss": 0.3694,
+      "step": 3722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579820290233092e-05,
+      "loss": 0.4255,
+      "step": 3723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5795905835090785e-05,
+      "loss": 0.3807,
+      "step": 3724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5793608307234824e-05,
+      "loss": 0.3726,
+      "step": 3725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791310318945624e-05,
+      "loss": 0.3731,
+      "step": 3726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5789011870405816e-05,
+      "loss": 0.3909,
+      "step": 3727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578671296179806e-05,
+      "loss": 0.3786,
+      "step": 3728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5784413593305058e-05,
+      "loss": 0.3681,
+      "step": 3729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5782113765109552e-05,
+      "loss": 0.3765,
+      "step": 3730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5779813477394305e-05,
+      "loss": 0.3675,
+      "step": 3731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5777512730342136e-05,
+      "loss": 0.3691,
+      "step": 3732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5775211524135888e-05,
+      "loss": 0.3633,
+      "step": 3733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5772909858958443e-05,
+      "loss": 0.3652,
+      "step": 3734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577060773499272e-05,
+      "loss": 0.4022,
+      "step": 3735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5768305152421678e-05,
+      "loss": 0.3898,
+      "step": 3736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576600211142831e-05,
+      "loss": 0.3763,
+      "step": 3737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576369861219564e-05,
+      "loss": 0.3797,
+      "step": 3738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5761394654906733e-05,
+      "loss": 0.3781,
+      "step": 3739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759090239744696e-05,
+      "loss": 0.3934,
+      "step": 3740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575678536689266e-05,
+      "loss": 0.383,
+      "step": 3741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5754480036533807e-05,
+      "loss": 0.3928,
+      "step": 3742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575217424885134e-05,
+      "loss": 0.3634,
+      "step": 3743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574986800402851e-05,
+      "loss": 0.3899,
+      "step": 3744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747561302248604e-05,
+      "loss": 0.3998,
+      "step": 3745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5745254143694936e-05,
+      "loss": 0.3874,
+      "step": 3746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574294652855086e-05,
+      "loss": 0.3324,
+      "step": 3747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5740638456999774e-05,
+      "loss": 0.4105,
+      "step": 3748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5738329929225105e-05,
+      "loss": 0.3775,
+      "step": 3749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5736020945410316e-05,
+      "loss": 0.377,
+      "step": 3750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573371150573891e-05,
+      "loss": 0.3907,
+      "step": 3751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731401610394418e-05,
+      "loss": 0.3436,
+      "step": 3752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5729091259560426e-05,
+      "loss": 0.3695,
+      "step": 3753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572678045342053e-05,
+      "loss": 0.368,
+      "step": 3754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572446919215839e-05,
+      "loss": 0.4046,
+      "step": 3755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722157475957675e-05,
+      "loss": 0.3727,
+      "step": 3756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5719845305002105e-05,
+      "loss": 0.3449,
+      "step": 3757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571753267947544e-05,
+      "loss": 0.418,
+      "step": 3758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571521959956147e-05,
+      "loss": 0.3662,
+      "step": 3759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5712906065444018e-05,
+      "loss": 0.3588,
+      "step": 3760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571059207730695e-05,
+      "loss": 0.3451,
+      "step": 3761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5708277635334162e-05,
+      "loss": 0.3402,
+      "step": 3762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5705962739709585e-05,
+      "loss": 0.3693,
+      "step": 3763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57036473906172e-05,
+      "loss": 0.3608,
+      "step": 3764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5701331588241003e-05,
+      "loss": 0.4132,
+      "step": 3765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569901533276505e-05,
+      "loss": 0.3786,
+      "step": 3766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5696698624373406e-05,
+      "loss": 0.3641,
+      "step": 3767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569438146325019e-05,
+      "loss": 0.362,
+      "step": 3768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5692063849579563e-05,
+      "loss": 0.3498,
+      "step": 3769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5689745783545696e-05,
+      "loss": 0.4161,
+      "step": 3770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568742726533282e-05,
+      "loss": 0.3587,
+      "step": 3771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5685108295125197e-05,
+      "loss": 0.3772,
+      "step": 3772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682788873107114e-05,
+      "loss": 0.3931,
+      "step": 3773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568046899946291e-05,
+      "loss": 0.3766,
+      "step": 3774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567814867437694e-05,
+      "loss": 0.3773,
+      "step": 3775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5675827898033615e-05,
+      "loss": 0.3362,
+      "step": 3776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567350667061737e-05,
+      "loss": 0.3993,
+      "step": 3777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671184992312687e-05,
+      "loss": 0.3699,
+      "step": 3778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5668862863304065e-05,
+      "loss": 0.414,
+      "step": 3779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5666540283776054e-05,
+      "loss": 0.3754,
+      "step": 3780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566421725391323e-05,
+      "loss": 0.3718,
+      "step": 3781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5661893773900225e-05,
+      "loss": 0.39,
+      "step": 3782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659569843921677e-05,
+      "loss": 0.3577,
+      "step": 3783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5657245464162284e-05,
+      "loss": 0.4157,
+      "step": 3784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565492063480676e-05,
+      "loss": 0.3878,
+      "step": 3785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5652595356039877e-05,
+      "loss": 0.3815,
+      "step": 3786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5650269628046423e-05,
+      "loss": 0.3786,
+      "step": 3787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564794345101124e-05,
+      "loss": 0.3739,
+      "step": 3788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564561682511918e-05,
+      "loss": 0.3546,
+      "step": 3789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564328975055516e-05,
+      "loss": 0.3756,
+      "step": 3790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640962227504113e-05,
+      "loss": 0.3725,
+      "step": 3791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563863425615101e-05,
+      "loss": 0.4047,
+      "step": 3792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5636305836680865e-05,
+      "loss": 0.4006,
+      "step": 3793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5633976969278725e-05,
+      "loss": 0.3575,
+      "step": 3794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563164765412967e-05,
+      "loss": 0.3557,
+      "step": 3795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562931789141881e-05,
+      "loss": 0.4095,
+      "step": 3796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562698768133131e-05,
+      "loss": 0.3915,
+      "step": 3797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5624657024052346e-05,
+      "loss": 0.3528,
+      "step": 3798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562232591976715e-05,
+      "loss": 0.3631,
+      "step": 3799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5619994368660977e-05,
+      "loss": 0.3758,
+      "step": 3800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5617662370919123e-05,
+      "loss": 0.3724,
+      "step": 3801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5615329926726915e-05,
+      "loss": 0.3569,
+      "step": 3802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561299703626972e-05,
+      "loss": 0.3546,
+      "step": 3803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5610663699732944e-05,
+      "loss": 0.4159,
+      "step": 3804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560832991730201e-05,
+      "loss": 0.3349,
+      "step": 3805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5605995689162407e-05,
+      "loss": 0.3782,
+      "step": 3806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5603661015499633e-05,
+      "loss": 0.3737,
+      "step": 3807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5601325896499232e-05,
+      "loss": 0.3296,
+      "step": 3808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5598990332346777e-05,
+      "loss": 0.3696,
+      "step": 3809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596654323227888e-05,
+      "loss": 0.3606,
+      "step": 3810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5594317869328214e-05,
+      "loss": 0.381,
+      "step": 3811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559198097083343e-05,
+      "loss": 0.3555,
+      "step": 3812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558964362792927e-05,
+      "loss": 0.3701,
+      "step": 3813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587305840801478e-05,
+      "loss": 0.3964,
+      "step": 3814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5584967609635847e-05,
+      "loss": 0.4006,
+      "step": 3815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55826289346182e-05,
+      "loss": 0.3919,
+      "step": 3816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55802898159344e-05,
+      "loss": 0.3603,
+      "step": 3817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5577950253770348e-05,
+      "loss": 0.3791,
+      "step": 3818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557561024831197e-05,
+      "loss": 0.3937,
+      "step": 3819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5573269799745226e-05,
+      "loss": 0.3714,
+      "step": 3820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557092890825613e-05,
+      "loss": 0.4279,
+      "step": 3821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556858757403071e-05,
+      "loss": 0.3498,
+      "step": 3822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5566245797255043e-05,
+      "loss": 0.3711,
+      "step": 3823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556390357811523e-05,
+      "loss": 0.369,
+      "step": 3824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556156091679742e-05,
+      "loss": 0.3493,
+      "step": 3825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555921781348779e-05,
+      "loss": 0.3543,
+      "step": 3826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555687426837255e-05,
+      "loss": 0.3759,
+      "step": 3827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555453028163794e-05,
+      "loss": 0.3645,
+      "step": 3828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555218585347026e-05,
+      "loss": 0.358,
+      "step": 3829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554984098405581e-05,
+      "loss": 0.3913,
+      "step": 3830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5547495673580962e-05,
+      "loss": 0.3857,
+      "step": 3831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5545149922232087e-05,
+      "loss": 0.3807,
+      "step": 3832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5542803730195616e-05,
+      "loss": 0.3969,
+      "step": 3833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5540457097658003e-05,
+      "loss": 0.3914,
+      "step": 3834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5538110024805746e-05,
+      "loss": 0.3718,
+      "step": 3835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5535762511825368e-05,
+      "loss": 0.4233,
+      "step": 3836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5533414558903436e-05,
+      "loss": 0.3548,
+      "step": 3837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5531066166226548e-05,
+      "loss": 0.3835,
+      "step": 3838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5528717333981333e-05,
+      "loss": 0.3556,
+      "step": 3839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5526368062354463e-05,
+      "loss": 0.3488,
+      "step": 3840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5524018351532637e-05,
+      "loss": 0.3884,
+      "step": 3841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5521668201702597e-05,
+      "loss": 0.3422,
+      "step": 3842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5519317613051113e-05,
+      "loss": 0.426,
+      "step": 3843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551696658576499e-05,
+      "loss": 0.352,
+      "step": 3844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5514615120031077e-05,
+      "loss": 0.3834,
+      "step": 3845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5512263216036244e-05,
+      "loss": 0.4111,
+      "step": 3846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.550991087396741e-05,
+      "loss": 0.3865,
+      "step": 3847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.550755809401151e-05,
+      "loss": 0.3446,
+      "step": 3848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5505204876355543e-05,
+      "loss": 0.382,
+      "step": 3849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550285122118651e-05,
+      "loss": 0.3403,
+      "step": 3850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550049712869147e-05,
+      "loss": 0.3888,
+      "step": 3851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5498142599057508e-05,
+      "loss": 0.3656,
+      "step": 3852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5495787632471742e-05,
+      "loss": 0.3697,
+      "step": 3853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549343222912133e-05,
+      "loss": 0.3952,
+      "step": 3854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549107638919346e-05,
+      "loss": 0.3659,
+      "step": 3855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5488720112875357e-05,
+      "loss": 0.3548,
+      "step": 3856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5486363400354283e-05,
+      "loss": 0.3582,
+      "step": 3857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548400625181753e-05,
+      "loss": 0.4135,
+      "step": 3858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5481648667452427e-05,
+      "loss": 0.3497,
+      "step": 3859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5479290647446335e-05,
+      "loss": 0.3756,
+      "step": 3860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547693219198666e-05,
+      "loss": 0.3637,
+      "step": 3861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5474573301260826e-05,
+      "loss": 0.3772,
+      "step": 3862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54722139754563e-05,
+      "loss": 0.3592,
+      "step": 3863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469854214760592e-05,
+      "loss": 0.3733,
+      "step": 3864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5467494019361236e-05,
+      "loss": 0.3591,
+      "step": 3865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5465133389445795e-05,
+      "loss": 0.343,
+      "step": 3866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5462772325201885e-05,
+      "loss": 0.3806,
+      "step": 3867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546041082681714e-05,
+      "loss": 0.3481,
+      "step": 3868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5458048894479228e-05,
+      "loss": 0.3757,
+      "step": 3869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5455686528375876e-05,
+      "loss": 0.378,
+      "step": 3870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545332372869481e-05,
+      "loss": 0.3672,
+      "step": 3871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5450960495623818e-05,
+      "loss": 0.354,
+      "step": 3872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448596829350706e-05,
+      "loss": 0.3934,
+      "step": 3873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5446232730063324e-05,
+      "loss": 0.3685,
+      "step": 3874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5443868197949555e-05,
+      "loss": 0.3985,
+      "step": 3875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544150323319731e-05,
+      "loss": 0.3639,
+      "step": 3876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5439137835994547e-05,
+      "loss": 0.3702,
+      "step": 3877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543677200652924e-05,
+      "loss": 0.4008,
+      "step": 3878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5434405744989414e-05,
+      "loss": 0.3539,
+      "step": 3879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5432039051563118e-05,
+      "loss": 0.3343,
+      "step": 3880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5429671926438444e-05,
+      "loss": 0.3731,
+      "step": 3881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542730436980351e-05,
+      "loss": 0.3968,
+      "step": 3882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5424936381846475e-05,
+      "loss": 0.4226,
+      "step": 3883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5422567962755527e-05,
+      "loss": 0.3839,
+      "step": 3884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420199112718888e-05,
+      "loss": 0.3876,
+      "step": 3885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5417829831924823e-05,
+      "loss": 0.3571,
+      "step": 3886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5415460120561624e-05,
+      "loss": 0.3768,
+      "step": 3887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5413089978817613e-05,
+      "loss": 0.3676,
+      "step": 3888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5410719406881153e-05,
+      "loss": 0.3862,
+      "step": 3889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5408348404940646e-05,
+      "loss": 0.3445,
+      "step": 3890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540597697318451e-05,
+      "loss": 0.3826,
+      "step": 3891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403605111801217e-05,
+      "loss": 0.3371,
+      "step": 3892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5401232820979262e-05,
+      "loss": 0.3711,
+      "step": 3893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5398860100907184e-05,
+      "loss": 0.3911,
+      "step": 3894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539648695177354e-05,
+      "loss": 0.3997,
+      "step": 3895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5394113373766936e-05,
+      "loss": 0.3886,
+      "step": 3896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391739367076006e-05,
+      "loss": 0.4112,
+      "step": 3897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389364931889413e-05,
+      "loss": 0.3342,
+      "step": 3898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386990068395867e-05,
+      "loss": 0.3496,
+      "step": 3899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53846147767841e-05,
+      "loss": 0.3758,
+      "step": 3900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538223905724289e-05,
+      "loss": 0.3682,
+      "step": 3901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5379862909961028e-05,
+      "loss": 0.3849,
+      "step": 3902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377486335127364e-05,
+      "loss": 0.3759,
+      "step": 3903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5375109332930768e-05,
+      "loss": 0.3991,
+      "step": 3904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5372731903560142e-05,
+      "loss": 0.3596,
+      "step": 3905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370354047204433e-05,
+      "loss": 0.3705,
+      "step": 3906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5367975764052613e-05,
+      "loss": 0.3941,
+      "step": 3907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536559705429369e-05,
+      "loss": 0.3596,
+      "step": 3908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5363217918116703e-05,
+      "loss": 0.358,
+      "step": 3909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5360838355710735e-05,
+      "loss": 0.3882,
+      "step": 3910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535845836726489e-05,
+      "loss": 0.3847,
+      "step": 3911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535607795296832e-05,
+      "loss": 0.3571,
+      "step": 3912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5353697113010194e-05,
+      "loss": 0.3732,
+      "step": 3913
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5351315847579725e-05,
+      "loss": 0.3662,
+      "step": 3914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5348934156866163e-05,
+      "loss": 0.3628,
+      "step": 3915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5346552041058786e-05,
+      "loss": 0.4228,
+      "step": 3916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53441695003469e-05,
+      "loss": 0.3891,
+      "step": 3917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5341786534919864e-05,
+      "loss": 0.3854,
+      "step": 3918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5339403144967045e-05,
+      "loss": 0.3882,
+      "step": 3919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533701933067787e-05,
+      "loss": 0.3875,
+      "step": 3920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5334635092241776e-05,
+      "loss": 0.38,
+      "step": 3921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5332250429848254e-05,
+      "loss": 0.3673,
+      "step": 3922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5329865343686815e-05,
+      "loss": 0.4318,
+      "step": 3923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5327479833947004e-05,
+      "loss": 0.3566,
+      "step": 3924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5325093900818412e-05,
+      "loss": 0.3432,
+      "step": 3925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322707544490648e-05,
+      "loss": 0.4054,
+      "step": 3926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5320320765153367e-05,
+      "loss": 0.3975,
+      "step": 3927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531793356299625e-05,
+      "loss": 0.3916,
+      "step": 3928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315545938209016e-05,
+      "loss": 0.3687,
+      "step": 3929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313157890981414e-05,
+      "loss": 0.358,
+      "step": 3930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531076942150323e-05,
+      "loss": 0.3603,
+      "step": 3931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5308380529964282e-05,
+      "loss": 0.3758,
+      "step": 3932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5305991216554416e-05,
+      "loss": 0.3672,
+      "step": 3933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5303601481463525e-05,
+      "loss": 0.3622,
+      "step": 3934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5301211324881525e-05,
+      "loss": 0.3982,
+      "step": 3935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5298820746998368e-05,
+      "loss": 0.4041,
+      "step": 3936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5296429748004032e-05,
+      "loss": 0.3702,
+      "step": 3937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294038328088546e-05,
+      "loss": 0.3761,
+      "step": 3938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529164648744196e-05,
+      "loss": 0.363,
+      "step": 3939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528925422625436e-05,
+      "loss": 0.3524,
+      "step": 3940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5286861544715864e-05,
+      "loss": 0.4339,
+      "step": 3941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5284468443016624e-05,
+      "loss": 0.3973,
+      "step": 3942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5282074921346827e-05,
+      "loss": 0.3535,
+      "step": 3943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527968097989669e-05,
+      "loss": 0.3879,
+      "step": 3944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527728661885647e-05,
+      "loss": 0.3776,
+      "step": 3945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5274891838416447e-05,
+      "loss": 0.3814,
+      "step": 3946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5272496638766948e-05,
+      "loss": 0.3634,
+      "step": 3947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5270101020098324e-05,
+      "loss": 0.3571,
+      "step": 3948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526770498260096e-05,
+      "loss": 0.3724,
+      "step": 3949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526530852646527e-05,
+      "loss": 0.3505,
+      "step": 3950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526291165188172e-05,
+      "loss": 0.3988,
+      "step": 3951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526051435904078e-05,
+      "loss": 0.3829,
+      "step": 3952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5258116648132976e-05,
+      "loss": 0.3935,
+      "step": 3953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5255718519348866e-05,
+      "loss": 0.3844,
+      "step": 3954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525331997287903e-05,
+      "loss": 0.3526,
+      "step": 3955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525092100891408e-05,
+      "loss": 0.347,
+      "step": 3956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248521627644684e-05,
+      "loss": 0.3959,
+      "step": 3957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5246121829261515e-05,
+      "loss": 0.3916,
+      "step": 3958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5243721613955296e-05,
+      "loss": 0.3475,
+      "step": 3959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5241320981916778e-05,
+      "loss": 0.3594,
+      "step": 3960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5238919933336742e-05,
+      "loss": 0.3807,
+      "step": 3961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5236518468406008e-05,
+      "loss": 0.4003,
+      "step": 3962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5234116587315428e-05,
+      "loss": 0.3678,
+      "step": 3963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5231714290255885e-05,
+      "loss": 0.37,
+      "step": 3964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5229311577418296e-05,
+      "loss": 0.3848,
+      "step": 3965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5226908448993609e-05,
+      "loss": 0.3881,
+      "step": 3966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522450490517281e-05,
+      "loss": 0.3464,
+      "step": 3967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5222100946146911e-05,
+      "loss": 0.3389,
+      "step": 3968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5219696572106966e-05,
+      "loss": 0.418,
+      "step": 3969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217291783244053e-05,
+      "loss": 0.3857,
+      "step": 3970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214886579749286e-05,
+      "loss": 0.3331,
+      "step": 3971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5212480961813815e-05,
+      "loss": 0.349,
+      "step": 3972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5210074929628827e-05,
+      "loss": 0.3846,
+      "step": 3973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520766848338552e-05,
+      "loss": 0.3501,
+      "step": 3974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5205261623275155e-05,
+      "loss": 0.3791,
+      "step": 3975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202854349489007e-05,
+      "loss": 0.3914,
+      "step": 3976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5200446662218387e-05,
+      "loss": 0.3393,
+      "step": 3977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519803856165464e-05,
+      "loss": 0.3775,
+      "step": 3978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5195630047989146e-05,
+      "loss": 0.3539,
+      "step": 3979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5193221121413313e-05,
+      "loss": 0.3884,
+      "step": 3980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519081178211859e-05,
+      "loss": 0.4055,
+      "step": 3981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188402030296448e-05,
+      "loss": 0.3795,
+      "step": 3982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185991866138396e-05,
+      "loss": 0.3713,
+      "step": 3983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5183581289835982e-05,
+      "loss": 0.3279,
+      "step": 3984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5181170301580776e-05,
+      "loss": 0.3662,
+      "step": 3985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178758901564386e-05,
+      "loss": 0.3535,
+      "step": 3986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5176347089978456e-05,
+      "loss": 0.3886,
+      "step": 3987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173934867014653e-05,
+      "loss": 0.3809,
+      "step": 3988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171522232864686e-05,
+      "loss": 0.3759,
+      "step": 3989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169109187720294e-05,
+      "loss": 0.3607,
+      "step": 3990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166695731773247e-05,
+      "loss": 0.4055,
+      "step": 3991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516428186521535e-05,
+      "loss": 0.3745,
+      "step": 3992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161867588238437e-05,
+      "loss": 0.3644,
+      "step": 3993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5159452901034377e-05,
+      "loss": 0.3692,
+      "step": 3994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5157037803795077e-05,
+      "loss": 0.3492,
+      "step": 3995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515462229671246e-05,
+      "loss": 0.3436,
+      "step": 3996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.51522063799785e-05,
+      "loss": 0.3819,
+      "step": 3997
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5149790053785199e-05,
+      "loss": 0.3648,
+      "step": 3998
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5147373318324588e-05,
+      "loss": 0.3606,
+      "step": 3999
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5144956173788722e-05,
+      "loss": 0.3751,
+      "step": 4000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5142538620369709e-05,
+      "loss": 0.3624,
+      "step": 4001
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5140120658259673e-05,
+      "loss": 0.3845,
+      "step": 4002
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5137702287650777e-05,
+      "loss": 0.3452,
+      "step": 4003
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135283508735214e-05,
+      "loss": 0.3839,
+      "step": 4004
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5132864321705213e-05,
+      "loss": 0.3724,
+      "step": 4005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5130444726753029e-05,
+      "loss": 0.3667,
+      "step": 4006
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128024724070955e-05,
+      "loss": 0.3472,
+      "step": 4007
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5125604313851318e-05,
+      "loss": 0.4056,
+      "step": 4008
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123183496286473e-05,
+      "loss": 0.3915,
+      "step": 4009
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120762271568809e-05,
+      "loss": 0.3614,
+      "step": 4010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118340639890745e-05,
+      "loss": 0.3946,
+      "step": 4011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5115918601444736e-05,
+      "loss": 0.3903,
+      "step": 4012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5113496156423271e-05,
+      "loss": 0.4019,
+      "step": 4013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511107330501886e-05,
+      "loss": 0.3573,
+      "step": 4014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108650047424056e-05,
+      "loss": 0.4158,
+      "step": 4015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5106226383831448e-05,
+      "loss": 0.3778,
+      "step": 4016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5103802314433646e-05,
+      "loss": 0.3841,
+      "step": 4017
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5101377839423298e-05,
+      "loss": 0.3722,
+      "step": 4018
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098952958993079e-05,
+      "loss": 0.3808,
+      "step": 4019
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5096527673335706e-05,
+      "loss": 0.3816,
+      "step": 4020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094101982643921e-05,
+      "loss": 0.3436,
+      "step": 4021
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5091675887110504e-05,
+      "loss": 0.3912,
+      "step": 4022
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5089249386928255e-05,
+      "loss": 0.3723,
+      "step": 4023
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086822482290023e-05,
+      "loss": 0.3936,
+      "step": 4024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5084395173388671e-05,
+      "loss": 0.3644,
+      "step": 4025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5081967460417114e-05,
+      "loss": 0.3532,
+      "step": 4026
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5079539343568281e-05,
+      "loss": 0.3922,
+      "step": 4027
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5077110823035144e-05,
+      "loss": 0.3636,
+      "step": 4028
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5074681899010703e-05,
+      "loss": 0.4197,
+      "step": 4029
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5072252571687991e-05,
+      "loss": 0.3559,
+      "step": 4030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5069822841260075e-05,
+      "loss": 0.3613,
+      "step": 4031
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5067392707920048e-05,
+      "loss": 0.3686,
+      "step": 4032
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064962171861042e-05,
+      "loss": 0.3501,
+      "step": 4033
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5062531233276218e-05,
+      "loss": 0.3649,
+      "step": 4034
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506009989235877e-05,
+      "loss": 0.3649,
+      "step": 4035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5057668149301921e-05,
+      "loss": 0.3885,
+      "step": 4036
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055236004298927e-05,
+      "loss": 0.3924,
+      "step": 4037
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052803457543077e-05,
+      "loss": 0.3633,
+      "step": 4038
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5050370509227694e-05,
+      "loss": 0.3618,
+      "step": 4039
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047937159546135e-05,
+      "loss": 0.3522,
+      "step": 4040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045503408691776e-05,
+      "loss": 0.3975,
+      "step": 4041
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043069256858039e-05,
+      "loss": 0.3551,
+      "step": 4042
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040634704238369e-05,
+      "loss": 0.35,
+      "step": 4043
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038199751026249e-05,
+      "loss": 0.399,
+      "step": 4044
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503576439741519e-05,
+      "loss": 0.3457,
+      "step": 4045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5033328643598734e-05,
+      "loss": 0.3658,
+      "step": 4046
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5030892489770463e-05,
+      "loss": 0.3808,
+      "step": 4047
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028455936123978e-05,
+      "loss": 0.3571,
+      "step": 4048
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502601898285292e-05,
+      "loss": 0.3642,
+      "step": 4049
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023581630150963e-05,
+      "loss": 0.3419,
+      "step": 4050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021143878211803e-05,
+      "loss": 0.3967,
+      "step": 4051
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018705727229184e-05,
+      "loss": 0.3849,
+      "step": 4052
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5016267177396863e-05,
+      "loss": 0.3951,
+      "step": 4053
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013828228908648e-05,
+      "loss": 0.4048,
+      "step": 4054
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5011388881958357e-05,
+      "loss": 0.378,
+      "step": 4055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5008949136739861e-05,
+      "loss": 0.3725,
+      "step": 4056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5006508993447049e-05,
+      "loss": 0.3846,
+      "step": 4057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5004068452273847e-05,
+      "loss": 0.3796,
+      "step": 4058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5001627513414207e-05,
+      "loss": 0.3788,
+      "step": 4059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4999186177062121e-05,
+      "loss": 0.3581,
+      "step": 4060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4996744443411606e-05,
+      "loss": 0.3803,
+      "step": 4061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4994302312656717e-05,
+      "loss": 0.3733,
+      "step": 4062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4991859784991532e-05,
+      "loss": 0.3474,
+      "step": 4063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989416860610167e-05,
+      "loss": 0.3561,
+      "step": 4064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986973539706766e-05,
+      "loss": 0.3885,
+      "step": 4065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4984529822475508e-05,
+      "loss": 0.3609,
+      "step": 4066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49820857091106e-05,
+      "loss": 0.3615,
+      "step": 4067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4979641199806285e-05,
+      "loss": 0.3302,
+      "step": 4068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977196294756832e-05,
+      "loss": 0.3843,
+      "step": 4069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4974750994156545e-05,
+      "loss": 0.3893,
+      "step": 4070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497230529819976e-05,
+      "loss": 0.3475,
+      "step": 4071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.496985920708084e-05,
+      "loss": 0.335,
+      "step": 4072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967412720994178e-05,
+      "loss": 0.3562,
+      "step": 4073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4964965840134214e-05,
+      "loss": 0.3656,
+      "step": 4074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49625185646954e-05,
+      "loss": 0.3756,
+      "step": 4075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4960070894872232e-05,
+      "loss": 0.4364,
+      "step": 4076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4957622830859228e-05,
+      "loss": 0.3682,
+      "step": 4077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4955174372850945e-05,
+      "loss": 0.3578,
+      "step": 4078
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4952725521041963e-05,
+      "loss": 0.3833,
+      "step": 4079
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4950276275626912e-05,
+      "loss": 0.4057,
+      "step": 4080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947826636800424e-05,
+      "loss": 0.413,
+      "step": 4081
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494537660475719e-05,
+      "loss": 0.3822,
+      "step": 4082
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4942926179691912e-05,
+      "loss": 0.3955,
+      "step": 4083
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4940475361799338e-05,
+      "loss": 0.3832,
+      "step": 4084
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4938024151274237e-05,
+      "loss": 0.4048,
+      "step": 4085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935572548311416e-05,
+      "loss": 0.3611,
+      "step": 4086
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933120553105709e-05,
+      "loss": 0.3357,
+      "step": 4087
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493066816585198e-05,
+      "loss": 0.3634,
+      "step": 4088
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4928215386745133e-05,
+      "loss": 0.3425,
+      "step": 4089
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4925762215980091e-05,
+      "loss": 0.3669,
+      "step": 4090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4923308653751813e-05,
+      "loss": 0.3813,
+      "step": 4091
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4920854700255296e-05,
+      "loss": 0.3938,
+      "step": 4092
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918400355685561e-05,
+      "loss": 0.3881,
+      "step": 4093
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4915945620237657e-05,
+      "loss": 0.4135,
+      "step": 4094
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913490494106668e-05,
+      "loss": 0.3694,
+      "step": 4095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4911034977487713e-05,
+      "loss": 0.3532,
+      "step": 4096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4908579070575936e-05,
+      "loss": 0.3323,
+      "step": 4097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490612277356652e-05,
+      "loss": 0.3917,
+      "step": 4098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4903666086654667e-05,
+      "loss": 0.4254,
+      "step": 4099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4901209010035616e-05,
+      "loss": 0.386,
+      "step": 4100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489875154390464e-05,
+      "loss": 0.3947,
+      "step": 4101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896293688457037e-05,
+      "loss": 0.3653,
+      "step": 4102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4893835443888147e-05,
+      "loss": 0.4132,
+      "step": 4103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891376810393325e-05,
+      "loss": 0.3386,
+      "step": 4104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4888917788167971e-05,
+      "loss": 0.3472,
+      "step": 4105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4886458377407504e-05,
+      "loss": 0.4096,
+      "step": 4106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883998578307384e-05,
+      "loss": 0.3784,
+      "step": 4107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4881538391063094e-05,
+      "loss": 0.3619,
+      "step": 4108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4879077815870158e-05,
+      "loss": 0.3634,
+      "step": 4109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487661685292412e-05,
+      "loss": 0.3458,
+      "step": 4110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874155502420558e-05,
+      "loss": 0.3758,
+      "step": 4111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4871693764555084e-05,
+      "loss": 0.3689,
+      "step": 4112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869231639523341e-05,
+      "loss": 0.3538,
+      "step": 4113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486676912752099e-05,
+      "loss": 0.3674,
+      "step": 4114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4864306228743749e-05,
+      "loss": 0.389,
+      "step": 4115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861842943387338e-05,
+      "loss": 0.3596,
+      "step": 4116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4859379271647532e-05,
+      "loss": 0.3847,
+      "step": 4117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4856915213720115e-05,
+      "loss": 0.3961,
+      "step": 4118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4854450769800915e-05,
+      "loss": 0.3981,
+      "step": 4119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851985940085791e-05,
+      "loss": 0.3398,
+      "step": 4120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484952072477063e-05,
+      "loss": 0.3419,
+      "step": 4121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847055124051345e-05,
+      "loss": 0.3421,
+      "step": 4122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4844589138123886e-05,
+      "loss": 0.3687,
+      "step": 4123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842122767184232e-05,
+      "loss": 0.3412,
+      "step": 4124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483965601142839e-05,
+      "loss": 0.3668,
+      "step": 4125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837188871052399e-05,
+      "loss": 0.3511,
+      "step": 4126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4834721346252332e-05,
+      "loss": 0.3894,
+      "step": 4127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4832253437224289e-05,
+      "loss": 0.3821,
+      "step": 4128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48297851441644e-05,
+      "loss": 0.371,
+      "step": 4129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4827316467268829e-05,
+      "loss": 0.3575,
+      "step": 4130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4824847406733762e-05,
+      "loss": 0.3687,
+      "step": 4131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482237796275543e-05,
+      "loss": 0.34,
+      "step": 4132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481990813553008e-05,
+      "loss": 0.3981,
+      "step": 4133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817437925254e-05,
+      "loss": 0.3615,
+      "step": 4134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4814967332123502e-05,
+      "loss": 0.3635,
+      "step": 4135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481249635633493e-05,
+      "loss": 0.3854,
+      "step": 4136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4810024998084656e-05,
+      "loss": 0.3746,
+      "step": 4137
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807553257569094e-05,
+      "loss": 0.3289,
+      "step": 4138
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4805081134984673e-05,
+      "loss": 0.3638,
+      "step": 4139
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480260863052786e-05,
+      "loss": 0.3459,
+      "step": 4140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4800135744395153e-05,
+      "loss": 0.3999,
+      "step": 4141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4797662476783077e-05,
+      "loss": 0.377,
+      "step": 4142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4795188827888187e-05,
+      "loss": 0.3782,
+      "step": 4143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4792714797907077e-05,
+      "loss": 0.3624,
+      "step": 4144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479024038703636e-05,
+      "loss": 0.3616,
+      "step": 4145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4787765595472685e-05,
+      "loss": 0.3627,
+      "step": 4146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4785290423412729e-05,
+      "loss": 0.3648,
+      "step": 4147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782814871053201e-05,
+      "loss": 0.4032,
+      "step": 4148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780338938590841e-05,
+      "loss": 0.35,
+      "step": 4149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777862626222414e-05,
+      "loss": 0.3505,
+      "step": 4150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4775385934144726e-05,
+      "loss": 0.4072,
+      "step": 4151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4772908862554598e-05,
+      "loss": 0.3919,
+      "step": 4152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4770431411648898e-05,
+      "loss": 0.3322,
+      "step": 4153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4767953581624507e-05,
+      "loss": 0.3786,
+      "step": 4154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476547537267835e-05,
+      "loss": 0.4115,
+      "step": 4155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4762996785007376e-05,
+      "loss": 0.3811,
+      "step": 4156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760517818808565e-05,
+      "loss": 0.3415,
+      "step": 4157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4758038474278923e-05,
+      "loss": 0.3579,
+      "step": 4158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755558751615495e-05,
+      "loss": 0.3752,
+      "step": 4159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4753078651015347e-05,
+      "loss": 0.3463,
+      "step": 4160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4750598172675585e-05,
+      "loss": 0.3793,
+      "step": 4161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4748117316793333e-05,
+      "loss": 0.338,
+      "step": 4162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4745636083565755e-05,
+      "loss": 0.3954,
+      "step": 4163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4743154473190037e-05,
+      "loss": 0.3804,
+      "step": 4164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740672485863405e-05,
+      "loss": 0.3789,
+      "step": 4165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738190121783102e-05,
+      "loss": 0.3554,
+      "step": 4166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4735707381146416e-05,
+      "loss": 0.3669,
+      "step": 4167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733224264150649e-05,
+      "loss": 0.3706,
+      "step": 4168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4730740770993147e-05,
+      "loss": 0.3843,
+      "step": 4169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4728256901871275e-05,
+      "loss": 0.3884,
+      "step": 4170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725772656982438e-05,
+      "loss": 0.3649,
+      "step": 4171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723288036524058e-05,
+      "loss": 0.3773,
+      "step": 4172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.47208030406936e-05,
+      "loss": 0.3657,
+      "step": 4173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4718317669688554e-05,
+      "loss": 0.3612,
+      "step": 4174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4715831923706434e-05,
+      "loss": 0.3444,
+      "step": 4175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4713345802944794e-05,
+      "loss": 0.376,
+      "step": 4176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4710859307601207e-05,
+      "loss": 0.3771,
+      "step": 4177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708372437873282e-05,
+      "loss": 0.3643,
+      "step": 4178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705885193958664e-05,
+      "loss": 0.3624,
+      "step": 4179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703397576055014e-05,
+      "loss": 0.3621,
+      "step": 4180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4700909584360032e-05,
+      "loss": 0.345,
+      "step": 4181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698421219071444e-05,
+      "loss": 0.3859,
+      "step": 4182
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469593248038701e-05,
+      "loss": 0.3746,
+      "step": 4183
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4693443368504513e-05,
+      "loss": 0.3548,
+      "step": 4184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4690953883621767e-05,
+      "loss": 0.3788,
+      "step": 4185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688464025936625e-05,
+      "loss": 0.3834,
+      "step": 4186
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4685973795646957e-05,
+      "loss": 0.369,
+      "step": 4187
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4683483192950671e-05,
+      "loss": 0.3854,
+      "step": 4188
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.46809922180457e-05,
+      "loss": 0.3725,
+      "step": 4189
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467850087113001e-05,
+      "loss": 0.4085,
+      "step": 4190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4676009152401594e-05,
+      "loss": 0.3696,
+      "step": 4191
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4673517062058476e-05,
+      "loss": 0.3855,
+      "step": 4192
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4671024600298709e-05,
+      "loss": 0.3669,
+      "step": 4193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466853176732037e-05,
+      "loss": 0.3553,
+      "step": 4194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666038563321578e-05,
+      "loss": 0.3791,
+      "step": 4195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4663544988500473e-05,
+      "loss": 0.3523,
+      "step": 4196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4661051043055225e-05,
+      "loss": 0.3717,
+      "step": 4197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4658556727184037e-05,
+      "loss": 0.3493,
+      "step": 4198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4656062041085133e-05,
+      "loss": 0.3515,
+      "step": 4199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4653566984956777e-05,
+      "loss": 0.3679,
+      "step": 4200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651071558997254e-05,
+      "loss": 0.3949,
+      "step": 4201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4648575763404887e-05,
+      "loss": 0.3679,
+      "step": 4202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646079598378017e-05,
+      "loss": 0.3751,
+      "step": 4203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4643583064115027e-05,
+      "loss": 0.3644,
+      "step": 4204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4641086160814322e-05,
+      "loss": 0.3522,
+      "step": 4205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4638588888674335e-05,
+      "loss": 0.3897,
+      "step": 4206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636091247893532e-05,
+      "loss": 0.3515,
+      "step": 4207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4633593238670406e-05,
+      "loss": 0.3565,
+      "step": 4208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4631094861203478e-05,
+      "loss": 0.3785,
+      "step": 4209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628596115691309e-05,
+      "loss": 0.3579,
+      "step": 4210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4626097002332475e-05,
+      "loss": 0.3764,
+      "step": 4211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4623597521325586e-05,
+      "loss": 0.3704,
+      "step": 4212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4621097672869282e-05,
+      "loss": 0.3648,
+      "step": 4213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4618597457162236e-05,
+      "loss": 0.3496,
+      "step": 4214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4616096874403144e-05,
+      "loss": 0.392,
+      "step": 4215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461359592479074e-05,
+      "loss": 0.3646,
+      "step": 4216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611094608523772e-05,
+      "loss": 0.3758,
+      "step": 4217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4608592925801032e-05,
+      "loss": 0.3911,
+      "step": 4218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460609087682133e-05,
+      "loss": 0.3598,
+      "step": 4219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460358846178352e-05,
+      "loss": 0.3707,
+      "step": 4220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601085680886466e-05,
+      "loss": 0.3765,
+      "step": 4221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4598582534329075e-05,
+      "loss": 0.3818,
+      "step": 4222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4596079022310277e-05,
+      "loss": 0.414,
+      "step": 4223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593575145029032e-05,
+      "loss": 0.3678,
+      "step": 4224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4591070902684333e-05,
+      "loss": 0.394,
+      "step": 4225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4588566295475195e-05,
+      "loss": 0.3956,
+      "step": 4226
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4586061323600666e-05,
+      "loss": 0.3865,
+      "step": 4227
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583555987259828e-05,
+      "loss": 0.3729,
+      "step": 4228
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458105028665178e-05,
+      "loss": 0.3623,
+      "step": 4229
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578544221975661e-05,
+      "loss": 0.3616,
+      "step": 4230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457603779343063e-05,
+      "loss": 0.391,
+      "step": 4231
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4573531001215884e-05,
+      "loss": 0.3543,
+      "step": 4232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571023845530642e-05,
+      "loss": 0.3683,
+      "step": 4233
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568516326574158e-05,
+      "loss": 0.359,
+      "step": 4234
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566008444545706e-05,
+      "loss": 0.3642,
+      "step": 4235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4563500199644592e-05,
+      "loss": 0.3588,
+      "step": 4236
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4560991592070159e-05,
+      "loss": 0.3944,
+      "step": 4237
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558482622021774e-05,
+      "loss": 0.3443,
+      "step": 4238
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4555973289698826e-05,
+      "loss": 0.3841,
+      "step": 4239
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4553463595300738e-05,
+      "loss": 0.3505,
+      "step": 4240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4550953539026965e-05,
+      "loss": 0.4286,
+      "step": 4241
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4548443121076986e-05,
+      "loss": 0.4025,
+      "step": 4242
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454593234165031e-05,
+      "loss": 0.3609,
+      "step": 4243
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4543421200946479e-05,
+      "loss": 0.3555,
+      "step": 4244
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4540909699165055e-05,
+      "loss": 0.3886,
+      "step": 4245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4538397836505637e-05,
+      "loss": 0.3793,
+      "step": 4246
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4535885613167848e-05,
+      "loss": 0.3772,
+      "step": 4247
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4533373029351343e-05,
+      "loss": 0.3667,
+      "step": 4248
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45308600852558e-05,
+      "loss": 0.3346,
+      "step": 4249
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528346781080931e-05,
+      "loss": 0.4166,
+      "step": 4250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4525833117026476e-05,
+      "loss": 0.4038,
+      "step": 4251
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4523319093292203e-05,
+      "loss": 0.3855,
+      "step": 4252
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4520804710077902e-05,
+      "loss": 0.3609,
+      "step": 4253
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518289967583406e-05,
+      "loss": 0.4073,
+      "step": 4254
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515774866008564e-05,
+      "loss": 0.3853,
+      "step": 4255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513259405553258e-05,
+      "loss": 0.3678,
+      "step": 4256
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510743586417397e-05,
+      "loss": 0.3633,
+      "step": 4257
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4508227408800921e-05,
+      "loss": 0.35,
+      "step": 4258
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505710872903795e-05,
+      "loss": 0.351,
+      "step": 4259
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450319397892602e-05,
+      "loss": 0.3792,
+      "step": 4260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4500676727067618e-05,
+      "loss": 0.3697,
+      "step": 4261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4498159117528638e-05,
+      "loss": 0.3699,
+      "step": 4262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495641150509165e-05,
+      "loss": 0.3952,
+      "step": 4263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493122826209307e-05,
+      "loss": 0.338,
+      "step": 4264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490604144829204e-05,
+      "loss": 0.3717,
+      "step": 4265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4488085106569017e-05,
+      "loss": 0.362,
+      "step": 4266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4485565711628944e-05,
+      "loss": 0.3633,
+      "step": 4267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448304596020921e-05,
+      "loss": 0.3703,
+      "step": 4268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4480525852510062e-05,
+      "loss": 0.3682,
+      "step": 4269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478005388731785e-05,
+      "loss": 0.3792,
+      "step": 4270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4475484569074678e-05,
+      "loss": 0.3546,
+      "step": 4271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4472963393739085e-05,
+      "loss": 0.3651,
+      "step": 4272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4470441862925369e-05,
+      "loss": 0.3554,
+      "step": 4273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4467919976833922e-05,
+      "loss": 0.4029,
+      "step": 4274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4465397735665163e-05,
+      "loss": 0.3733,
+      "step": 4275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4462875139619543e-05,
+      "loss": 0.4128,
+      "step": 4276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460352188897536e-05,
+      "loss": 0.3835,
+      "step": 4277
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4457828883699652e-05,
+      "loss": 0.3715,
+      "step": 4278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4455305224226429e-05,
+      "loss": 0.3602,
+      "step": 4279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452781210678417e-05,
+      "loss": 0.39,
+      "step": 4280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4450256843256211e-05,
+      "loss": 0.3542,
+      "step": 4281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4447732122160431e-05,
+      "loss": 0.3564,
+      "step": 4282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4445207047591725e-05,
+      "loss": 0.3548,
+      "step": 4283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442681619750763e-05,
+      "loss": 0.3642,
+      "step": 4284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4440155838838249e-05,
+      "loss": 0.3192,
+      "step": 4285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4437629705054912e-05,
+      "loss": 0.3541,
+      "step": 4286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443510321860151e-05,
+      "loss": 0.332,
+      "step": 4287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432576379678835e-05,
+      "loss": 0.3592,
+      "step": 4288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430049188487699e-05,
+      "loss": 0.38,
+      "step": 4289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442752164522894e-05,
+      "loss": 0.4046,
+      "step": 4290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4424993750103432e-05,
+      "loss": 0.3472,
+      "step": 4291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4422465503312074e-05,
+      "loss": 0.3976,
+      "step": 4292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4419936905055794e-05,
+      "loss": 0.4188,
+      "step": 4293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4417407955535541e-05,
+      "loss": 0.3368,
+      "step": 4294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4414878654952299e-05,
+      "loss": 0.3781,
+      "step": 4295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441234900350708e-05,
+      "loss": 0.363,
+      "step": 4296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409819001400927e-05,
+      "loss": 0.3649,
+      "step": 4297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4407288648834894e-05,
+      "loss": 0.3655,
+      "step": 4298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4404757946010084e-05,
+      "loss": 0.3648,
+      "step": 4299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4402226893127611e-05,
+      "loss": 0.3435,
+      "step": 4300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4399695490388634e-05,
+      "loss": 0.3687,
+      "step": 4301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397163737994324e-05,
+      "loss": 0.3335,
+      "step": 4302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4394631636145888e-05,
+      "loss": 0.3882,
+      "step": 4303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392099185044557e-05,
+      "loss": 0.3619,
+      "step": 4304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438956638489159e-05,
+      "loss": 0.3822,
+      "step": 4305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4387033235888281e-05,
+      "loss": 0.3933,
+      "step": 4306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4384499738235942e-05,
+      "loss": 0.3837,
+      "step": 4307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4381965892135918e-05,
+      "loss": 0.3631,
+      "step": 4308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4379431697789579e-05,
+      "loss": 0.3469,
+      "step": 4309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4376897155398326e-05,
+      "loss": 0.3544,
+      "step": 4310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4374362265163586e-05,
+      "loss": 0.3465,
+      "step": 4311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4371827027286808e-05,
+      "loss": 0.357,
+      "step": 4312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436929144196948e-05,
+      "loss": 0.3624,
+      "step": 4313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436675550941311e-05,
+      "loss": 0.3556,
+      "step": 4314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364219229819237e-05,
+      "loss": 0.3587,
+      "step": 4315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436168260338942e-05,
+      "loss": 0.3484,
+      "step": 4316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4359145630325256e-05,
+      "loss": 0.3539,
+      "step": 4317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4356608310828363e-05,
+      "loss": 0.371,
+      "step": 4318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354070645100393e-05,
+      "loss": 0.4363,
+      "step": 4319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4351532633343013e-05,
+      "loss": 0.3319,
+      "step": 4320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348994275757933e-05,
+      "loss": 0.3861,
+      "step": 4321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4346455572546875e-05,
+      "loss": 0.3454,
+      "step": 4322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4343916523911605e-05,
+      "loss": 0.3598,
+      "step": 4323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4341377130053905e-05,
+      "loss": 0.3983,
+      "step": 4324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.3715,
+      "step": 4325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4336297307478485e-05,
+      "loss": 0.3391,
+      "step": 4326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433375687916447e-05,
+      "loss": 0.3545,
+      "step": 4327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4331216106435441e-05,
+      "loss": 0.3794,
+      "step": 4328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4328674989493316e-05,
+      "loss": 0.3825,
+      "step": 4329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326133528540045e-05,
+      "loss": 0.3857,
+      "step": 4330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4323591723777603e-05,
+      "loss": 0.3842,
+      "step": 4331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4321049575407995e-05,
+      "loss": 0.3631,
+      "step": 4332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4318507083633256e-05,
+      "loss": 0.3434,
+      "step": 4333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4315964248655436e-05,
+      "loss": 0.3125,
+      "step": 4334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4313421070676626e-05,
+      "loss": 0.3671,
+      "step": 4335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310877549898939e-05,
+      "loss": 0.3954,
+      "step": 4336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4308333686524515e-05,
+      "loss": 0.3973,
+      "step": 4337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4305789480755522e-05,
+      "loss": 0.3194,
+      "step": 4338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4303244932794153e-05,
+      "loss": 0.3991,
+      "step": 4339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4300700042842632e-05,
+      "loss": 0.3346,
+      "step": 4340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4298154811103204e-05,
+      "loss": 0.3336,
+      "step": 4341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4295609237778153e-05,
+      "loss": 0.3444,
+      "step": 4342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4293063323069777e-05,
+      "loss": 0.3928,
+      "step": 4343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290517067180404e-05,
+      "loss": 0.3898,
+      "step": 4344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4287970470312393e-05,
+      "loss": 0.4054,
+      "step": 4345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285423532668134e-05,
+      "loss": 0.3591,
+      "step": 4346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4282876254450033e-05,
+      "loss": 0.3631,
+      "step": 4347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280328635860535e-05,
+      "loss": 0.3715,
+      "step": 4348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4277780677102098e-05,
+      "loss": 0.3444,
+      "step": 4349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427523237837722e-05,
+      "loss": 0.3543,
+      "step": 4350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4272683739888423e-05,
+      "loss": 0.3801,
+      "step": 4351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4270134761838245e-05,
+      "loss": 0.3523,
+      "step": 4352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4267585444429268e-05,
+      "loss": 0.3826,
+      "step": 4353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4265035787864091e-05,
+      "loss": 0.3679,
+      "step": 4354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4262485792345344e-05,
+      "loss": 0.3718,
+      "step": 4355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425993545807568e-05,
+      "loss": 0.3582,
+      "step": 4356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257384785257777e-05,
+      "loss": 0.4445,
+      "step": 4357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254833774094347e-05,
+      "loss": 0.377,
+      "step": 4358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4252282424788126e-05,
+      "loss": 0.3654,
+      "step": 4359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4249730737541878e-05,
+      "loss": 0.4014,
+      "step": 4360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247178712558389e-05,
+      "loss": 0.3303,
+      "step": 4361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244626350040477e-05,
+      "loss": 0.3874,
+      "step": 4362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4242073650190984e-05,
+      "loss": 0.3559,
+      "step": 4363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4239520613212778e-05,
+      "loss": 0.3722,
+      "step": 4364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4236967239308758e-05,
+      "loss": 0.3589,
+      "step": 4365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423441352868185e-05,
+      "loss": 0.3366,
+      "step": 4366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4231859481534998e-05,
+      "loss": 0.3559,
+      "step": 4367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229305098071182e-05,
+      "loss": 0.3594,
+      "step": 4368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4226750378493409e-05,
+      "loss": 0.3729,
+      "step": 4369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42241953230047e-05,
+      "loss": 0.3491,
+      "step": 4370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4221639931808117e-05,
+      "loss": 0.3548,
+      "step": 4371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421908420510675e-05,
+      "loss": 0.3843,
+      "step": 4372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42165281431037e-05,
+      "loss": 0.353,
+      "step": 4373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4213971746002109e-05,
+      "loss": 0.3507,
+      "step": 4374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4211415014005136e-05,
+      "loss": 0.3487,
+      "step": 4375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4208857947315974e-05,
+      "loss": 0.3904,
+      "step": 4376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4206300546137844e-05,
+      "loss": 0.3741,
+      "step": 4377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4203742810673984e-05,
+      "loss": 0.3333,
+      "step": 4378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4201184741127664e-05,
+      "loss": 0.3459,
+      "step": 4379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198626337702181e-05,
+      "loss": 0.3595,
+      "step": 4380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419606760060086e-05,
+      "loss": 0.3915,
+      "step": 4381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419350853002705e-05,
+      "loss": 0.3917,
+      "step": 4382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4190949126184125e-05,
+      "loss": 0.3682,
+      "step": 4383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418838938927549e-05,
+      "loss": 0.3821,
+      "step": 4384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4185829319504572e-05,
+      "loss": 0.3205,
+      "step": 4385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183268917074828e-05,
+      "loss": 0.3643,
+      "step": 4386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4180708182189741e-05,
+      "loss": 0.41,
+      "step": 4387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4178147115052813e-05,
+      "loss": 0.3227,
+      "step": 4388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4175585715867586e-05,
+      "loss": 0.3761,
+      "step": 4389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417302398483762e-05,
+      "loss": 0.3734,
+      "step": 4390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4170461922166499e-05,
+      "loss": 0.3793,
+      "step": 4391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4167899528057838e-05,
+      "loss": 0.3634,
+      "step": 4392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4165336802715277e-05,
+      "loss": 0.4234,
+      "step": 4393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4162773746342487e-05,
+      "loss": 0.3902,
+      "step": 4394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160210359143153e-05,
+      "loss": 0.3227,
+      "step": 4395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4157646641321003e-05,
+      "loss": 0.3475,
+      "step": 4396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4155082593079774e-05,
+      "loss": 0.3269,
+      "step": 4397
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4152518214623242e-05,
+      "loss": 0.3324,
+      "step": 4398
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4149953506155205e-05,
+      "loss": 0.3551,
+      "step": 4399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4147388467879485e-05,
+      "loss": 0.3802,
+      "step": 4400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4144823099999936e-05,
+      "loss": 0.3428,
+      "step": 4401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414225740272043e-05,
+      "loss": 0.3516,
+      "step": 4402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4139691376244873e-05,
+      "loss": 0.3403,
+      "step": 4403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137125020777192e-05,
+      "loss": 0.3628,
+      "step": 4404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4134558336521342e-05,
+      "loss": 0.3462,
+      "step": 4405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131991323681305e-05,
+      "loss": 0.3569,
+      "step": 4406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4129423982461088e-05,
+      "loss": 0.3352,
+      "step": 4407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126856313064726e-05,
+      "loss": 0.3577,
+      "step": 4408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4124288315696274e-05,
+      "loss": 0.3598,
+      "step": 4409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121719990559823e-05,
+      "loss": 0.3413,
+      "step": 4410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4119151337859479e-05,
+      "loss": 0.3653,
+      "step": 4411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4116582357799384e-05,
+      "loss": 0.3343,
+      "step": 4412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41140130505837e-05,
+      "loss": 0.3863,
+      "step": 4413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111443416416617e-05,
+      "loss": 0.3804,
+      "step": 4414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.410887345550235e-05,
+      "loss": 0.3494,
+      "step": 4415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4106303168045139e-05,
+      "loss": 0.3658,
+      "step": 4416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4103732554249252e-05,
+      "loss": 0.3457,
+      "step": 4417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101161614318986e-05,
+      "loss": 0.3816,
+      "step": 4418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4098590348458658e-05,
+      "loss": 0.3511,
+      "step": 4419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096018756872612e-05,
+      "loss": 0.3388,
+      "step": 4420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4093446839765219e-05,
+      "loss": 0.3681,
+      "step": 4421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4090874597340878e-05,
+      "loss": 0.3654,
+      "step": 4422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.408830202980401e-05,
+      "loss": 0.3567,
+      "step": 4423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4085729137359065e-05,
+      "loss": 0.3373,
+      "step": 4424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4083155920210518e-05,
+      "loss": 0.3747,
+      "step": 4425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4080582378562871e-05,
+      "loss": 0.3442,
+      "step": 4426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4078008512620645e-05,
+      "loss": 0.3532,
+      "step": 4427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4075434322588395e-05,
+      "loss": 0.3543,
+      "step": 4428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072859808670696e-05,
+      "loss": 0.3781,
+      "step": 4429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4070284971072155e-05,
+      "loss": 0.3666,
+      "step": 4430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067709809997401e-05,
+      "loss": 0.3618,
+      "step": 4431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4065134325651086e-05,
+      "loss": 0.3672,
+      "step": 4432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4062558518237893e-05,
+      "loss": 0.3651,
+      "step": 4433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4059982387962523e-05,
+      "loss": 0.3589,
+      "step": 4434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057405935029715e-05,
+      "loss": 0.4198,
+      "step": 4435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4054829159644225e-05,
+      "loss": 0.4029,
+      "step": 4436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4052252062010831e-05,
+      "loss": 0.3425,
+      "step": 4437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4049674642334348e-05,
+      "loss": 0.336,
+      "step": 4438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4047096900819606e-05,
+      "loss": 0.3162,
+      "step": 4439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4044518837671463e-05,
+      "loss": 0.3322,
+      "step": 4440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4041940453094813e-05,
+      "loss": 0.4088,
+      "step": 4441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403936174729456e-05,
+      "loss": 0.3911,
+      "step": 4442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4036782720475642e-05,
+      "loss": 0.3796,
+      "step": 4443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403420337284302e-05,
+      "loss": 0.3671,
+      "step": 4444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031623704601685e-05,
+      "loss": 0.3675,
+      "step": 4445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4029043715956645e-05,
+      "loss": 0.3563,
+      "step": 4446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4026463407112943e-05,
+      "loss": 0.3389,
+      "step": 4447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023882778275641e-05,
+      "loss": 0.3677,
+      "step": 4448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4021301829649829e-05,
+      "loss": 0.3636,
+      "step": 4449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4018720561440622e-05,
+      "loss": 0.3815,
+      "step": 4450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016138973853158e-05,
+      "loss": 0.3651,
+      "step": 4451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40135570670926e-05,
+      "loss": 0.3874,
+      "step": 4452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4010974841364146e-05,
+      "loss": 0.3882,
+      "step": 4453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400839229687301e-05,
+      "loss": 0.3459,
+      "step": 4454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4005809433824433e-05,
+      "loss": 0.3372,
+      "step": 4455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400322625242368e-05,
+      "loss": 0.3227,
+      "step": 4456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4000642752876044e-05,
+      "loss": 0.3692,
+      "step": 4457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3998058935386845e-05,
+      "loss": 0.3202,
+      "step": 4458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3995474800161424e-05,
+      "loss": 0.3646,
+      "step": 4459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992890347405148e-05,
+      "loss": 0.399,
+      "step": 4460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399030557732341e-05,
+      "loss": 0.3353,
+      "step": 4461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987720490121632e-05,
+      "loss": 0.4135,
+      "step": 4462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3985135086005256e-05,
+      "loss": 0.3978,
+      "step": 4463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982549365179745e-05,
+      "loss": 0.3701,
+      "step": 4464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3979963327850603e-05,
+      "loss": 0.3807,
+      "step": 4465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977376974223342e-05,
+      "loss": 0.3566,
+      "step": 4466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397479030450351e-05,
+      "loss": 0.3879,
+      "step": 4467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3972203318896673e-05,
+      "loss": 0.376,
+      "step": 4468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3969616017608427e-05,
+      "loss": 0.4035,
+      "step": 4469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967028400844394e-05,
+      "loss": 0.3729,
+      "step": 4470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3964440468810216e-05,
+      "loss": 0.3646,
+      "step": 4471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3961852221711565e-05,
+      "loss": 0.3857,
+      "step": 4472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959263659754135e-05,
+      "loss": 0.3449,
+      "step": 4473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3956674783143644e-05,
+      "loss": 0.3538,
+      "step": 4474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954085592085835e-05,
+      "loss": 0.3997,
+      "step": 4475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3951496086786485e-05,
+      "loss": 0.3667,
+      "step": 4476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3948906267451383e-05,
+      "loss": 0.3723,
+      "step": 4477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3946316134286351e-05,
+      "loss": 0.3843,
+      "step": 4478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943725687497231e-05,
+      "loss": 0.3582,
+      "step": 4479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3941134927289894e-05,
+      "loss": 0.3807,
+      "step": 4480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3938543853870236e-05,
+      "loss": 0.3896,
+      "step": 4481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3935952467444179e-05,
+      "loss": 0.3456,
+      "step": 4482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3933360768217658e-05,
+      "loss": 0.3375,
+      "step": 4483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393076875639665e-05,
+      "loss": 0.4125,
+      "step": 4484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3928176432187148e-05,
+      "loss": 0.3614,
+      "step": 4485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392558379579517e-05,
+      "loss": 0.3631,
+      "step": 4486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3922990847426756e-05,
+      "loss": 0.3573,
+      "step": 4487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392039758728798e-05,
+      "loss": 0.3844,
+      "step": 4488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3917804015584932e-05,
+      "loss": 0.3897,
+      "step": 4489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3915210132523732e-05,
+      "loss": 0.3643,
+      "step": 4490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3912615938310521e-05,
+      "loss": 0.3606,
+      "step": 4491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910021433151463e-05,
+      "loss": 0.3872,
+      "step": 4492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390742661725276e-05,
+      "loss": 0.3575,
+      "step": 4493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3904831490820618e-05,
+      "loss": 0.3791,
+      "step": 4494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3902236054061289e-05,
+      "loss": 0.3867,
+      "step": 4495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899640307181028e-05,
+      "loss": 0.3909,
+      "step": 4496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3897044250386134e-05,
+      "loss": 0.3688,
+      "step": 4497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894447883882916e-05,
+      "loss": 0.3802,
+      "step": 4498
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891851207877722e-05,
+      "loss": 0.3886,
+      "step": 4499
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3889254222576911e-05,
+      "loss": 0.3812,
+      "step": 4500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3886656928186874e-05,
+      "loss": 0.4006,
+      "step": 4501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884059324914023e-05,
+      "loss": 0.3664,
+      "step": 4502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3881461412964799e-05,
+      "loss": 0.345,
+      "step": 4503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3878863192545662e-05,
+      "loss": 0.3842,
+      "step": 4504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3876264663863102e-05,
+      "loss": 0.3848,
+      "step": 4505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873665827123631e-05,
+      "loss": 0.4202,
+      "step": 4506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3871066682533782e-05,
+      "loss": 0.3493,
+      "step": 4507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386846723030012e-05,
+      "loss": 0.364,
+      "step": 4508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3865867470629229e-05,
+      "loss": 0.3745,
+      "step": 4509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3863267403727713e-05,
+      "loss": 0.3568,
+      "step": 4510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3860667029802218e-05,
+      "loss": 0.4156,
+      "step": 4511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858066349059392e-05,
+      "loss": 0.392,
+      "step": 4512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855465361705922e-05,
+      "loss": 0.3436,
+      "step": 4513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3852864067948514e-05,
+      "loss": 0.3665,
+      "step": 4514
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850262467993902e-05,
+      "loss": 0.3681,
+      "step": 4515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384766056204884e-05,
+      "loss": 0.373,
+      "step": 4516
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3845058350320109e-05,
+      "loss": 0.3952,
+      "step": 4517
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3842455833014515e-05,
+      "loss": 0.3473,
+      "step": 4518
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3839853010338883e-05,
+      "loss": 0.3492,
+      "step": 4519
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383724988250007e-05,
+      "loss": 0.3682,
+      "step": 4520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3834646449704949e-05,
+      "loss": 0.3541,
+      "step": 4521
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3832042712160427e-05,
+      "loss": 0.3404,
+      "step": 4522
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3829438670073427e-05,
+      "loss": 0.4027,
+      "step": 4523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.3704,
+      "step": 4524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3824229673099819e-05,
+      "loss": 0.3456,
+      "step": 4525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382162471862718e-05,
+      "loss": 0.3576,
+      "step": 4526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3819019460440013e-05,
+      "loss": 0.3488,
+      "step": 4527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816413898745358e-05,
+      "loss": 0.3462,
+      "step": 4528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813808033750289e-05,
+      "loss": 0.3772,
+      "step": 4529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3811201865661899e-05,
+      "loss": 0.3699,
+      "step": 4530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808595394687309e-05,
+      "loss": 0.3491,
+      "step": 4531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380598862103366e-05,
+      "loss": 0.3765,
+      "step": 4532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3803381544908118e-05,
+      "loss": 0.3663,
+      "step": 4533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3800774166517879e-05,
+      "loss": 0.3703,
+      "step": 4534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3798166486070156e-05,
+      "loss": 0.3739,
+      "step": 4535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795558503772189e-05,
+      "loss": 0.4086,
+      "step": 4536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3792950219831238e-05,
+      "loss": 0.395,
+      "step": 4537
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379034163445459e-05,
+      "loss": 0.4121,
+      "step": 4538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3787732747849561e-05,
+      "loss": 0.4212,
+      "step": 4539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3785123560223484e-05,
+      "loss": 0.3503,
+      "step": 4540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3782514071783717e-05,
+      "loss": 0.3685,
+      "step": 4541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3779904282737642e-05,
+      "loss": 0.3389,
+      "step": 4542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3777294193292666e-05,
+      "loss": 0.3294,
+      "step": 4543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3774683803656221e-05,
+      "loss": 0.3555,
+      "step": 4544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3772073114035762e-05,
+      "loss": 0.3843,
+      "step": 4545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769462124638766e-05,
+      "loss": 0.3456,
+      "step": 4546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3766850835672733e-05,
+      "loss": 0.4017,
+      "step": 4547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3764239247345197e-05,
+      "loss": 0.3419,
+      "step": 4548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761627359863699e-05,
+      "loss": 0.3685,
+      "step": 4549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3759015173435816e-05,
+      "loss": 0.3588,
+      "step": 4550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756402688269145e-05,
+      "loss": 0.3509,
+      "step": 4551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3753789904571308e-05,
+      "loss": 0.3615,
+      "step": 4552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3751176822549947e-05,
+      "loss": 0.377,
+      "step": 4553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3748563442412736e-05,
+      "loss": 0.3571,
+      "step": 4554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745949764367364e-05,
+      "loss": 0.3585,
+      "step": 4555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3743335788621543e-05,
+      "loss": 0.4136,
+      "step": 4556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3740721515383015e-05,
+      "loss": 0.3962,
+      "step": 4557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3738106944859548e-05,
+      "loss": 0.3663,
+      "step": 4558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3735492077258924e-05,
+      "loss": 0.3546,
+      "step": 4559
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732876912788955e-05,
+      "loss": 0.3162,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3730261451657471e-05,
+      "loss": 0.364,
+      "step": 4561
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3727645694072336e-05,
+      "loss": 0.3447,
+      "step": 4562
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3725029640241428e-05,
+      "loss": 0.3876,
+      "step": 4563
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3722413290372649e-05,
+      "loss": 0.3957,
+      "step": 4564
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719796644673931e-05,
+      "loss": 0.3805,
+      "step": 4565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3717179703353222e-05,
+      "loss": 0.3806,
+      "step": 4566
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714562466618504e-05,
+      "loss": 0.3407,
+      "step": 4567
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371194493467777e-05,
+      "loss": 0.3558,
+      "step": 4568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3709327107739042e-05,
+      "loss": 0.3717,
+      "step": 4569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3706708986010367e-05,
+      "loss": 0.3543,
+      "step": 4570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704090569699814e-05,
+      "loss": 0.3428,
+      "step": 4571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3701471859015478e-05,
+      "loss": 0.4051,
+      "step": 4572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369885285416547e-05,
+      "loss": 0.3905,
+      "step": 4573
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696233555357933e-05,
+      "loss": 0.3754,
+      "step": 4574
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3693613962801024e-05,
+      "loss": 0.3703,
+      "step": 4575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3690994076702937e-05,
+      "loss": 0.3733,
+      "step": 4576
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3688373897271876e-05,
+      "loss": 0.354,
+      "step": 4577
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3685753424716072e-05,
+      "loss": 0.3706,
+      "step": 4578
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3683132659243788e-05,
+      "loss": 0.3495,
+      "step": 4579
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3680511601063296e-05,
+      "loss": 0.3816,
+      "step": 4580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36778902503829e-05,
+      "loss": 0.3829,
+      "step": 4581
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3675268607410925e-05,
+      "loss": 0.3717,
+      "step": 4582
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3672646672355723e-05,
+      "loss": 0.3445,
+      "step": 4583
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670024445425665e-05,
+      "loss": 0.3411,
+      "step": 4584
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3667401926829146e-05,
+      "loss": 0.3407,
+      "step": 4585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3664779116774582e-05,
+      "loss": 0.3761,
+      "step": 4586
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3662156015470419e-05,
+      "loss": 0.3519,
+      "step": 4587
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659532623125118e-05,
+      "loss": 0.3376,
+      "step": 4588
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3656908939947166e-05,
+      "loss": 0.3772,
+      "step": 4589
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3654284966145082e-05,
+      "loss": 0.4178,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651660701927392e-05,
+      "loss": 0.3392,
+      "step": 4591
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364903614750265e-05,
+      "loss": 0.3912,
+      "step": 4592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3646411303079446e-05,
+      "loss": 0.3558,
+      "step": 4593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364378616886638e-05,
+      "loss": 0.3814,
+      "step": 4594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3641160745072076e-05,
+      "loss": 0.3968,
+      "step": 4595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3638535031905184e-05,
+      "loss": 0.3582,
+      "step": 4596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3635909029574375e-05,
+      "loss": 0.3652,
+      "step": 4597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3633282738288346e-05,
+      "loss": 0.3452,
+      "step": 4598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3630656158255822e-05,
+      "loss": 0.3536,
+      "step": 4599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3628029289685532e-05,
+      "loss": 0.3983,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3625402132786247e-05,
+      "loss": 0.3487,
+      "step": 4601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622774687766752e-05,
+      "loss": 0.3891,
+      "step": 4602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3620146954835858e-05,
+      "loss": 0.3634,
+      "step": 4603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3617518934202396e-05,
+      "loss": 0.3611,
+      "step": 4604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614890626075226e-05,
+      "loss": 0.3649,
+      "step": 4605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361226203066322e-05,
+      "loss": 0.38,
+      "step": 4606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3609633148175288e-05,
+      "loss": 0.3519,
+      "step": 4607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3607003978820347e-05,
+      "loss": 0.3397,
+      "step": 4608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3604374522807345e-05,
+      "loss": 0.3499,
+      "step": 4609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3601744780345251e-05,
+      "loss": 0.3309,
+      "step": 4610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3599114751643063e-05,
+      "loss": 0.376,
+      "step": 4611
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3596484436909791e-05,
+      "loss": 0.3787,
+      "step": 4612
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593853836354481e-05,
+      "loss": 0.3693,
+      "step": 4613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359122295018618e-05,
+      "loss": 0.3394,
+      "step": 4614
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358859177861398e-05,
+      "loss": 0.3542,
+      "step": 4615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3585960321846988e-05,
+      "loss": 0.3901,
+      "step": 4616
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358332858009433e-05,
+      "loss": 0.3901,
+      "step": 4617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580696553565158e-05,
+      "loss": 0.3158,
+      "step": 4618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3578064242468647e-05,
+      "loss": 0.3463,
+      "step": 4619
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3575431647013993e-05,
+      "loss": 0.3795,
+      "step": 4620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3572798767410413e-05,
+      "loss": 0.3679,
+      "step": 4621
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3570165603867154e-05,
+      "loss": 0.3661,
+      "step": 4622
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567532156593477e-05,
+      "loss": 0.344,
+      "step": 4623
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3564898425798667e-05,
+      "loss": 0.3665,
+      "step": 4624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3562264411692038e-05,
+      "loss": 0.3416,
+      "step": 4625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3559630114482918e-05,
+      "loss": 0.4056,
+      "step": 4626
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3556995534380666e-05,
+      "loss": 0.3957,
+      "step": 4627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3554360671594652e-05,
+      "loss": 0.3401,
+      "step": 4628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3551725526334286e-05,
+      "loss": 0.3351,
+      "step": 4629
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354909009880898e-05,
+      "loss": 0.3658,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546454389228182e-05,
+      "loss": 0.3837,
+      "step": 4631
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3543818397801358e-05,
+      "loss": 0.3698,
+      "step": 4632
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541182124738e-05,
+      "loss": 0.3675,
+      "step": 4633
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3538545570247616e-05,
+      "loss": 0.4027,
+      "step": 4634
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3535908734539744e-05,
+      "loss": 0.339,
+      "step": 4635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3533271617823935e-05,
+      "loss": 0.3965,
+      "step": 4636
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3530634220309773e-05,
+      "loss": 0.3327,
+      "step": 4637
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.352799654220685e-05,
+      "loss": 0.3428,
+      "step": 4638
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3525358583724803e-05,
+      "loss": 0.3363,
+      "step": 4639
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3522720345073266e-05,
+      "loss": 0.3353,
+      "step": 4640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520081826461911e-05,
+      "loss": 0.3682,
+      "step": 4641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517443028100426e-05,
+      "loss": 0.3464,
+      "step": 4642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3514803950198526e-05,
+      "loss": 0.3721,
+      "step": 4643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3512164592965944e-05,
+      "loss": 0.3717,
+      "step": 4644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3509524956612436e-05,
+      "loss": 0.3558,
+      "step": 4645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506885041347781e-05,
+      "loss": 0.3694,
+      "step": 4646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3504244847381781e-05,
+      "loss": 0.3396,
+      "step": 4647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3501604374924262e-05,
+      "loss": 0.3639,
+      "step": 4648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3498963624185065e-05,
+      "loss": 0.3426,
+      "step": 4649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3496322595374055e-05,
+      "loss": 0.3663,
+      "step": 4650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3493681288701126e-05,
+      "loss": 0.3751,
+      "step": 4651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349103970437619e-05,
+      "loss": 0.3829,
+      "step": 4652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.348839784260918e-05,
+      "loss": 0.3523,
+      "step": 4653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485755703610048e-05,
+      "loss": 0.3995,
+      "step": 4654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3483113287588775e-05,
+      "loss": 0.3268,
+      "step": 4655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348047059475536e-05,
+      "loss": 0.399,
+      "step": 4656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3477827625319826e-05,
+      "loss": 0.3448,
+      "step": 4657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3475184379492215e-05,
+      "loss": 0.378,
+      "step": 4658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3472540857482592e-05,
+      "loss": 0.4074,
+      "step": 4659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3469897059501049e-05,
+      "loss": 0.3977,
+      "step": 4660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467252985757689e-05,
+      "loss": 0.4172,
+      "step": 4661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3464608636462649e-05,
+      "loss": 0.3513,
+      "step": 4662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461964011826082e-05,
+      "loss": 0.3472,
+      "step": 4663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3459319112058156e-05,
+      "loss": 0.3771,
+      "step": 4664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3456673937369078e-05,
+      "loss": 0.3485,
+      "step": 4665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3454028487969062e-05,
+      "loss": 0.3879,
+      "step": 4666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3451382764068347e-05,
+      "loss": 0.3528,
+      "step": 4667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34487367658772e-05,
+      "loss": 0.3404,
+      "step": 4668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3446090493605903e-05,
+      "loss": 0.3814,
+      "step": 4669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3443443947464765e-05,
+      "loss": 0.3308,
+      "step": 4670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3440797127664112e-05,
+      "loss": 0.3941,
+      "step": 4671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343815003441429e-05,
+      "loss": 0.3501,
+      "step": 4672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435502667925673e-05,
+      "loss": 0.3829,
+      "step": 4673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3432855028408655e-05,
+      "loss": 0.3691,
+      "step": 4674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3430207116073654e-05,
+      "loss": 0.3658,
+      "step": 4675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3427558931131101e-05,
+      "loss": 0.332,
+      "step": 4676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3424910473791458e-05,
+      "loss": 0.36,
+      "step": 4677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3422261744265201e-05,
+      "loss": 0.3729,
+      "step": 4678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3419612742762835e-05,
+      "loss": 0.3811,
+      "step": 4679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3416963469494883e-05,
+      "loss": 0.3625,
+      "step": 4680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3414313924671887e-05,
+      "loss": 0.3769,
+      "step": 4681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411664108504416e-05,
+      "loss": 0.3786,
+      "step": 4682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3409014021203056e-05,
+      "loss": 0.3911,
+      "step": 4683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3406363662978417e-05,
+      "loss": 0.3724,
+      "step": 4684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340371303404113e-05,
+      "loss": 0.3807,
+      "step": 4685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3401062134601847e-05,
+      "loss": 0.3732,
+      "step": 4686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398410964871242e-05,
+      "loss": 0.3764,
+      "step": 4687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3395759525060012e-05,
+      "loss": 0.4012,
+      "step": 4688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393107815378873e-05,
+      "loss": 0.373,
+      "step": 4689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3390455836038565e-05,
+      "loss": 0.377,
+      "step": 4690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3387803587249839e-05,
+      "loss": 0.3554,
+      "step": 4691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3385151069223488e-05,
+      "loss": 0.3832,
+      "step": 4692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382498282170306e-05,
+      "loss": 0.3931,
+      "step": 4693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3379845226301125e-05,
+      "loss": 0.364,
+      "step": 4694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3377191901826782e-05,
+      "loss": 0.3468,
+      "step": 4695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374538308958147e-05,
+      "loss": 0.3642,
+      "step": 4696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3371884447906109e-05,
+      "loss": 0.3652,
+      "step": 4697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3369230318881582e-05,
+      "loss": 0.3769,
+      "step": 4698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3366575922095484e-05,
+      "loss": 0.4072,
+      "step": 4699
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3363921257758777e-05,
+      "loss": 0.3537,
+      "step": 4700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336126632608243e-05,
+      "loss": 0.3655,
+      "step": 4701
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3358611127277439e-05,
+      "loss": 0.3623,
+      "step": 4702
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3355955661554818e-05,
+      "loss": 0.4059,
+      "step": 4703
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3353299929125607e-05,
+      "loss": 0.4202,
+      "step": 4704
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3350643930200862e-05,
+      "loss": 0.3659,
+      "step": 4705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334798766499166e-05,
+      "loss": 0.4186,
+      "step": 4706
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3345331133709108e-05,
+      "loss": 0.3831,
+      "step": 4707
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3342674336564316e-05,
+      "loss": 0.3516,
+      "step": 4708
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3340017273768437e-05,
+      "loss": 0.3327,
+      "step": 4709
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3337359945532633e-05,
+      "loss": 0.385,
+      "step": 4710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3334702352068086e-05,
+      "loss": 0.3392,
+      "step": 4711
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3332044493586004e-05,
+      "loss": 0.3637,
+      "step": 4712
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329386370297615e-05,
+      "loss": 0.3895,
+      "step": 4713
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3326727982414158e-05,
+      "loss": 0.3702,
+      "step": 4714
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324069330146916e-05,
+      "loss": 0.403,
+      "step": 4715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3321410413707171e-05,
+      "loss": 0.3665,
+      "step": 4716
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3318751233306234e-05,
+      "loss": 0.3843,
+      "step": 4717
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3316091789155442e-05,
+      "loss": 0.3787,
+      "step": 4718
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3313432081466138e-05,
+      "loss": 0.3949,
+      "step": 4719
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3310772110449706e-05,
+      "loss": 0.4076,
+      "step": 4720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3308111876317538e-05,
+      "loss": 0.3605,
+      "step": 4721
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3305451379281046e-05,
+      "loss": 0.3965,
+      "step": 4722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3302790619551673e-05,
+      "loss": 0.3462,
+      "step": 4723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300129597340871e-05,
+      "loss": 0.4034,
+      "step": 4724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3297468312860122e-05,
+      "loss": 0.4004,
+      "step": 4725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3294806766320923e-05,
+      "loss": 0.3597,
+      "step": 4726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3292144957934795e-05,
+      "loss": 0.3701,
+      "step": 4727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328948288791328e-05,
+      "loss": 0.4113,
+      "step": 4728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3286820556467937e-05,
+      "loss": 0.3417,
+      "step": 4729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3284157963810352e-05,
+      "loss": 0.3647,
+      "step": 4730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281495110152121e-05,
+      "loss": 0.3845,
+      "step": 4731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3278831995704877e-05,
+      "loss": 0.3625,
+      "step": 4732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276168620680259e-05,
+      "loss": 0.3597,
+      "step": 4733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3273504985289938e-05,
+      "loss": 0.3549,
+      "step": 4734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270841089745594e-05,
+      "loss": 0.3511,
+      "step": 4735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3268176934258937e-05,
+      "loss": 0.3395,
+      "step": 4736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326551251904169e-05,
+      "loss": 0.381,
+      "step": 4737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326284784430561e-05,
+      "loss": 0.3689,
+      "step": 4738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3260182910262455e-05,
+      "loss": 0.3668,
+      "step": 4739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3257517717124022e-05,
+      "loss": 0.3917,
+      "step": 4740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3254852265102118e-05,
+      "loss": 0.4234,
+      "step": 4741
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3252186554408575e-05,
+      "loss": 0.3745,
+      "step": 4742
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3249520585255242e-05,
+      "loss": 0.3951,
+      "step": 4743
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246854357853991e-05,
+      "loss": 0.3566,
+      "step": 4744
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3244187872416716e-05,
+      "loss": 0.3593,
+      "step": 4745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3241521129155326e-05,
+      "loss": 0.3634,
+      "step": 4746
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3238854128281758e-05,
+      "loss": 0.4118,
+      "step": 4747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3236186870007963e-05,
+      "loss": 0.3688,
+      "step": 4748
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3233519354545915e-05,
+      "loss": 0.3584,
+      "step": 4749
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323085158210761e-05,
+      "loss": 0.3535,
+      "step": 4750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228183552905061e-05,
+      "loss": 0.3495,
+      "step": 4751
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225515267150305e-05,
+      "loss": 0.3821,
+      "step": 4752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3222846725055397e-05,
+      "loss": 0.3394,
+      "step": 4753
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220177926832416e-05,
+      "loss": 0.3312,
+      "step": 4754
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321750887269345e-05,
+      "loss": 0.3428,
+      "step": 4755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3214839562850625e-05,
+      "loss": 0.3706,
+      "step": 4756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3212169997516074e-05,
+      "loss": 0.3834,
+      "step": 4757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3209500176901953e-05,
+      "loss": 0.339,
+      "step": 4758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320683010122044e-05,
+      "loss": 0.338,
+      "step": 4759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3204159770683736e-05,
+      "loss": 0.3287,
+      "step": 4760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3201489185504056e-05,
+      "loss": 0.3709,
+      "step": 4761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319881834589364e-05,
+      "loss": 0.364,
+      "step": 4762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3196147252064748e-05,
+      "loss": 0.3393,
+      "step": 4763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3193475904229655e-05,
+      "loss": 0.3503,
+      "step": 4764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3190804302600664e-05,
+      "loss": 0.3591,
+      "step": 4765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3188132447390091e-05,
+      "loss": 0.3799,
+      "step": 4766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3185460338810279e-05,
+      "loss": 0.3572,
+      "step": 4767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3182787977073582e-05,
+      "loss": 0.4153,
+      "step": 4768
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3180115362392383e-05,
+      "loss": 0.368,
+      "step": 4769
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3177442494979085e-05,
+      "loss": 0.3672,
+      "step": 4770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31747693750461e-05,
+      "loss": 0.3659,
+      "step": 4771
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3172096002805872e-05,
+      "loss": 0.3149,
+      "step": 4772
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3169422378470862e-05,
+      "loss": 0.3788,
+      "step": 4773
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316674850225355e-05,
+      "loss": 0.3499,
+      "step": 4774
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3164074374366435e-05,
+      "loss": 0.3393,
+      "step": 4775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3161399995022036e-05,
+      "loss": 0.3735,
+      "step": 4776
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3158725364432893e-05,
+      "loss": 0.3665,
+      "step": 4777
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3156050482811566e-05,
+      "loss": 0.327,
+      "step": 4778
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3153375350370639e-05,
+      "loss": 0.3705,
+      "step": 4779
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3150699967322705e-05,
+      "loss": 0.3478,
+      "step": 4780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3148024333880387e-05,
+      "loss": 0.359,
+      "step": 4781
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3145348450256325e-05,
+      "loss": 0.3999,
+      "step": 4782
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3142672316663178e-05,
+      "loss": 0.3747,
+      "step": 4783
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3139995933313627e-05,
+      "loss": 0.3866,
+      "step": 4784
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3137319300420368e-05,
+      "loss": 0.3525,
+      "step": 4785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3134642418196122e-05,
+      "loss": 0.3666,
+      "step": 4786
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131965286853627e-05,
+      "loss": 0.3308,
+      "step": 4787
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3129287906605641e-05,
+      "loss": 0.339,
+      "step": 4788
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3126610277664949e-05,
+      "loss": 0.3533,
+      "step": 4789
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123932400244337e-05,
+      "loss": 0.3758,
+      "step": 4790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121254274556631e-05,
+      "loss": 0.3304,
+      "step": 4791
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3118575900814669e-05,
+      "loss": 0.392,
+      "step": 4792
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3115897279231309e-05,
+      "loss": 0.3704,
+      "step": 4793
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3113218410019421e-05,
+      "loss": 0.3698,
+      "step": 4794
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3110539293391909e-05,
+      "loss": 0.3827,
+      "step": 4795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3107859929561686e-05,
+      "loss": 0.348,
+      "step": 4796
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105180318741691e-05,
+      "loss": 0.3637,
+      "step": 4797
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102500461144876e-05,
+      "loss": 0.3362,
+      "step": 4798
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3099820356984218e-05,
+      "loss": 0.3641,
+      "step": 4799
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3097140006472712e-05,
+      "loss": 0.3651,
+      "step": 4800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3094459409823372e-05,
+      "loss": 0.3444,
+      "step": 4801
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3091778567249234e-05,
+      "loss": 0.3773,
+      "step": 4802
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3089097478963348e-05,
+      "loss": 0.3926,
+      "step": 4803
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308641614517879e-05,
+      "loss": 0.3849,
+      "step": 4804
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3083734566108652e-05,
+      "loss": 0.3557,
+      "step": 4805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3081052741966047e-05,
+      "loss": 0.3661,
+      "step": 4806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078370672964105e-05,
+      "loss": 0.3329,
+      "step": 4807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3075688359315977e-05,
+      "loss": 0.3658,
+      "step": 4808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3073005801234833e-05,
+      "loss": 0.3737,
+      "step": 4809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070322998933864e-05,
+      "loss": 0.3451,
+      "step": 4810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306763995262628e-05,
+      "loss": 0.3377,
+      "step": 4811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306495666252531e-05,
+      "loss": 0.3685,
+      "step": 4812
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.30622731288442e-05,
+      "loss": 0.3573,
+      "step": 4813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059589351796218e-05,
+      "loss": 0.3456,
+      "step": 4814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3056905331594653e-05,
+      "loss": 0.3376,
+      "step": 4815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305422106845281e-05,
+      "loss": 0.3486,
+      "step": 4816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3051536562584014e-05,
+      "loss": 0.353,
+      "step": 4817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048851814201607e-05,
+      "loss": 0.3513,
+      "step": 4818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3046166823518956e-05,
+      "loss": 0.3703,
+      "step": 4819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3043481590749445e-05,
+      "loss": 0.3882,
+      "step": 4820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040796116106473e-05,
+      "loss": 0.3509,
+      "step": 4821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3038110399803468e-05,
+      "loss": 0.376,
+      "step": 4822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3035424442053864e-05,
+      "loss": 0.3993,
+      "step": 4823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3032738243071124e-05,
+      "loss": 0.3809,
+      "step": 4824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3030051803068729e-05,
+      "loss": 0.3932,
+      "step": 4825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3027365122260174e-05,
+      "loss": 0.3305,
+      "step": 4826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302467820085898e-05,
+      "loss": 0.3691,
+      "step": 4827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302199103907868e-05,
+      "loss": 0.385,
+      "step": 4828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3019303637132835e-05,
+      "loss": 0.3394,
+      "step": 4829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016615995235013e-05,
+      "loss": 0.3993,
+      "step": 4830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3013928113598815e-05,
+      "loss": 0.3867,
+      "step": 4831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011239992437847e-05,
+      "loss": 0.3748,
+      "step": 4832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300855163196575e-05,
+      "loss": 0.3584,
+      "step": 4833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3005863032396167e-05,
+      "loss": 0.3264,
+      "step": 4834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3003174193942769e-05,
+      "loss": 0.4251,
+      "step": 4835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3000485116819246e-05,
+      "loss": 0.3645,
+      "step": 4836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2997795801239308e-05,
+      "loss": 0.3447,
+      "step": 4837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2995106247416684e-05,
+      "loss": 0.423,
+      "step": 4838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992416455565114e-05,
+      "loss": 0.3238,
+      "step": 4839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2989726425898366e-05,
+      "loss": 0.3842,
+      "step": 4840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2987036158630223e-05,
+      "loss": 0.3381,
+      "step": 4841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2984345653974488e-05,
+      "loss": 0.4481,
+      "step": 4842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298165491214498e-05,
+      "loss": 0.3478,
+      "step": 4843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2978963933355545e-05,
+      "loss": 0.345,
+      "step": 4844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2976272717820036e-05,
+      "loss": 0.3416,
+      "step": 4845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2973581265752335e-05,
+      "loss": 0.3295,
+      "step": 4846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2970889577366339e-05,
+      "loss": 0.3773,
+      "step": 4847
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2968197652875958e-05,
+      "loss": 0.3583,
+      "step": 4848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2965505492495132e-05,
+      "loss": 0.3707,
+      "step": 4849
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962813096437813e-05,
+      "loss": 0.3515,
+      "step": 4850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2960120464917971e-05,
+      "loss": 0.3668,
+      "step": 4851
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29574275981496e-05,
+      "loss": 0.3873,
+      "step": 4852
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954734496346704e-05,
+      "loss": 0.3373,
+      "step": 4853
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295204115972331e-05,
+      "loss": 0.3567,
+      "step": 4854
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2949347588493472e-05,
+      "loss": 0.3635,
+      "step": 4855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2946653782871254e-05,
+      "loss": 0.3469,
+      "step": 4856
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2943959743070735e-05,
+      "loss": 0.4003,
+      "step": 4857
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2941265469306016e-05,
+      "loss": 0.347,
+      "step": 4858
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2938570961791225e-05,
+      "loss": 0.3709,
+      "step": 4859
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2935876220740498e-05,
+      "loss": 0.386,
+      "step": 4860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933181246367991e-05,
+      "loss": 0.3586,
+      "step": 4861
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2930486038887883e-05,
+      "loss": 0.3388,
+      "step": 4862
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927790598514371e-05,
+      "loss": 0.3688,
+      "step": 4863
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2925094925461665e-05,
+      "loss": 0.3373,
+      "step": 4864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2922399019944001e-05,
+      "loss": 0.3832,
+      "step": 4865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2919702882175624e-05,
+      "loss": 0.3702,
+      "step": 4866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917006512370809e-05,
+      "loss": 0.3573,
+      "step": 4867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291430991074384e-05,
+      "loss": 0.3398,
+      "step": 4868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2911613077509026e-05,
+      "loss": 0.3633,
+      "step": 4869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2908916012880693e-05,
+      "loss": 0.364,
+      "step": 4870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906218717073173e-05,
+      "loss": 0.3656,
+      "step": 4871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290352119030084e-05,
+      "loss": 0.3769,
+      "step": 4872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900823432778064e-05,
+      "loss": 0.3712,
+      "step": 4873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2898125444719251e-05,
+      "loss": 0.3773,
+      "step": 4874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2895427226338812e-05,
+      "loss": 0.3846,
+      "step": 4875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2892728777851185e-05,
+      "loss": 0.3921,
+      "step": 4876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2890030099470813e-05,
+      "loss": 0.3625,
+      "step": 4877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2887331191412183e-05,
+      "loss": 0.39,
+      "step": 4878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2884632053889772e-05,
+      "loss": 0.3783,
+      "step": 4879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288193268711809e-05,
+      "loss": 0.3557,
+      "step": 4880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879233091311667e-05,
+      "loss": 0.3898,
+      "step": 4881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287653326668504e-05,
+      "loss": 0.3677,
+      "step": 4882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287383321345278e-05,
+      "loss": 0.3517,
+      "step": 4883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.287113293182946e-05,
+      "loss": 0.3816,
+      "step": 4884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868432422029681e-05,
+      "loss": 0.3418,
+      "step": 4885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2865731684268061e-05,
+      "loss": 0.351,
+      "step": 4886
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286303071875923e-05,
+      "loss": 0.3507,
+      "step": 4887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286032952571785e-05,
+      "loss": 0.3579,
+      "step": 4888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285762810535858e-05,
+      "loss": 0.367,
+      "step": 4889
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2854926457896117e-05,
+      "loss": 0.3552,
+      "step": 4890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2852224583545166e-05,
+      "loss": 0.3513,
+      "step": 4891
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849522482520453e-05,
+      "loss": 0.365,
+      "step": 4892
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284682015503672e-05,
+      "loss": 0.3814,
+      "step": 4893
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2844117601308726e-05,
+      "loss": 0.3499,
+      "step": 4894
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841414821551252e-05,
+      "loss": 0.3591,
+      "step": 4895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2838711815979096e-05,
+      "loss": 0.3433,
+      "step": 4896
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2836008584807074e-05,
+      "loss": 0.3344,
+      "step": 4897
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2833305128250016e-05,
+      "loss": 0.3482,
+      "step": 4898
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2830601446522773e-05,
+      "loss": 0.3389,
+      "step": 4899
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2827897539840213e-05,
+      "loss": 0.3451,
+      "step": 4900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825193408417225e-05,
+      "loss": 0.3612,
+      "step": 4901
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822489052468712e-05,
+      "loss": 0.4112,
+      "step": 4902
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2819784472209598e-05,
+      "loss": 0.3804,
+      "step": 4903
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817079667854819e-05,
+      "loss": 0.3644,
+      "step": 4904
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2814374639619338e-05,
+      "loss": 0.3417,
+      "step": 4905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2811669387718127e-05,
+      "loss": 0.3758,
+      "step": 4906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2808963912366182e-05,
+      "loss": 0.3428,
+      "step": 4907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280625821377851e-05,
+      "loss": 0.3748,
+      "step": 4908
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2803552292170145e-05,
+      "loss": 0.3542,
+      "step": 4909
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2800846147756131e-05,
+      "loss": 0.3795,
+      "step": 4910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2798139780751535e-05,
+      "loss": 0.3357,
+      "step": 4911
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795433191371434e-05,
+      "loss": 0.3519,
+      "step": 4912
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.279272637983093e-05,
+      "loss": 0.3606,
+      "step": 4913
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2790019346345146e-05,
+      "loss": 0.3494,
+      "step": 4914
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2787312091129208e-05,
+      "loss": 0.4172,
+      "step": 4915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2784604614398274e-05,
+      "loss": 0.3753,
+      "step": 4916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2781896916367513e-05,
+      "loss": 0.3417,
+      "step": 4917
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2779188997252113e-05,
+      "loss": 0.348,
+      "step": 4918
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277648085726728e-05,
+      "loss": 0.3335,
+      "step": 4919
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2773772496628236e-05,
+      "loss": 0.4035,
+      "step": 4920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277106391555022e-05,
+      "loss": 0.3544,
+      "step": 4921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2768355114248493e-05,
+      "loss": 0.3779,
+      "step": 4922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2765646092938333e-05,
+      "loss": 0.3495,
+      "step": 4923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2762936851835026e-05,
+      "loss": 0.3938,
+      "step": 4924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2760227391153888e-05,
+      "loss": 0.3984,
+      "step": 4925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2757517711110244e-05,
+      "loss": 0.3523,
+      "step": 4926
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2754807811919438e-05,
+      "loss": 0.36,
+      "step": 4927
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2752097693796841e-05,
+      "loss": 0.342,
+      "step": 4928
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2749387356957824e-05,
+      "loss": 0.4001,
+      "step": 4929
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2746676801617784e-05,
+      "loss": 0.3447,
+      "step": 4930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743966027992142e-05,
+      "loss": 0.3647,
+      "step": 4931
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2741255036296334e-05,
+      "loss": 0.3436,
+      "step": 4932
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738543826745798e-05,
+      "loss": 0.3544,
+      "step": 4933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2735832399556007e-05,
+      "loss": 0.3642,
+      "step": 4934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2733120754942446e-05,
+      "loss": 0.3882,
+      "step": 4935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730408893120614e-05,
+      "loss": 0.36,
+      "step": 4936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727696814306034e-05,
+      "loss": 0.3607,
+      "step": 4937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2724984518714236e-05,
+      "loss": 0.3837,
+      "step": 4938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2722272006560779e-05,
+      "loss": 0.3816,
+      "step": 4939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2719559278061229e-05,
+      "loss": 0.3601,
+      "step": 4940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2716846333431176e-05,
+      "loss": 0.3644,
+      "step": 4941
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714133172886222e-05,
+      "loss": 0.3791,
+      "step": 4942
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711419796641995e-05,
+      "loss": 0.3406,
+      "step": 4943
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270870620491413e-05,
+      "loss": 0.3754,
+      "step": 4944
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270599239791828e-05,
+      "loss": 0.3655,
+      "step": 4945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2703278375870127e-05,
+      "loss": 0.3913,
+      "step": 4946
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2700564138985354e-05,
+      "loss": 0.3686,
+      "step": 4947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697849687479672e-05,
+      "loss": 0.3357,
+      "step": 4948
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2695135021568803e-05,
+      "loss": 0.3331,
+      "step": 4949
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2692420141468489e-05,
+      "loss": 0.4399,
+      "step": 4950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689705047394494e-05,
+      "loss": 0.3994,
+      "step": 4951
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2686989739562587e-05,
+      "loss": 0.3505,
+      "step": 4952
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268427421818856e-05,
+      "loss": 0.3646,
+      "step": 4953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268155848348823e-05,
+      "loss": 0.4008,
+      "step": 4954
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267884253567742e-05,
+      "loss": 0.3647,
+      "step": 4955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267612637497197e-05,
+      "loss": 0.3636,
+      "step": 4956
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673410001587745e-05,
+      "loss": 0.3496,
+      "step": 4957
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2670693415740621e-05,
+      "loss": 0.3615,
+      "step": 4958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2667976617646488e-05,
+      "loss": 0.4005,
+      "step": 4959
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665259607521263e-05,
+      "loss": 0.3521,
+      "step": 4960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2662542385580871e-05,
+      "loss": 0.3579,
+      "step": 4961
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2659824952041257e-05,
+      "loss": 0.3426,
+      "step": 4962
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2657107307118385e-05,
+      "loss": 0.3487,
+      "step": 4963
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2654389451028228e-05,
+      "loss": 0.3294,
+      "step": 4964
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2651671383986788e-05,
+      "loss": 0.3578,
+      "step": 4965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2648953106210071e-05,
+      "loss": 0.3976,
+      "step": 4966
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2646234617914108e-05,
+      "loss": 0.3449,
+      "step": 4967
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2643515919314945e-05,
+      "loss": 0.3725,
+      "step": 4968
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2640797010628647e-05,
+      "loss": 0.365,
+      "step": 4969
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2638077892071284e-05,
+      "loss": 0.3872,
+      "step": 4970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263535856385896e-05,
+      "loss": 0.3486,
+      "step": 4971
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263263902620778e-05,
+      "loss": 0.3528,
+      "step": 4972
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2629919279333884e-05,
+      "loss": 0.3533,
+      "step": 4973
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627199323453406e-05,
+      "loss": 0.3191,
+      "step": 4974
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262447915878251e-05,
+      "loss": 0.3491,
+      "step": 4975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2621758785537376e-05,
+      "loss": 0.3436,
+      "step": 4976
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2619038203934205e-05,
+      "loss": 0.389,
+      "step": 4977
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26163174141892e-05,
+      "loss": 0.3483,
+      "step": 4978
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2613596416518594e-05,
+      "loss": 0.3397,
+      "step": 4979
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2610875211138632e-05,
+      "loss": 0.3659,
+      "step": 4980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2608153798265573e-05,
+      "loss": 0.3618,
+      "step": 4981
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2605432178115695e-05,
+      "loss": 0.3429,
+      "step": 4982
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2602710350905295e-05,
+      "loss": 0.3725,
+      "step": 4983
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2599988316850682e-05,
+      "loss": 0.3769,
+      "step": 4984
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2597266076168182e-05,
+      "loss": 0.3678,
+      "step": 4985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2594543629074138e-05,
+      "loss": 0.3516,
+      "step": 4986
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591820975784916e-05,
+      "loss": 0.3298,
+      "step": 4987
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2589098116516887e-05,
+      "loss": 0.3639,
+      "step": 4988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258637505148644e-05,
+      "loss": 0.3688,
+      "step": 4989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2583651780909995e-05,
+      "loss": 0.3489,
+      "step": 4990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2580928305003972e-05,
+      "loss": 0.3793,
+      "step": 4991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2578204623984812e-05,
+      "loss": 0.3677,
+      "step": 4992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2575480738068971e-05,
+      "loss": 0.3536,
+      "step": 4993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257275664747292e-05,
+      "loss": 0.3905,
+      "step": 4994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2570032352413164e-05,
+      "loss": 0.3818,
+      "step": 4995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2567307853106199e-05,
+      "loss": 0.351,
+      "step": 4996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.256458314976855e-05,
+      "loss": 0.3631,
+      "step": 4997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2561858242616756e-05,
+      "loss": 0.3384,
+      "step": 4998
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2559133131867372e-05,
+      "loss": 0.3572,
+      "step": 4999
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2556407817736973e-05,
+      "loss": 0.3962,
+      "step": 5000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553682300442143e-05,
+      "loss": 0.3925,
+      "step": 5001
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2550956580199487e-05,
+      "loss": 0.3538,
+      "step": 5002
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2548230657225629e-05,
+      "loss": 0.3353,
+      "step": 5003
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.25455045317372e-05,
+      "loss": 0.4043,
+      "step": 5004
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542778203950854e-05,
+      "loss": 0.3509,
+      "step": 5005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254005167408326e-05,
+      "loss": 0.3462,
+      "step": 5006
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537324942351101e-05,
+      "loss": 0.3334,
+      "step": 5007
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2534598008971082e-05,
+      "loss": 0.3431,
+      "step": 5008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2531870874159915e-05,
+      "loss": 0.3582,
+      "step": 5009
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529143538134334e-05,
+      "loss": 0.3917,
+      "step": 5010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2526416001111087e-05,
+      "loss": 0.3553,
+      "step": 5011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2523688263306936e-05,
+      "loss": 0.3549,
+      "step": 5012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2520960324938667e-05,
+      "loss": 0.3362,
+      "step": 5013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518232186223077e-05,
+      "loss": 0.3576,
+      "step": 5014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515503847376974e-05,
+      "loss": 0.3537,
+      "step": 5015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2512775308617183e-05,
+      "loss": 0.384,
+      "step": 5016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2510046570160552e-05,
+      "loss": 0.3622,
+      "step": 5017
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2507317632223948e-05,
+      "loss": 0.3595,
+      "step": 5018
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2504588495024236e-05,
+      "loss": 0.3463,
+      "step": 5019
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2501859158778314e-05,
+      "loss": 0.3575,
+      "step": 5020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499129623703086e-05,
+      "loss": 0.3602,
+      "step": 5021
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2496399890015477e-05,
+      "loss": 0.3251,
+      "step": 5022
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2493669957932425e-05,
+      "loss": 0.3931,
+      "step": 5023
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2490939827670886e-05,
+      "loss": 0.3459,
+      "step": 5024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2488209499447832e-05,
+      "loss": 0.3989,
+      "step": 5025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2485478973480247e-05,
+      "loss": 0.4041,
+      "step": 5026
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2482748249985137e-05,
+      "loss": 0.4138,
+      "step": 5027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480017329179513e-05,
+      "loss": 0.3415,
+      "step": 5028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247728621128041e-05,
+      "loss": 0.3862,
+      "step": 5029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2474554896504882e-05,
+      "loss": 0.3566,
+      "step": 5030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247182338506999e-05,
+      "loss": 0.365,
+      "step": 5031
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2469091677192817e-05,
+      "loss": 0.3704,
+      "step": 5032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2466359773090455e-05,
+      "loss": 0.3677,
+      "step": 5033
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2463627672980017e-05,
+      "loss": 0.3903,
+      "step": 5034
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2460895377078632e-05,
+      "loss": 0.3776,
+      "step": 5035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458162885603444e-05,
+      "loss": 0.3754,
+      "step": 5036
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2455430198771607e-05,
+      "loss": 0.3073,
+      "step": 5037
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2452697316800296e-05,
+      "loss": 0.3506,
+      "step": 5038
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2449964239906703e-05,
+      "loss": 0.3802,
+      "step": 5039
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2447230968308028e-05,
+      "loss": 0.3569,
+      "step": 5040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2444497502221495e-05,
+      "loss": 0.359,
+      "step": 5041
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2441763841864337e-05,
+      "loss": 0.3484,
+      "step": 5042
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.243902998745381e-05,
+      "loss": 0.3821,
+      "step": 5043
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2436295939207177e-05,
+      "loss": 0.3196,
+      "step": 5044
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2433561697341721e-05,
+      "loss": 0.3832,
+      "step": 5045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2430827262074737e-05,
+      "loss": 0.3372,
+      "step": 5046
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2428092633623539e-05,
+      "loss": 0.3772,
+      "step": 5047
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2425357812205457e-05,
+      "loss": 0.3587,
+      "step": 5048
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2422622798037833e-05,
+      "loss": 0.36,
+      "step": 5049
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2419887591338026e-05,
+      "loss": 0.3714,
+      "step": 5050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417152192323409e-05,
+      "loss": 0.3777,
+      "step": 5051
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414416601211372e-05,
+      "loss": 0.3994,
+      "step": 5052
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2411680818219321e-05,
+      "loss": 0.3687,
+      "step": 5053
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2408944843564674e-05,
+      "loss": 0.3586,
+      "step": 5054
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2406208677464868e-05,
+      "loss": 0.3621,
+      "step": 5055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240347232013735e-05,
+      "loss": 0.3631,
+      "step": 5056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2400735771799588e-05,
+      "loss": 0.4061,
+      "step": 5057
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2397999032669064e-05,
+      "loss": 0.3437,
+      "step": 5058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395262102963268e-05,
+      "loss": 0.3475,
+      "step": 5059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392524982899718e-05,
+      "loss": 0.3543,
+      "step": 5060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2389787672695937e-05,
+      "loss": 0.3627,
+      "step": 5061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2387050172569463e-05,
+      "loss": 0.3857,
+      "step": 5062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2384312482737858e-05,
+      "loss": 0.3818,
+      "step": 5063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2381574603418691e-05,
+      "loss": 0.3502,
+      "step": 5064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237883653482955e-05,
+      "loss": 0.3962,
+      "step": 5065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2376098277188033e-05,
+      "loss": 0.3542,
+      "step": 5066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2373359830711758e-05,
+      "loss": 0.3681,
+      "step": 5067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237062119561836e-05,
+      "loss": 0.3335,
+      "step": 5068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367882372125477e-05,
+      "loss": 0.3422,
+      "step": 5069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236514336045078e-05,
+      "loss": 0.3631,
+      "step": 5070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236240416081194e-05,
+      "loss": 0.3793,
+      "step": 5071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2359664773426651e-05,
+      "loss": 0.3678,
+      "step": 5072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356925198512617e-05,
+      "loss": 0.3816,
+      "step": 5073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235418543628756e-05,
+      "loss": 0.3532,
+      "step": 5074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2351445486969216e-05,
+      "loss": 0.3549,
+      "step": 5075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2348705350775339e-05,
+      "loss": 0.3521,
+      "step": 5076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234596502792369e-05,
+      "loss": 0.3476,
+      "step": 5077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2343224518632051e-05,
+      "loss": 0.361,
+      "step": 5078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2340483823118219e-05,
+      "loss": 0.3816,
+      "step": 5079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2337742941600002e-05,
+      "loss": 0.3477,
+      "step": 5080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335001874295227e-05,
+      "loss": 0.3582,
+      "step": 5081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2332260621421735e-05,
+      "loss": 0.3914,
+      "step": 5082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329519183197378e-05,
+      "loss": 0.3778,
+      "step": 5083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2326777559840024e-05,
+      "loss": 0.3538,
+      "step": 5084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2324035751567562e-05,
+      "loss": 0.3468,
+      "step": 5085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2321293758597887e-05,
+      "loss": 0.3373,
+      "step": 5086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2318551581148912e-05,
+      "loss": 0.3515,
+      "step": 5087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2315809219438565e-05,
+      "loss": 0.3537,
+      "step": 5088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2313066673684793e-05,
+      "loss": 0.3839,
+      "step": 5089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2310323944105551e-05,
+      "loss": 0.3211,
+      "step": 5090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307581030918807e-05,
+      "loss": 0.3715,
+      "step": 5091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2304837934342553e-05,
+      "loss": 0.3524,
+      "step": 5092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2302094654594782e-05,
+      "loss": 0.3357,
+      "step": 5093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229935119189352e-05,
+      "loss": 0.3432,
+      "step": 5094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2296607546456793e-05,
+      "loss": 0.3466,
+      "step": 5095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2293863718502645e-05,
+      "loss": 0.3669,
+      "step": 5096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2291119708249135e-05,
+      "loss": 0.397,
+      "step": 5097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2288375515914335e-05,
+      "loss": 0.3644,
+      "step": 5098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2285631141716337e-05,
+      "loss": 0.3606,
+      "step": 5099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2282886585873239e-05,
+      "loss": 0.3723,
+      "step": 5100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2280141848603165e-05,
+      "loss": 0.3538,
+      "step": 5101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227739693012424e-05,
+      "loss": 0.3784,
+      "step": 5102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274651830654612e-05,
+      "loss": 0.3575,
+      "step": 5103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227190655041244e-05,
+      "loss": 0.3568,
+      "step": 5104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2269161089615902e-05,
+      "loss": 0.3473,
+      "step": 5105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266415448483184e-05,
+      "loss": 0.3478,
+      "step": 5106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2263669627232493e-05,
+      "loss": 0.3518,
+      "step": 5107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2260923626082042e-05,
+      "loss": 0.3735,
+      "step": 5108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2258177445250066e-05,
+      "loss": 0.3808,
+      "step": 5109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2255431084954805e-05,
+      "loss": 0.3582,
+      "step": 5110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2252684545414528e-05,
+      "loss": 0.3575,
+      "step": 5111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2249937826847507e-05,
+      "loss": 0.3596,
+      "step": 5112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2247190929472031e-05,
+      "loss": 0.335,
+      "step": 5113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.22444438535064e-05,
+      "loss": 0.3482,
+      "step": 5114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2241696599168936e-05,
+      "loss": 0.3311,
+      "step": 5115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238949166677962e-05,
+      "loss": 0.3862,
+      "step": 5116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2236201556251835e-05,
+      "loss": 0.3903,
+      "step": 5117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233453768108908e-05,
+      "loss": 0.3701,
+      "step": 5118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2230705802467557e-05,
+      "loss": 0.343,
+      "step": 5119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2227957659546168e-05,
+      "loss": 0.3897,
+      "step": 5120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.3585,
+      "step": 5121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222460842736905e-05,
+      "loss": 0.3638,
+      "step": 5122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2219712169285876e-05,
+      "loss": 0.3866,
+      "step": 5123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2216963319428503e-05,
+      "loss": 0.3447,
+      "step": 5124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2214214293383243e-05,
+      "loss": 0.354,
+      "step": 5125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2211465091368572e-05,
+      "loss": 0.3398,
+      "step": 5126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2208715713602976e-05,
+      "loss": 0.3657,
+      "step": 5127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2205966160304951e-05,
+      "loss": 0.3413,
+      "step": 5128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2203216431693014e-05,
+      "loss": 0.3452,
+      "step": 5129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200466527985692e-05,
+      "loss": 0.3611,
+      "step": 5130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2197716449401532e-05,
+      "loss": 0.3574,
+      "step": 5131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194966196159081e-05,
+      "loss": 0.3178,
+      "step": 5132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2192215768476915e-05,
+      "loss": 0.4029,
+      "step": 5133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2189465166573615e-05,
+      "loss": 0.3354,
+      "step": 5134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2186714390667782e-05,
+      "loss": 0.3387,
+      "step": 5135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2183963440978027e-05,
+      "loss": 0.3799,
+      "step": 5136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218121231772297e-05,
+      "loss": 0.3456,
+      "step": 5137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2178461021121253e-05,
+      "loss": 0.3392,
+      "step": 5138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217570955139153e-05,
+      "loss": 0.341,
+      "step": 5139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2172957908752468e-05,
+      "loss": 0.3532,
+      "step": 5140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170206093422746e-05,
+      "loss": 0.3529,
+      "step": 5141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2167454105621055e-05,
+      "loss": 0.3723,
+      "step": 5142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2164701945566106e-05,
+      "loss": 0.3588,
+      "step": 5143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2161949613476623e-05,
+      "loss": 0.3088,
+      "step": 5144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2159197109571337e-05,
+      "loss": 0.352,
+      "step": 5145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2156444434068994e-05,
+      "loss": 0.4122,
+      "step": 5146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2153691587188363e-05,
+      "loss": 0.3403,
+      "step": 5147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2150938569148219e-05,
+      "loss": 0.4052,
+      "step": 5148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2148185380167347e-05,
+      "loss": 0.4183,
+      "step": 5149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2145432020464553e-05,
+      "loss": 0.3606,
+      "step": 5150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2142678490258653e-05,
+      "loss": 0.3749,
+      "step": 5151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2139924789768478e-05,
+      "loss": 0.3347,
+      "step": 5152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2137170919212873e-05,
+      "loss": 0.3678,
+      "step": 5153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2134416878810693e-05,
+      "loss": 0.3689,
+      "step": 5154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2131662668780808e-05,
+      "loss": 0.3639,
+      "step": 5155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2128908289342105e-05,
+      "loss": 0.3536,
+      "step": 5156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2126153740713477e-05,
+      "loss": 0.3767,
+      "step": 5157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2123399023113844e-05,
+      "loss": 0.3754,
+      "step": 5158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120644136762123e-05,
+      "loss": 0.3726,
+      "step": 5159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2117889081877257e-05,
+      "loss": 0.3739,
+      "step": 5160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115133858678192e-05,
+      "loss": 0.3693,
+      "step": 5161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2112378467383896e-05,
+      "loss": 0.366,
+      "step": 5162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2109622908213348e-05,
+      "loss": 0.3573,
+      "step": 5163
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2106867181385538e-05,
+      "loss": 0.3593,
+      "step": 5164
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104111287119471e-05,
+      "loss": 0.3578,
+      "step": 5165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2101355225634164e-05,
+      "loss": 0.381,
+      "step": 5166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2098598997148654e-05,
+      "loss": 0.3404,
+      "step": 5167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2095842601881978e-05,
+      "loss": 0.3236,
+      "step": 5168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2093086040053198e-05,
+      "loss": 0.3541,
+      "step": 5169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2090329311881383e-05,
+      "loss": 0.3344,
+      "step": 5170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2087572417585623e-05,
+      "loss": 0.3483,
+      "step": 5171
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2084815357385009e-05,
+      "loss": 0.3763,
+      "step": 5172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2082058131498656e-05,
+      "loss": 0.3478,
+      "step": 5173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2079300740145681e-05,
+      "loss": 0.3444,
+      "step": 5174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207654318354523e-05,
+      "loss": 0.3537,
+      "step": 5175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073785461916453e-05,
+      "loss": 0.368,
+      "step": 5176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2071027575478508e-05,
+      "loss": 0.3326,
+      "step": 5177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2068269524450573e-05,
+      "loss": 0.3535,
+      "step": 5178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2065511309051841e-05,
+      "loss": 0.3777,
+      "step": 5179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206275292950151e-05,
+      "loss": 0.3455,
+      "step": 5180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2059994386018794e-05,
+      "loss": 0.3274,
+      "step": 5181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205723567882293e-05,
+      "loss": 0.3446,
+      "step": 5182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2054476808133151e-05,
+      "loss": 0.3982,
+      "step": 5183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051717774168718e-05,
+      "loss": 0.3432,
+      "step": 5184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2048958577148897e-05,
+      "loss": 0.3303,
+      "step": 5185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2046199217292964e-05,
+      "loss": 0.3447,
+      "step": 5186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2043439694820218e-05,
+      "loss": 0.341,
+      "step": 5187
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2040680009949966e-05,
+      "loss": 0.3597,
+      "step": 5188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037920162901522e-05,
+      "loss": 0.3408,
+      "step": 5189
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2035160153894225e-05,
+      "loss": 0.3546,
+      "step": 5190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032399983147414e-05,
+      "loss": 0.3639,
+      "step": 5191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202963965088045e-05,
+      "loss": 0.3396,
+      "step": 5192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026879157312701e-05,
+      "loss": 0.3475,
+      "step": 5193
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2024118502663557e-05,
+      "loss": 0.3635,
+      "step": 5194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2021357687152411e-05,
+      "loss": 0.3664,
+      "step": 5195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018596710998671e-05,
+      "loss": 0.3867,
+      "step": 5196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201583557442176e-05,
+      "loss": 0.3767,
+      "step": 5197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201307427764111e-05,
+      "loss": 0.373,
+      "step": 5198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010312820876174e-05,
+      "loss": 0.3939,
+      "step": 5199
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2007551204346406e-05,
+      "loss": 0.3565,
+      "step": 5200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2004789428271283e-05,
+      "loss": 0.3319,
+      "step": 5201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002027492870291e-05,
+      "loss": 0.3794,
+      "step": 5202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1999265398362932e-05,
+      "loss": 0.3889,
+      "step": 5203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1996503144968702e-05,
+      "loss": 0.363,
+      "step": 5204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1993740732907143e-05,
+      "loss": 0.3332,
+      "step": 5205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1990978162397777e-05,
+      "loss": 0.3464,
+      "step": 5206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198821543366016e-05,
+      "loss": 0.3543,
+      "step": 5207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1985452546913854e-05,
+      "loss": 0.3452,
+      "step": 5208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1982689502378425e-05,
+      "loss": 0.3418,
+      "step": 5209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197992630027347e-05,
+      "loss": 0.3258,
+      "step": 5210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197716294081858e-05,
+      "loss": 0.3811,
+      "step": 5211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197439942423337e-05,
+      "loss": 0.3537,
+      "step": 5212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1971635750737464e-05,
+      "loss": 0.3428,
+      "step": 5213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1968871920550498e-05,
+      "loss": 0.3585,
+      "step": 5214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966107933892117e-05,
+      "loss": 0.3471,
+      "step": 5215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1963343790981989e-05,
+      "loss": 0.3552,
+      "step": 5216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1960579492039783e-05,
+      "loss": 0.3702,
+      "step": 5217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1957815037285186e-05,
+      "loss": 0.3635,
+      "step": 5218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1955050426937896e-05,
+      "loss": 0.3548,
+      "step": 5219
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952285661217623e-05,
+      "loss": 0.3986,
+      "step": 5220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1949520740344095e-05,
+      "loss": 0.384,
+      "step": 5221
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946755664537045e-05,
+      "loss": 0.3529,
+      "step": 5222
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1943990434016217e-05,
+      "loss": 0.3187,
+      "step": 5223
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1941225049001375e-05,
+      "loss": 0.375,
+      "step": 5224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193845950971229e-05,
+      "loss": 0.3482,
+      "step": 5225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935693816368751e-05,
+      "loss": 0.3772,
+      "step": 5226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1932927969190546e-05,
+      "loss": 0.3231,
+      "step": 5227
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1930161968397493e-05,
+      "loss": 0.3718,
+      "step": 5228
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927395814209408e-05,
+      "loss": 0.3763,
+      "step": 5229
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1924629506846129e-05,
+      "loss": 0.336,
+      "step": 5230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921863046527498e-05,
+      "loss": 0.3716,
+      "step": 5231
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1919096433473371e-05,
+      "loss": 0.3319,
+      "step": 5232
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191632966790362e-05,
+      "loss": 0.3685,
+      "step": 5233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191356275003813e-05,
+      "loss": 0.3907,
+      "step": 5234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1910795680096795e-05,
+      "loss": 0.3796,
+      "step": 5235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1908028458299514e-05,
+      "loss": 0.3785,
+      "step": 5236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905261084866216e-05,
+      "loss": 0.3873,
+      "step": 5237
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902493560016818e-05,
+      "loss": 0.3661,
+      "step": 5238
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189972588397128e-05,
+      "loss": 0.3668,
+      "step": 5239
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896958056949543e-05,
+      "loss": 0.4872,
+      "step": 5240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1894190079171576e-05,
+      "loss": 0.3601,
+      "step": 5241
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1891421950857362e-05,
+      "loss": 0.3522,
+      "step": 5242
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1888653672226887e-05,
+      "loss": 0.368,
+      "step": 5243
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1885885243500154e-05,
+      "loss": 0.359,
+      "step": 5244
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1883116664897179e-05,
+      "loss": 0.404,
+      "step": 5245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1880347936637988e-05,
+      "loss": 0.3501,
+      "step": 5246
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1877579058942616e-05,
+      "loss": 0.3449,
+      "step": 5247
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187481003203112e-05,
+      "loss": 0.3812,
+      "step": 5248
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1872040856123552e-05,
+      "loss": 0.3932,
+      "step": 5249
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186927153143999e-05,
+      "loss": 0.3537,
+      "step": 5250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1866502058200523e-05,
+      "loss": 0.3646,
+      "step": 5251
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863732436625246e-05,
+      "loss": 0.3464,
+      "step": 5252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1860962666934268e-05,
+      "loss": 0.3893,
+      "step": 5253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1858192749347708e-05,
+      "loss": 0.3305,
+      "step": 5254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18554226840857e-05,
+      "loss": 0.4024,
+      "step": 5255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1852652471368385e-05,
+      "loss": 0.3768,
+      "step": 5256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184988211141593e-05,
+      "loss": 0.3873,
+      "step": 5257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847111604448488e-05,
+      "loss": 0.3445,
+      "step": 5258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184434095068625e-05,
+      "loss": 0.3919,
+      "step": 5259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1841570150349402e-05,
+      "loss": 0.3304,
+      "step": 5260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838799203658149e-05,
+      "loss": 0.3761,
+      "step": 5261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18360281108327e-05,
+      "loss": 0.3639,
+      "step": 5262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1833256872093288e-05,
+      "loss": 0.3473,
+      "step": 5263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1830485487660148e-05,
+      "loss": 0.3636,
+      "step": 5264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827713957753529e-05,
+      "loss": 0.3651,
+      "step": 5265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182494228259369e-05,
+      "loss": 0.3897,
+      "step": 5266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1822170462400906e-05,
+      "loss": 0.3756,
+      "step": 5267
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1819398497395455e-05,
+      "loss": 0.3879,
+      "step": 5268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181662638779764e-05,
+      "loss": 0.339,
+      "step": 5269
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1813854133827766e-05,
+      "loss": 0.3319,
+      "step": 5270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811081735706154e-05,
+      "loss": 0.3324,
+      "step": 5271
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1808309193653124e-05,
+      "loss": 0.3478,
+      "step": 5272
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805536507889021e-05,
+      "loss": 0.3699,
+      "step": 5273
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1802763678634204e-05,
+      "loss": 0.3654,
+      "step": 5274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1799990706109035e-05,
+      "loss": 0.3733,
+      "step": 5275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797217590533882e-05,
+      "loss": 0.379,
+      "step": 5276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794444332129137e-05,
+      "loss": 0.3488,
+      "step": 5277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17916709311152e-05,
+      "loss": 0.3528,
+      "step": 5278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1788897387712477e-05,
+      "loss": 0.3456,
+      "step": 5279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786123702141392e-05,
+      "loss": 0.3695,
+      "step": 5280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783349874622372e-05,
+      "loss": 0.3579,
+      "step": 5281
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1780575905375865e-05,
+      "loss": 0.3624,
+      "step": 5282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1777801794622325e-05,
+      "loss": 0.363,
+      "step": 5283
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1775027542582218e-05,
+      "loss": 0.3596,
+      "step": 5284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1772253149476017e-05,
+      "loss": 0.3865,
+      "step": 5285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1769478615524214e-05,
+      "loss": 0.4014,
+      "step": 5286
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1766703940947309e-05,
+      "loss": 0.3336,
+      "step": 5287
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1763929125965809e-05,
+      "loss": 0.3984,
+      "step": 5288
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1761154170800244e-05,
+      "loss": 0.3645,
+      "step": 5289
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1758379075671135e-05,
+      "loss": 0.3153,
+      "step": 5290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1755603840799035e-05,
+      "loss": 0.3558,
+      "step": 5291
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752828466404496e-05,
+      "loss": 0.3536,
+      "step": 5292
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1750052952708086e-05,
+      "loss": 0.3865,
+      "step": 5293
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174727729993038e-05,
+      "loss": 0.3559,
+      "step": 5294
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1744501508291968e-05,
+      "loss": 0.3137,
+      "step": 5295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741725578013448e-05,
+      "loss": 0.3406,
+      "step": 5296
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1738949509315432e-05,
+      "loss": 0.3569,
+      "step": 5297
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173617330241854e-05,
+      "loss": 0.3811,
+      "step": 5298
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733396957543409e-05,
+      "loss": 0.3712,
+      "step": 5299
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730620474910674e-05,
+      "loss": 0.342,
+      "step": 5300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1727843854740997e-05,
+      "loss": 0.3331,
+      "step": 5301
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172506709725504e-05,
+      "loss": 0.379,
+      "step": 5302
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1722290202673479e-05,
+      "loss": 0.3768,
+      "step": 5303
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1719513171217001e-05,
+      "loss": 0.3571,
+      "step": 5304
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1716736003106306e-05,
+      "loss": 0.3475,
+      "step": 5305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17139586985621e-05,
+      "loss": 0.3275,
+      "step": 5306
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1711181257805106e-05,
+      "loss": 0.3322,
+      "step": 5307
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1708403681056051e-05,
+      "loss": 0.3344,
+      "step": 5308
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1705625968535677e-05,
+      "loss": 0.3732,
+      "step": 5309
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1702848120464738e-05,
+      "loss": 0.3165,
+      "step": 5310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1700070137064e-05,
+      "loss": 0.3632,
+      "step": 5311
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1697292018554227e-05,
+      "loss": 0.3176,
+      "step": 5312
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1694513765156212e-05,
+      "loss": 0.3354,
+      "step": 5313
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1691735377090743e-05,
+      "loss": 0.3552,
+      "step": 5314
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1688956854578634e-05,
+      "loss": 0.3919,
+      "step": 5315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1686178197840696e-05,
+      "loss": 0.3759,
+      "step": 5316
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1683399407097756e-05,
+      "loss": 0.3508,
+      "step": 5317
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1680620482570655e-05,
+      "loss": 0.353,
+      "step": 5318
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1677841424480239e-05,
+      "loss": 0.4179,
+      "step": 5319
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1675062233047365e-05,
+      "loss": 0.3482,
+      "step": 5320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1672282908492907e-05,
+      "loss": 0.3847,
+      "step": 5321
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669503451037745e-05,
+      "loss": 0.388,
+      "step": 5322
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1666723860902767e-05,
+      "loss": 0.3514,
+      "step": 5323
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663944138308873e-05,
+      "loss": 0.3702,
+      "step": 5324
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1661164283476982e-05,
+      "loss": 0.37,
+      "step": 5325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1658384296628007e-05,
+      "loss": 0.3627,
+      "step": 5326
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165560417798289e-05,
+      "loss": 0.3833,
+      "step": 5327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1652823927762564e-05,
+      "loss": 0.3728,
+      "step": 5328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1650043546187994e-05,
+      "loss": 0.3422,
+      "step": 5329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164726303348014e-05,
+      "loss": 0.3761,
+      "step": 5330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1644482389859971e-05,
+      "loss": 0.3832,
+      "step": 5331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1641701615548475e-05,
+      "loss": 0.3276,
+      "step": 5332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163892071076665e-05,
+      "loss": 0.3769,
+      "step": 5333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1636139675735504e-05,
+      "loss": 0.3459,
+      "step": 5334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163335851067605e-05,
+      "loss": 0.3761,
+      "step": 5335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1630577215809311e-05,
+      "loss": 0.3543,
+      "step": 5336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1627795791356329e-05,
+      "loss": 0.3736,
+      "step": 5337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1625014237538148e-05,
+      "loss": 0.3824,
+      "step": 5338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1622232554575828e-05,
+      "loss": 0.3734,
+      "step": 5339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1619450742690434e-05,
+      "loss": 0.3446,
+      "step": 5340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1616668802103046e-05,
+      "loss": 0.3578,
+      "step": 5341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161388673303475e-05,
+      "loss": 0.3438,
+      "step": 5342
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1611104535706645e-05,
+      "loss": 0.3579,
+      "step": 5343
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1608322210339842e-05,
+      "loss": 0.35,
+      "step": 5344
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1605539757155458e-05,
+      "loss": 0.3652,
+      "step": 5345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1602757176374619e-05,
+      "loss": 0.3484,
+      "step": 5346
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159997446821847e-05,
+      "loss": 0.3436,
+      "step": 5347
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597191632908153e-05,
+      "loss": 0.3409,
+      "step": 5348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1594408670664832e-05,
+      "loss": 0.3444,
+      "step": 5349
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1591625581709676e-05,
+      "loss": 0.3701,
+      "step": 5350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1588842366263861e-05,
+      "loss": 0.3723,
+      "step": 5351
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586059024548584e-05,
+      "loss": 0.3389,
+      "step": 5352
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1583275556785036e-05,
+      "loss": 0.4104,
+      "step": 5353
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1580491963194431e-05,
+      "loss": 0.3303,
+      "step": 5354
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577708243997986e-05,
+      "loss": 0.3744,
+      "step": 5355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1574924399416933e-05,
+      "loss": 0.3584,
+      "step": 5356
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157214042967251e-05,
+      "loss": 0.36,
+      "step": 5357
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569356334985967e-05,
+      "loss": 0.3526,
+      "step": 5358
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1566572115578561e-05,
+      "loss": 0.3882,
+      "step": 5359
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1563787771671564e-05,
+      "loss": 0.3788,
+      "step": 5360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1561003303486255e-05,
+      "loss": 0.3704,
+      "step": 5361
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1558218711243923e-05,
+      "loss": 0.3597,
+      "step": 5362
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1555433995165864e-05,
+      "loss": 0.3449,
+      "step": 5363
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155264915547339e-05,
+      "loss": 0.3595,
+      "step": 5364
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1549864192387819e-05,
+      "loss": 0.3208,
+      "step": 5365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1547079106130478e-05,
+      "loss": 0.3347,
+      "step": 5366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1544293896922705e-05,
+      "loss": 0.3573,
+      "step": 5367
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1541508564985848e-05,
+      "loss": 0.3359,
+      "step": 5368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1538723110541267e-05,
+      "loss": 0.3673,
+      "step": 5369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1535937533810328e-05,
+      "loss": 0.3349,
+      "step": 5370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1533151835014406e-05,
+      "loss": 0.3982,
+      "step": 5371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1530366014374892e-05,
+      "loss": 0.394,
+      "step": 5372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1527580072113176e-05,
+      "loss": 0.3469,
+      "step": 5373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152479400845067e-05,
+      "loss": 0.3806,
+      "step": 5374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152200782360879e-05,
+      "loss": 0.3823,
+      "step": 5375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1519221517808956e-05,
+      "loss": 0.3836,
+      "step": 5376
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1516435091272606e-05,
+      "loss": 0.3425,
+      "step": 5377
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1513648544221188e-05,
+      "loss": 0.3283,
+      "step": 5378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151086187687615e-05,
+      "loss": 0.3661,
+      "step": 5379
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508075089458959e-05,
+      "loss": 0.3512,
+      "step": 5380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150528818219109e-05,
+      "loss": 0.3585,
+      "step": 5381
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502501155294024e-05,
+      "loss": 0.3554,
+      "step": 5382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1499714008989252e-05,
+      "loss": 0.3373,
+      "step": 5383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1496926743498275e-05,
+      "loss": 0.3648,
+      "step": 5384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1494139359042612e-05,
+      "loss": 0.335,
+      "step": 5385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1491351855843773e-05,
+      "loss": 0.3408,
+      "step": 5386
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1488564234123297e-05,
+      "loss": 0.3504,
+      "step": 5387
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1485776494102718e-05,
+      "loss": 0.3826,
+      "step": 5388
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148298863600359e-05,
+      "loss": 0.3445,
+      "step": 5389
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1480200660047462e-05,
+      "loss": 0.3336,
+      "step": 5390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1477412566455912e-05,
+      "loss": 0.3298,
+      "step": 5391
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1474624355450515e-05,
+      "loss": 0.3304,
+      "step": 5392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471836027252854e-05,
+      "loss": 0.363,
+      "step": 5393
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1469047582084524e-05,
+      "loss": 0.3622,
+      "step": 5394
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1466259020167135e-05,
+      "loss": 0.4049,
+      "step": 5395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1463470341722298e-05,
+      "loss": 0.3313,
+      "step": 5396
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1460681546971637e-05,
+      "loss": 0.3445,
+      "step": 5397
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1457892636136787e-05,
+      "loss": 0.4003,
+      "step": 5398
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1455103609439387e-05,
+      "loss": 0.3786,
+      "step": 5399
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145231446710109e-05,
+      "loss": 0.3806,
+      "step": 5400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449525209343552e-05,
+      "loss": 0.3374,
+      "step": 5401
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144673583638845e-05,
+      "loss": 0.3421,
+      "step": 5402
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144394634845746e-05,
+      "loss": 0.3519,
+      "step": 5403
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441156745772269e-05,
+      "loss": 0.3559,
+      "step": 5404
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1438367028554574e-05,
+      "loss": 0.3855,
+      "step": 5405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1435577197026082e-05,
+      "loss": 0.3602,
+      "step": 5406
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432787251408506e-05,
+      "loss": 0.3571,
+      "step": 5407
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429997191923572e-05,
+      "loss": 0.3451,
+      "step": 5408
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1427207018793014e-05,
+      "loss": 0.3754,
+      "step": 5409
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1424416732238574e-05,
+      "loss": 0.3414,
+      "step": 5410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1421626332482003e-05,
+      "loss": 0.348,
+      "step": 5411
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1418835819745061e-05,
+      "loss": 0.3751,
+      "step": 5412
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1416045194249517e-05,
+      "loss": 0.3706,
+      "step": 5413
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413254456217152e-05,
+      "loss": 0.3457,
+      "step": 5414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141046360586975e-05,
+      "loss": 0.4174,
+      "step": 5415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1407672643429108e-05,
+      "loss": 0.3694,
+      "step": 5416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404881569117034e-05,
+      "loss": 0.3718,
+      "step": 5417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402090383155338e-05,
+      "loss": 0.3544,
+      "step": 5418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399299085765845e-05,
+      "loss": 0.3152,
+      "step": 5419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1396507677170386e-05,
+      "loss": 0.3546,
+      "step": 5420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1393716157590801e-05,
+      "loss": 0.3532,
+      "step": 5421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1390924527248941e-05,
+      "loss": 0.3668,
+      "step": 5422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1388132786366663e-05,
+      "loss": 0.3356,
+      "step": 5423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1385340935165838e-05,
+      "loss": 0.4186,
+      "step": 5424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1382548973868333e-05,
+      "loss": 0.3634,
+      "step": 5425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379756902696038e-05,
+      "loss": 0.3581,
+      "step": 5426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1376964721870847e-05,
+      "loss": 0.3519,
+      "step": 5427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1374172431614663e-05,
+      "loss": 0.3665,
+      "step": 5428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371380032149393e-05,
+      "loss": 0.3563,
+      "step": 5429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1368587523696955e-05,
+      "loss": 0.3423,
+      "step": 5430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1365794906479279e-05,
+      "loss": 0.3923,
+      "step": 5431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1363002180718303e-05,
+      "loss": 0.3411,
+      "step": 5432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1360209346635975e-05,
+      "loss": 0.3732,
+      "step": 5433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1357416404454244e-05,
+      "loss": 0.3624,
+      "step": 5434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354623354395074e-05,
+      "loss": 0.3619,
+      "step": 5435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1351830196680433e-05,
+      "loss": 0.3394,
+      "step": 5436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1349036931532305e-05,
+      "loss": 0.3692,
+      "step": 5437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1346243559172674e-05,
+      "loss": 0.3292,
+      "step": 5438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343450079823542e-05,
+      "loss": 0.3412,
+      "step": 5439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134065649370691e-05,
+      "loss": 0.3536,
+      "step": 5440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1337862801044792e-05,
+      "loss": 0.3467,
+      "step": 5441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1335069002059211e-05,
+      "loss": 0.3673,
+      "step": 5442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13322750969722e-05,
+      "loss": 0.3402,
+      "step": 5443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329481086005791e-05,
+      "loss": 0.3329,
+      "step": 5444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132668696938204e-05,
+      "loss": 0.3436,
+      "step": 5445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1323892747322995e-05,
+      "loss": 0.343,
+      "step": 5446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1321098420050727e-05,
+      "loss": 0.3612,
+      "step": 5447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13183039877873e-05,
+      "loss": 0.3455,
+      "step": 5448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1315509450754805e-05,
+      "loss": 0.3609,
+      "step": 5449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1312714809175322e-05,
+      "loss": 0.3214,
+      "step": 5450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1309920063270959e-05,
+      "loss": 0.3841,
+      "step": 5451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130712521326381e-05,
+      "loss": 0.3857,
+      "step": 5452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1304330259375995e-05,
+      "loss": 0.3207,
+      "step": 5453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301535201829634e-05,
+      "loss": 0.3602,
+      "step": 5454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1298740040846863e-05,
+      "loss": 0.3637,
+      "step": 5455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1295944776649817e-05,
+      "loss": 0.3707,
+      "step": 5456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1293149409460642e-05,
+      "loss": 0.3027,
+      "step": 5457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1290353939501492e-05,
+      "loss": 0.3429,
+      "step": 5458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287558366994536e-05,
+      "loss": 0.3892,
+      "step": 5459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128476269216194e-05,
+      "loss": 0.3492,
+      "step": 5460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281966915225886e-05,
+      "loss": 0.3721,
+      "step": 5461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127917103640856e-05,
+      "loss": 0.388,
+      "step": 5462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1276375055932163e-05,
+      "loss": 0.3588,
+      "step": 5463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1273578974018893e-05,
+      "loss": 0.3396,
+      "step": 5464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1270782790890965e-05,
+      "loss": 0.3286,
+      "step": 5465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1267986506770594e-05,
+      "loss": 0.3326,
+      "step": 5466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1265190121880014e-05,
+      "loss": 0.3755,
+      "step": 5467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126239363644146e-05,
+      "loss": 0.3548,
+      "step": 5468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1259597050677178e-05,
+      "loss": 0.4015,
+      "step": 5469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1256800364809415e-05,
+      "loss": 0.3821,
+      "step": 5470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1254003579060431e-05,
+      "loss": 0.357,
+      "step": 5471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1251206693652494e-05,
+      "loss": 0.3601,
+      "step": 5472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1248409708807889e-05,
+      "loss": 0.3492,
+      "step": 5473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1245612624748887e-05,
+      "loss": 0.3641,
+      "step": 5474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1242815441697788e-05,
+      "loss": 0.3518,
+      "step": 5475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1240018159876886e-05,
+      "loss": 0.3513,
+      "step": 5476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1237220779508493e-05,
+      "loss": 0.3833,
+      "step": 5477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123442330081492e-05,
+      "loss": 0.3373,
+      "step": 5478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1231625724018494e-05,
+      "loss": 0.3749,
+      "step": 5479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228828049341543e-05,
+      "loss": 0.3555,
+      "step": 5480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1226030277006404e-05,
+      "loss": 0.3813,
+      "step": 5481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1223232407235427e-05,
+      "loss": 0.4132,
+      "step": 5482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1220434440250966e-05,
+      "loss": 0.3717,
+      "step": 5483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121763637627538e-05,
+      "loss": 0.3759,
+      "step": 5484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121483821553104e-05,
+      "loss": 0.3342,
+      "step": 5485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1212039958240323e-05,
+      "loss": 0.3221,
+      "step": 5486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1209241604625618e-05,
+      "loss": 0.3522,
+      "step": 5487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120644315490931e-05,
+      "loss": 0.3524,
+      "step": 5488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12036446093138e-05,
+      "loss": 0.3934,
+      "step": 5489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12008459680615e-05,
+      "loss": 0.3785,
+      "step": 5490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198047231374827e-05,
+      "loss": 0.3312,
+      "step": 5491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1195248399476197e-05,
+      "loss": 0.383,
+      "step": 5492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192449472588046e-05,
+      "loss": 0.3703,
+      "step": 5493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1189650450932808e-05,
+      "loss": 0.3954,
+      "step": 5494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118685133473293e-05,
+      "loss": 0.3373,
+      "step": 5495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1184052124210869e-05,
+      "loss": 0.4068,
+      "step": 5496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1181252819589081e-05,
+      "loss": 0.3761,
+      "step": 5497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1178453421090037e-05,
+      "loss": 0.3501,
+      "step": 5498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175653928936207e-05,
+      "loss": 0.3635,
+      "step": 5499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117285434335008e-05,
+      "loss": 0.3659,
+      "step": 5500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170054664554147e-05,
+      "loss": 0.3663,
+      "step": 5501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11672548927709e-05,
+      "loss": 0.3515,
+      "step": 5502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116445502822285e-05,
+      "loss": 0.3607,
+      "step": 5503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1161655071132504e-05,
+      "loss": 0.3571,
+      "step": 5504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1158855021722388e-05,
+      "loss": 0.3382,
+      "step": 5505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1156054880215027e-05,
+      "loss": 0.4044,
+      "step": 5506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1153254646832947e-05,
+      "loss": 0.3975,
+      "step": 5507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1150454321798705e-05,
+      "loss": 0.3695,
+      "step": 5508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147653905334841e-05,
+      "loss": 0.3159,
+      "step": 5509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1144853397663915e-05,
+      "loss": 0.3752,
+      "step": 5510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1142052799008486e-05,
+      "loss": 0.3293,
+      "step": 5511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139252109591127e-05,
+      "loss": 0.3756,
+      "step": 5512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1136451329634418e-05,
+      "loss": 0.3573,
+      "step": 5513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133650459360948e-05,
+      "loss": 0.3934,
+      "step": 5514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1130849498993299e-05,
+      "loss": 0.3916,
+      "step": 5515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1128048448754078e-05,
+      "loss": 0.3491,
+      "step": 5516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1125247308865892e-05,
+      "loss": 0.3399,
+      "step": 5517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1122446079551352e-05,
+      "loss": 0.3847,
+      "step": 5518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119644761033079e-05,
+      "loss": 0.3767,
+      "step": 5519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1116843353533704e-05,
+      "loss": 0.3573,
+      "step": 5520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1114041857275862e-05,
+      "loss": 0.3219,
+      "step": 5521
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111124027248219e-05,
+      "loss": 0.3377,
+      "step": 5522
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108438599375346e-05,
+      "loss": 0.353,
+      "step": 5523
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1105636838177976e-05,
+      "loss": 0.3531,
+      "step": 5524
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102834989112752e-05,
+      "loss": 0.3587,
+      "step": 5525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110003305240234e-05,
+      "loss": 0.3604,
+      "step": 5526
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097231028269417e-05,
+      "loss": 0.3464,
+      "step": 5527
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1094428916936672e-05,
+      "loss": 0.3475,
+      "step": 5528
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1091626718626791e-05,
+      "loss": 0.3268,
+      "step": 5529
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088824433562471e-05,
+      "loss": 0.3571,
+      "step": 5530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086022061966422e-05,
+      "loss": 0.3508,
+      "step": 5531
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083219604061354e-05,
+      "loss": 0.3631,
+      "step": 5532
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1080417060069986e-05,
+      "loss": 0.3212,
+      "step": 5533
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107761443021504e-05,
+      "loss": 0.3592,
+      "step": 5534
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1074811714719251e-05,
+      "loss": 0.3735,
+      "step": 5535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107200891380536e-05,
+      "loss": 0.3743,
+      "step": 5536
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106920602769611e-05,
+      "loss": 0.3848,
+      "step": 5537
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1066403056614252e-05,
+      "loss": 0.3447,
+      "step": 5538
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106360000078255e-05,
+      "loss": 0.3757,
+      "step": 5539
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106079686042377e-05,
+      "loss": 0.3479,
+      "step": 5540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1057993635760683e-05,
+      "loss": 0.3396,
+      "step": 5541
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055190327016065e-05,
+      "loss": 0.3686,
+      "step": 5542
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1052386934412711e-05,
+      "loss": 0.3834,
+      "step": 5543
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1049583458173408e-05,
+      "loss": 0.3678,
+      "step": 5544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1046779898520954e-05,
+      "loss": 0.351,
+      "step": 5545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1043976255678163e-05,
+      "loss": 0.3161,
+      "step": 5546
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1041172529867839e-05,
+      "loss": 0.34,
+      "step": 5547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1038368721312804e-05,
+      "loss": 0.347,
+      "step": 5548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1035564830235887e-05,
+      "loss": 0.3698,
+      "step": 5549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1032760856859923e-05,
+      "loss": 0.3362,
+      "step": 5550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1029956801407742e-05,
+      "loss": 0.3324,
+      "step": 5551
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1027152664102197e-05,
+      "loss": 0.3402,
+      "step": 5552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1024348445166133e-05,
+      "loss": 0.369,
+      "step": 5553
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1021544144822418e-05,
+      "loss": 0.3419,
+      "step": 5554
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1018739763293912e-05,
+      "loss": 0.3745,
+      "step": 5555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015935300803487e-05,
+      "loss": 0.3802,
+      "step": 5556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101313075757402e-05,
+      "loss": 0.3612,
+      "step": 5557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1010326133828395e-05,
+      "loss": 0.3318,
+      "step": 5558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007521429789506e-05,
+      "loss": 0.3247,
+      "step": 5559
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100471664568025e-05,
+      "loss": 0.3358,
+      "step": 5560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1001911781723524e-05,
+      "loss": 0.3461,
+      "step": 5561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999106838142246e-05,
+      "loss": 0.4264,
+      "step": 5562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0996301815159329e-05,
+      "loss": 0.3662,
+      "step": 5563
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0993496712997698e-05,
+      "loss": 0.3692,
+      "step": 5564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0990691531880275e-05,
+      "loss": 0.3405,
+      "step": 5565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987886272030002e-05,
+      "loss": 0.3478,
+      "step": 5566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0985080933669816e-05,
+      "loss": 0.3145,
+      "step": 5567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098227551702267e-05,
+      "loss": 0.3919,
+      "step": 5568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0979470022311512e-05,
+      "loss": 0.3683,
+      "step": 5569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0976664449759302e-05,
+      "loss": 0.3515,
+      "step": 5570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0973858799589011e-05,
+      "loss": 0.3618,
+      "step": 5571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0971053072023607e-05,
+      "loss": 0.3617,
+      "step": 5572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0968247267286073e-05,
+      "loss": 0.3526,
+      "step": 5573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096544138559939e-05,
+      "loss": 0.3287,
+      "step": 5574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0962635427186547e-05,
+      "loss": 0.3391,
+      "step": 5575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0959829392270547e-05,
+      "loss": 0.3225,
+      "step": 5576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0957023281074388e-05,
+      "loss": 0.3765,
+      "step": 5577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0954217093821082e-05,
+      "loss": 0.3938,
+      "step": 5578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0951410830733642e-05,
+      "loss": 0.3732,
+      "step": 5579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094860449203509e-05,
+      "loss": 0.3372,
+      "step": 5580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094579807794845e-05,
+      "loss": 0.359,
+      "step": 5581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094299158869676e-05,
+      "loss": 0.3257,
+      "step": 5582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0940185024503059e-05,
+      "loss": 0.3499,
+      "step": 5583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0937378385590387e-05,
+      "loss": 0.402,
+      "step": 5584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.09345716721818e-05,
+      "loss": 0.3788,
+      "step": 5585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0931764884500353e-05,
+      "loss": 0.3558,
+      "step": 5586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928958022769108e-05,
+      "loss": 0.3826,
+      "step": 5587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092615108721113e-05,
+      "loss": 0.3725,
+      "step": 5588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923344078049503e-05,
+      "loss": 0.3734,
+      "step": 5589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0920536995507298e-05,
+      "loss": 0.3225,
+      "step": 5590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0917729839807612e-05,
+      "loss": 0.3372,
+      "step": 5591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0914922611173526e-05,
+      "loss": 0.35,
+      "step": 5592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0912115309828141e-05,
+      "loss": 0.3922,
+      "step": 5593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0909307935994563e-05,
+      "loss": 0.389,
+      "step": 5594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0906500489895901e-05,
+      "loss": 0.3494,
+      "step": 5595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903692971755269e-05,
+      "loss": 0.3435,
+      "step": 5596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0900885381795788e-05,
+      "loss": 0.3423,
+      "step": 5597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0898077720240582e-05,
+      "loss": 0.3367,
+      "step": 5598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0895269987312788e-05,
+      "loss": 0.4058,
+      "step": 5599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892462183235541e-05,
+      "loss": 0.3691,
+      "step": 5600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0889654308231987e-05,
+      "loss": 0.3328,
+      "step": 5601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886846362525271e-05,
+      "loss": 0.3354,
+      "step": 5602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0884038346338552e-05,
+      "loss": 0.3745,
+      "step": 5603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0881230259894991e-05,
+      "loss": 0.335,
+      "step": 5604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087842210341775e-05,
+      "loss": 0.3767,
+      "step": 5605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0875613877129998e-05,
+      "loss": 0.3418,
+      "step": 5606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872805581254919e-05,
+      "loss": 0.34,
+      "step": 5607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0869997216015693e-05,
+      "loss": 0.3505,
+      "step": 5608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086718878163551e-05,
+      "loss": 0.3503,
+      "step": 5609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086438027833756e-05,
+      "loss": 0.3701,
+      "step": 5610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0861571706345044e-05,
+      "loss": 0.3622,
+      "step": 5611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0858763065881164e-05,
+      "loss": 0.3696,
+      "step": 5612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0855954357169137e-05,
+      "loss": 0.3668,
+      "step": 5613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0853145580432172e-05,
+      "loss": 0.3857,
+      "step": 5614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085033673589349e-05,
+      "loss": 0.39,
+      "step": 5615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084752782377632e-05,
+      "loss": 0.3379,
+      "step": 5616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0844718844303894e-05,
+      "loss": 0.3444,
+      "step": 5617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0841909797699446e-05,
+      "loss": 0.3422,
+      "step": 5618
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0839100684186219e-05,
+      "loss": 0.3441,
+      "step": 5619
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0836291503987463e-05,
+      "loss": 0.3622,
+      "step": 5620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0833482257326429e-05,
+      "loss": 0.3396,
+      "step": 5621
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0830672944426374e-05,
+      "loss": 0.3513,
+      "step": 5622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0827863565510565e-05,
+      "loss": 0.3734,
+      "step": 5623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825054120802267e-05,
+      "loss": 0.3996,
+      "step": 5624
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082224461052476e-05,
+      "loss": 0.382,
+      "step": 5625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0819435034901314e-05,
+      "loss": 0.3398,
+      "step": 5626
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0816625394155223e-05,
+      "loss": 0.3348,
+      "step": 5627
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0813815688509766e-05,
+      "loss": 0.3414,
+      "step": 5628
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0811005918188244e-05,
+      "loss": 0.3389,
+      "step": 5629
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0808196083413958e-05,
+      "loss": 0.3454,
+      "step": 5630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0805386184410214e-05,
+      "loss": 0.3391,
+      "step": 5631
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0802576221400314e-05,
+      "loss": 0.3525,
+      "step": 5632
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0799766194607581e-05,
+      "loss": 0.3525,
+      "step": 5633
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0796956104255332e-05,
+      "loss": 0.3577,
+      "step": 5634
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0794145950566893e-05,
+      "loss": 0.3562,
+      "step": 5635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0791335733765591e-05,
+      "loss": 0.3656,
+      "step": 5636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0788525454074765e-05,
+      "loss": 0.362,
+      "step": 5637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785715111717756e-05,
+      "loss": 0.3533,
+      "step": 5638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078290470691791e-05,
+      "loss": 0.3291,
+      "step": 5639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780094239898571e-05,
+      "loss": 0.363,
+      "step": 5640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07772837108831e-05,
+      "loss": 0.324,
+      "step": 5641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0774473120094855e-05,
+      "loss": 0.3604,
+      "step": 5642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0771662467757203e-05,
+      "loss": 0.3794,
+      "step": 5643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076885175409351e-05,
+      "loss": 0.4079,
+      "step": 5644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0766040979327155e-05,
+      "loss": 0.3479,
+      "step": 5645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0763230143681514e-05,
+      "loss": 0.3302,
+      "step": 5646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760419247379972e-05,
+      "loss": 0.3606,
+      "step": 5647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075760829064592e-05,
+      "loss": 0.3577,
+      "step": 5648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0754797273702755e-05,
+      "loss": 0.3761,
+      "step": 5649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751986196773871e-05,
+      "loss": 0.3426,
+      "step": 5650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0749175060082671e-05,
+      "loss": 0.3886,
+      "step": 5651
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0746363863852565e-05,
+      "loss": 0.3598,
+      "step": 5652
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074355260830697e-05,
+      "loss": 0.3667,
+      "step": 5653
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0740741293669298e-05,
+      "loss": 0.3529,
+      "step": 5654
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0737929920162975e-05,
+      "loss": 0.3308,
+      "step": 5655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0735118488011427e-05,
+      "loss": 0.3659,
+      "step": 5656
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0732306997438084e-05,
+      "loss": 0.3552,
+      "step": 5657
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0729495448666387e-05,
+      "loss": 0.3707,
+      "step": 5658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0726683841919774e-05,
+      "loss": 0.3505,
+      "step": 5659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723872177421692e-05,
+      "loss": 0.3392,
+      "step": 5660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.072106045539559e-05,
+      "loss": 0.3579,
+      "step": 5661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718248676064923e-05,
+      "loss": 0.3513,
+      "step": 5662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0715436839653157e-05,
+      "loss": 0.3539,
+      "step": 5663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0712624946383743e-05,
+      "loss": 0.3996,
+      "step": 5664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070981299648016e-05,
+      "loss": 0.3441,
+      "step": 5665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707000990165877e-05,
+      "loss": 0.3478,
+      "step": 5666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0704188927664375e-05,
+      "loss": 0.3675,
+      "step": 5667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0701376809199131e-05,
+      "loss": 0.3447,
+      "step": 5668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0698564634993634e-05,
+      "loss": 0.339,
+      "step": 5669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0695752405271371e-05,
+      "loss": 0.3615,
+      "step": 5670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0692940120255846e-05,
+      "loss": 0.3921,
+      "step": 5671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0690127780170553e-05,
+      "loss": 0.3463,
+      "step": 5672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0687315385238997e-05,
+      "loss": 0.4023,
+      "step": 5673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0684502935684684e-05,
+      "loss": 0.3353,
+      "step": 5674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0681690431731129e-05,
+      "loss": 0.3418,
+      "step": 5675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678877873601847e-05,
+      "loss": 0.3115,
+      "step": 5676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0676065261520362e-05,
+      "loss": 0.3847,
+      "step": 5677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673252595710202e-05,
+      "loss": 0.3447,
+      "step": 5678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.067043987639489e-05,
+      "loss": 0.36,
+      "step": 5679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0667627103797967e-05,
+      "loss": 0.3328,
+      "step": 5680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0664814278142965e-05,
+      "loss": 0.3375,
+      "step": 5681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0662001399653433e-05,
+      "loss": 0.3499,
+      "step": 5682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659188468552912e-05,
+      "loss": 0.3788,
+      "step": 5683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0656375485064956e-05,
+      "loss": 0.3601,
+      "step": 5684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0653562449413122e-05,
+      "loss": 0.4133,
+      "step": 5685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0650749361820967e-05,
+      "loss": 0.3257,
+      "step": 5686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0647936222512052e-05,
+      "loss": 0.3326,
+      "step": 5687
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.064512303170995e-05,
+      "loss": 0.363,
+      "step": 5688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0642309789638228e-05,
+      "loss": 0.343,
+      "step": 5689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0639496496520466e-05,
+      "loss": 0.352,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063668315258024e-05,
+      "loss": 0.3737,
+      "step": 5691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0633869758041135e-05,
+      "loss": 0.3369,
+      "step": 5692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0631056313126736e-05,
+      "loss": 0.3791,
+      "step": 5693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0628242818060644e-05,
+      "loss": 0.3454,
+      "step": 5694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0625429273066445e-05,
+      "loss": 0.3435,
+      "step": 5695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0622615678367742e-05,
+      "loss": 0.3336,
+      "step": 5696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061980203418814e-05,
+      "loss": 0.3595,
+      "step": 5697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616988340751247e-05,
+      "loss": 0.3504,
+      "step": 5698
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0614174598280673e-05,
+      "loss": 0.3136,
+      "step": 5699
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611360807000033e-05,
+      "loss": 0.3557,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0608546967132949e-05,
+      "loss": 0.3833,
+      "step": 5701
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060573307890304e-05,
+      "loss": 0.39,
+      "step": 5702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602919142533939e-05,
+      "loss": 0.3456,
+      "step": 5703
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0600105158249272e-05,
+      "loss": 0.361,
+      "step": 5704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0597291126272673e-05,
+      "loss": 0.3091,
+      "step": 5705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0594477046827785e-05,
+      "loss": 0.3562,
+      "step": 5706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0591662920138248e-05,
+      "loss": 0.3767,
+      "step": 5707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0588848746427709e-05,
+      "loss": 0.3621,
+      "step": 5708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586034525919817e-05,
+      "loss": 0.3437,
+      "step": 5709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0583220258838222e-05,
+      "loss": 0.3805,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0580405945406585e-05,
+      "loss": 0.3608,
+      "step": 5711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057759158584857e-05,
+      "loss": 0.333,
+      "step": 5712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0574777180387836e-05,
+      "loss": 0.3809,
+      "step": 5713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571962729248053e-05,
+      "loss": 0.3804,
+      "step": 5714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0569148232652893e-05,
+      "loss": 0.3546,
+      "step": 5715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0566333690826029e-05,
+      "loss": 0.3912,
+      "step": 5716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0563519103991145e-05,
+      "loss": 0.3409,
+      "step": 5717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560704472371919e-05,
+      "loss": 0.3169,
+      "step": 5718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055788979619204e-05,
+      "loss": 0.398,
+      "step": 5719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0555075075675196e-05,
+      "loss": 0.4136,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552260311045082e-05,
+      "loss": 0.3088,
+      "step": 5721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549445502525394e-05,
+      "loss": 0.3565,
+      "step": 5722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054663065033983e-05,
+      "loss": 0.3474,
+      "step": 5723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0543815754712097e-05,
+      "loss": 0.3493,
+      "step": 5724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05410008158659e-05,
+      "loss": 0.315,
+      "step": 5725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0538185834024955e-05,
+      "loss": 0.3469,
+      "step": 5726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0535370809412965e-05,
+      "loss": 0.3534,
+      "step": 5727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0532555742253654e-05,
+      "loss": 0.3602,
+      "step": 5728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0529740632770743e-05,
+      "loss": 0.3332,
+      "step": 5729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052692548118796e-05,
+      "loss": 0.3217,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0524110287729024e-05,
+      "loss": 0.3926,
+      "step": 5731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052129505261767e-05,
+      "loss": 0.39,
+      "step": 5732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0518479776077633e-05,
+      "loss": 0.3533,
+      "step": 5733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0515664458332646e-05,
+      "loss": 0.3787,
+      "step": 5734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512849099606458e-05,
+      "loss": 0.373,
+      "step": 5735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0510033700122804e-05,
+      "loss": 0.3946,
+      "step": 5736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0507218260105437e-05,
+      "loss": 0.3374,
+      "step": 5737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0504402779778104e-05,
+      "loss": 0.3601,
+      "step": 5738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050158725936456e-05,
+      "loss": 0.313,
+      "step": 5739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0498771699088566e-05,
+      "loss": 0.3535,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0495956099173874e-05,
+      "loss": 0.3718,
+      "step": 5741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049314045984425e-05,
+      "loss": 0.3629,
+      "step": 5742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0490324781323464e-05,
+      "loss": 0.3759,
+      "step": 5743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0487509063835283e-05,
+      "loss": 0.3854,
+      "step": 5744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0484693307603478e-05,
+      "loss": 0.3578,
+      "step": 5745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0481877512851823e-05,
+      "loss": 0.3881,
+      "step": 5746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04790616798041e-05,
+      "loss": 0.3581,
+      "step": 5747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476245808684094e-05,
+      "loss": 0.3407,
+      "step": 5748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047342989971558e-05,
+      "loss": 0.3423,
+      "step": 5749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0470613953122355e-05,
+      "loss": 0.3578,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0467797969128202e-05,
+      "loss": 0.3521,
+      "step": 5751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046498194795692e-05,
+      "loss": 0.3705,
+      "step": 5752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0462165889832308e-05,
+      "loss": 0.3443,
+      "step": 5753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0459349794978157e-05,
+      "loss": 0.3584,
+      "step": 5754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0456533663618273e-05,
+      "loss": 0.3506,
+      "step": 5755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453717495976466e-05,
+      "loss": 0.3957,
+      "step": 5756
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0450901292276536e-05,
+      "loss": 0.3541,
+      "step": 5757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0448085052742302e-05,
+      "loss": 0.3366,
+      "step": 5758
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0445268777597573e-05,
+      "loss": 0.385,
+      "step": 5759
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442452467066169e-05,
+      "loss": 0.3524,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0439636121371906e-05,
+      "loss": 0.3676,
+      "step": 5761
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043681974073861e-05,
+      "loss": 0.3627,
+      "step": 5762
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04340033253901e-05,
+      "loss": 0.3165,
+      "step": 5763
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0431186875550212e-05,
+      "loss": 0.3762,
+      "step": 5764
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0428370391442773e-05,
+      "loss": 0.3355,
+      "step": 5765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0425553873291614e-05,
+      "loss": 0.36,
+      "step": 5766
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0422737321320577e-05,
+      "loss": 0.3741,
+      "step": 5767
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0419920735753495e-05,
+      "loss": 0.3716,
+      "step": 5768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041710411681421e-05,
+      "loss": 0.3652,
+      "step": 5769
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0414287464726567e-05,
+      "loss": 0.3435,
+      "step": 5770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411470779714419e-05,
+      "loss": 0.3555,
+      "step": 5771
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0408654062001604e-05,
+      "loss": 0.3656,
+      "step": 5772
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0405837311811983e-05,
+      "loss": 0.3612,
+      "step": 5773
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0403020529369405e-05,
+      "loss": 0.3376,
+      "step": 5774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0400203714897733e-05,
+      "loss": 0.378,
+      "step": 5775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0397386868620818e-05,
+      "loss": 0.3094,
+      "step": 5776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0394569990762528e-05,
+      "loss": 0.3991,
+      "step": 5777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0391753081546729e-05,
+      "loss": 0.359,
+      "step": 5778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0388936141197284e-05,
+      "loss": 0.3247,
+      "step": 5779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0386119169938068e-05,
+      "loss": 0.3767,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0383302167992944e-05,
+      "loss": 0.3596,
+      "step": 5781
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0380485135585796e-05,
+      "loss": 0.3691,
+      "step": 5782
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0377668072940494e-05,
+      "loss": 0.3421,
+      "step": 5783
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0374850980280924e-05,
+      "loss": 0.3452,
+      "step": 5784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0372033857830962e-05,
+      "loss": 0.356,
+      "step": 5785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0369216705814496e-05,
+      "loss": 0.3571,
+      "step": 5786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366399524455408e-05,
+      "loss": 0.3981,
+      "step": 5787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0363582313977591e-05,
+      "loss": 0.3844,
+      "step": 5788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0360765074604938e-05,
+      "loss": 0.342,
+      "step": 5789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0357947806561335e-05,
+      "loss": 0.3507,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355130510070682e-05,
+      "loss": 0.3444,
+      "step": 5791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0352313185356875e-05,
+      "loss": 0.3535,
+      "step": 5792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0349495832643819e-05,
+      "loss": 0.287,
+      "step": 5793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0346678452155412e-05,
+      "loss": 0.3508,
+      "step": 5794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0343861044115559e-05,
+      "loss": 0.3453,
+      "step": 5795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341043608748169e-05,
+      "loss": 0.3444,
+      "step": 5796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0338226146277148e-05,
+      "loss": 0.3222,
+      "step": 5797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335408656926409e-05,
+      "loss": 0.3938,
+      "step": 5798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0332591140919866e-05,
+      "loss": 0.3361,
+      "step": 5799
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0329773598481432e-05,
+      "loss": 0.3496,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0326956029835027e-05,
+      "loss": 0.3247,
+      "step": 5801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.032413843520457e-05,
+      "loss": 0.3486,
+      "step": 5802
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321320814813983e-05,
+      "loss": 0.4021,
+      "step": 5803
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0318503168887183e-05,
+      "loss": 0.3319,
+      "step": 5804
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0315685497648107e-05,
+      "loss": 0.3886,
+      "step": 5805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0312867801320674e-05,
+      "loss": 0.365,
+      "step": 5806
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031005008012882e-05,
+      "loss": 0.3695,
+      "step": 5807
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0307232334296472e-05,
+      "loss": 0.3409,
+      "step": 5808
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0304414564047565e-05,
+      "loss": 0.3624,
+      "step": 5809
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0301596769606032e-05,
+      "loss": 0.3567,
+      "step": 5810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0298778951195818e-05,
+      "loss": 0.3753,
+      "step": 5811
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0295961109040855e-05,
+      "loss": 0.3656,
+      "step": 5812
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0293143243365087e-05,
+      "loss": 0.3473,
+      "step": 5813
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0290325354392458e-05,
+      "loss": 0.3631,
+      "step": 5814
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028750744234691e-05,
+      "loss": 0.3556,
+      "step": 5815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0284689507452397e-05,
+      "loss": 0.3588,
+      "step": 5816
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0281871549932858e-05,
+      "loss": 0.3518,
+      "step": 5817
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027905357001225e-05,
+      "loss": 0.3553,
+      "step": 5818
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0276235567914523e-05,
+      "loss": 0.3819,
+      "step": 5819
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027341754386363e-05,
+      "loss": 0.3522,
+      "step": 5820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0270599498083532e-05,
+      "loss": 0.3415,
+      "step": 5821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.026778143079818e-05,
+      "loss": 0.3505,
+      "step": 5822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0264963342231536e-05,
+      "loss": 0.3542,
+      "step": 5823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.026214523260756e-05,
+      "loss": 0.3259,
+      "step": 5824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0259327102150218e-05,
+      "loss": 0.348,
+      "step": 5825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025650895108347e-05,
+      "loss": 0.3725,
+      "step": 5826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253690779631282e-05,
+      "loss": 0.4059,
+      "step": 5827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0250872588017625e-05,
+      "loss": 0.3659,
+      "step": 5828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0248054376466465e-05,
+      "loss": 0.3688,
+      "step": 5829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0245236145201776e-05,
+      "loss": 0.3316,
+      "step": 5830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0242417894447524e-05,
+      "loss": 0.4516,
+      "step": 5831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0239599624427688e-05,
+      "loss": 0.3601,
+      "step": 5832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0236781335366239e-05,
+      "loss": 0.4079,
+      "step": 5833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0233963027487162e-05,
+      "loss": 0.3392,
+      "step": 5834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0231144701014428e-05,
+      "loss": 0.404,
+      "step": 5835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0228326356172017e-05,
+      "loss": 0.3556,
+      "step": 5836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0225507993183915e-05,
+      "loss": 0.3522,
+      "step": 5837
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02226896122741e-05,
+      "loss": 0.3578,
+      "step": 5838
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219871213666557e-05,
+      "loss": 0.3683,
+      "step": 5839
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0217052797585274e-05,
+      "loss": 0.38,
+      "step": 5840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0214234364254236e-05,
+      "loss": 0.3717,
+      "step": 5841
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0211415913897431e-05,
+      "loss": 0.3586,
+      "step": 5842
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020859744673885e-05,
+      "loss": 0.3256,
+      "step": 5843
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205778963002484e-05,
+      "loss": 0.3422,
+      "step": 5844
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0202960462912321e-05,
+      "loss": 0.337,
+      "step": 5845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0200141946692362e-05,
+      "loss": 0.3487,
+      "step": 5846
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197323414566596e-05,
+      "loss": 0.378,
+      "step": 5847
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0194504866759023e-05,
+      "loss": 0.3763,
+      "step": 5848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0191686303493636e-05,
+      "loss": 0.3388,
+      "step": 5849
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0188867724994434e-05,
+      "loss": 0.3612,
+      "step": 5850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0186049131485422e-05,
+      "loss": 0.3436,
+      "step": 5851
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01832305231906e-05,
+      "loss": 0.37,
+      "step": 5852
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0180411900333966e-05,
+      "loss": 0.3221,
+      "step": 5853
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177593263139525e-05,
+      "loss": 0.3724,
+      "step": 5854
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174774611831281e-05,
+      "loss": 0.3719,
+      "step": 5855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0171955946633243e-05,
+      "loss": 0.3819,
+      "step": 5856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0169137267769414e-05,
+      "loss": 0.3554,
+      "step": 5857
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166318575463802e-05,
+      "loss": 0.3758,
+      "step": 5858
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0163499869940419e-05,
+      "loss": 0.3843,
+      "step": 5859
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016068115142327e-05,
+      "loss": 0.3531,
+      "step": 5860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0157862420136371e-05,
+      "loss": 0.3528,
+      "step": 5861
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015504367630373e-05,
+      "loss": 0.3904,
+      "step": 5862
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0152224920149363e-05,
+      "loss": 0.3301,
+      "step": 5863
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0149406151897283e-05,
+      "loss": 0.3535,
+      "step": 5864
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0146587371771503e-05,
+      "loss": 0.3436,
+      "step": 5865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0143768579996044e-05,
+      "loss": 0.4176,
+      "step": 5866
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0140949776794917e-05,
+      "loss": 0.3556,
+      "step": 5867
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0138130962392139e-05,
+      "loss": 0.3511,
+      "step": 5868
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135312137011732e-05,
+      "loss": 0.3237,
+      "step": 5869
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013249330087772e-05,
+      "loss": 0.3452,
+      "step": 5870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0129674454214116e-05,
+      "loss": 0.3407,
+      "step": 5871
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012685559724494e-05,
+      "loss": 0.338,
+      "step": 5872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012403673019422e-05,
+      "loss": 0.3243,
+      "step": 5873
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0121217853285974e-05,
+      "loss": 0.3402,
+      "step": 5874
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011839896674423e-05,
+      "loss": 0.3062,
+      "step": 5875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0115580070793007e-05,
+      "loss": 0.3696,
+      "step": 5876
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0112761165656335e-05,
+      "loss": 0.3702,
+      "step": 5877
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0109942251558238e-05,
+      "loss": 0.3694,
+      "step": 5878
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.010712332872274e-05,
+      "loss": 0.3458,
+      "step": 5879
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0104304397373872e-05,
+      "loss": 0.3518,
+      "step": 5880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0101485457735658e-05,
+      "loss": 0.333,
+      "step": 5881
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0098666510032132e-05,
+      "loss": 0.3702,
+      "step": 5882
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0095847554487316e-05,
+      "loss": 0.4059,
+      "step": 5883
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0093028591325249e-05,
+      "loss": 0.3547,
+      "step": 5884
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009020962076995e-05,
+      "loss": 0.2998,
+      "step": 5885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087390643045459e-05,
+      "loss": 0.3369,
+      "step": 5886
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0084571658375803e-05,
+      "loss": 0.3569,
+      "step": 5887
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0081752666985021e-05,
+      "loss": 0.3612,
+      "step": 5888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0078933669097135e-05,
+      "loss": 0.3415,
+      "step": 5889
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0076114664936185e-05,
+      "loss": 0.3384,
+      "step": 5890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00732956547262e-05,
+      "loss": 0.3305,
+      "step": 5891
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0070476638691222e-05,
+      "loss": 0.3664,
+      "step": 5892
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006765761705528e-05,
+      "loss": 0.3809,
+      "step": 5893
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0064838590042407e-05,
+      "loss": 0.3608,
+      "step": 5894
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062019557876643e-05,
+      "loss": 0.3605,
+      "step": 5895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059200520782023e-05,
+      "loss": 0.3809,
+      "step": 5896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056381478982582e-05,
+      "loss": 0.3591,
+      "step": 5897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0053562432702355e-05,
+      "loss": 0.3706,
+      "step": 5898
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050743382165385e-05,
+      "loss": 0.331,
+      "step": 5899
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0047924327595701e-05,
+      "loss": 0.3657,
+      "step": 5900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045105269217347e-05,
+      "loss": 0.3738,
+      "step": 5901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042286207254361e-05,
+      "loss": 0.3577,
+      "step": 5902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0039467141930776e-05,
+      "loss": 0.3397,
+      "step": 5903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036648073470636e-05,
+      "loss": 0.3951,
+      "step": 5904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033829002097977e-05,
+      "loss": 0.3628,
+      "step": 5905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0031009928036841e-05,
+      "loss": 0.3771,
+      "step": 5906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002819085151126e-05,
+      "loss": 0.3501,
+      "step": 5907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0025371772745283e-05,
+      "loss": 0.3794,
+      "step": 5908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002255269196294e-05,
+      "loss": 0.4042,
+      "step": 5909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0019733609388276e-05,
+      "loss": 0.3359,
+      "step": 5910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0016914525245334e-05,
+      "loss": 0.3421,
+      "step": 5911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0014095439758148e-05,
+      "loss": 0.3799,
+      "step": 5912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0011276353150761e-05,
+      "loss": 0.3243,
+      "step": 5913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0008457265647213e-05,
+      "loss": 0.3474,
+      "step": 5914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0005638177471544e-05,
+      "loss": 0.3488,
+      "step": 5915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0002819088847794e-05,
+      "loss": 0.3398,
+      "step": 5916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.3608,
+      "step": 5917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997180911152208e-06,
+      "loss": 0.3447,
+      "step": 5918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99436182252846e-06,
+      "loss": 0.3555,
+      "step": 5919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99154273435279e-06,
+      "loss": 0.3415,
+      "step": 5920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.988723646849244e-06,
+      "loss": 0.374,
+      "step": 5921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985904560241855e-06,
+      "loss": 0.3691,
+      "step": 5922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983085474754667e-06,
+      "loss": 0.3266,
+      "step": 5923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.980266390611727e-06,
+      "loss": 0.3711,
+      "step": 5924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97744730803706e-06,
+      "loss": 0.347,
+      "step": 5925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.974628227254724e-06,
+      "loss": 0.3254,
+      "step": 5926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.971809148488741e-06,
+      "loss": 0.3937,
+      "step": 5927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.968990071963165e-06,
+      "loss": 0.3518,
+      "step": 5928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.966170997902027e-06,
+      "loss": 0.3655,
+      "step": 5929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963351926529364e-06,
+      "loss": 0.3688,
+      "step": 5930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.960532858069226e-06,
+      "loss": 0.3413,
+      "step": 5931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95771379274564e-06,
+      "loss": 0.3514,
+      "step": 5932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954894730782655e-06,
+      "loss": 0.3493,
+      "step": 5933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9520756724043e-06,
+      "loss": 0.3318,
+      "step": 5934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949256617834617e-06,
+      "loss": 0.3516,
+      "step": 5935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.946437567297646e-06,
+      "loss": 0.3505,
+      "step": 5936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943618521017421e-06,
+      "loss": 0.3706,
+      "step": 5937
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940799479217982e-06,
+      "loss": 0.3342,
+      "step": 5938
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937980442123359e-06,
+      "loss": 0.3549,
+      "step": 5939
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.935161409957598e-06,
+      "loss": 0.3348,
+      "step": 5940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.932342382944723e-06,
+      "loss": 0.3638,
+      "step": 5941
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.92952336130878e-06,
+      "loss": 0.3666,
+      "step": 5942
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.926704345273803e-06,
+      "loss": 0.3301,
+      "step": 5943
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.923885335063817e-06,
+      "loss": 0.3167,
+      "step": 5944
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.92106633090287e-06,
+      "loss": 0.3305,
+      "step": 5945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.918247333014984e-06,
+      "loss": 0.3307,
+      "step": 5946
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.915428341624197e-06,
+      "loss": 0.3639,
+      "step": 5947
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912609356954545e-06,
+      "loss": 0.3811,
+      "step": 5948
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.909790379230051e-06,
+      "loss": 0.3661,
+      "step": 5949
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906971408674756e-06,
+      "loss": 0.3815,
+      "step": 5950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.904152445512685e-06,
+      "loss": 0.3279,
+      "step": 5951
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90133348996787e-06,
+      "loss": 0.3868,
+      "step": 5952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.898514542264345e-06,
+      "loss": 0.332,
+      "step": 5953
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89569560262613e-06,
+      "loss": 0.3726,
+      "step": 5954
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.892876671277262e-06,
+      "loss": 0.3552,
+      "step": 5955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890057748441766e-06,
+      "loss": 0.3903,
+      "step": 5956
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88723883434367e-06,
+      "loss": 0.3682,
+      "step": 5957
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.884419929206995e-06,
+      "loss": 0.3456,
+      "step": 5958
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88160103325577e-06,
+      "loss": 0.3847,
+      "step": 5959
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.878782146714028e-06,
+      "loss": 0.3469,
+      "step": 5960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875963269805783e-06,
+      "loss": 0.3313,
+      "step": 5961
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.873144402755063e-06,
+      "loss": 0.33,
+      "step": 5962
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.870325545785889e-06,
+      "loss": 0.3374,
+      "step": 5963
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.867506699122282e-06,
+      "loss": 0.3856,
+      "step": 5964
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86468786298827e-06,
+      "loss": 0.3595,
+      "step": 5965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.861869037607861e-06,
+      "loss": 0.3784,
+      "step": 5966
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859050223205088e-06,
+      "loss": 0.3259,
+      "step": 5967
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856231420003959e-06,
+      "loss": 0.3469,
+      "step": 5968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8534126282285e-06,
+      "loss": 0.3392,
+      "step": 5969
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85059384810272e-06,
+      "loss": 0.3821,
+      "step": 5970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847775079850637e-06,
+      "loss": 0.3771,
+      "step": 5971
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844956323696273e-06,
+      "loss": 0.3472,
+      "step": 5972
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.842137579863632e-06,
+      "loss": 0.3479,
+      "step": 5973
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839318848576733e-06,
+      "loss": 0.3306,
+      "step": 5974
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.836500130059585e-06,
+      "loss": 0.3702,
+      "step": 5975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8336814245362e-06,
+      "loss": 0.4157,
+      "step": 5976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83086273223059e-06,
+      "loss": 0.3158,
+      "step": 5977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82804405336676e-06,
+      "loss": 0.3808,
+      "step": 5978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825225388168724e-06,
+      "loss": 0.3323,
+      "step": 5979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822406736860479e-06,
+      "loss": 0.3358,
+      "step": 5980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819588099666039e-06,
+      "loss": 0.3679,
+      "step": 5981
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.816769476809404e-06,
+      "loss": 0.3355,
+      "step": 5982
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81395086851458e-06,
+      "loss": 0.3309,
+      "step": 5983
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811132275005568e-06,
+      "loss": 0.3774,
+      "step": 5984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808313696506368e-06,
+      "loss": 0.3806,
+      "step": 5985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805495133240984e-06,
+      "loss": 0.3398,
+      "step": 5986
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802676585433408e-06,
+      "loss": 0.3646,
+      "step": 5987
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79985805330764e-06,
+      "loss": 0.3606,
+      "step": 5988
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79703953708768e-06,
+      "loss": 0.3488,
+      "step": 5989
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79422103699752e-06,
+      "loss": 0.3647,
+      "step": 5990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.791402553261153e-06,
+      "loss": 0.4031,
+      "step": 5991
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.788584086102572e-06,
+      "loss": 0.3884,
+      "step": 5992
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.785765635745765e-06,
+      "loss": 0.3205,
+      "step": 5993
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78294720241473e-06,
+      "loss": 0.3567,
+      "step": 5994
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780128786333443e-06,
+      "loss": 0.3596,
+      "step": 5995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.777310387725904e-06,
+      "loss": 0.378,
+      "step": 5996
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.774492006816088e-06,
+      "loss": 0.3426,
+      "step": 5997
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.771673643827988e-06,
+      "loss": 0.3759,
+      "step": 5998
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.768855298985574e-06,
+      "loss": 0.3515,
+      "step": 5999
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76603697251284e-06,
+      "loss": 0.3508,
+      "step": 6000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763218664633763e-06,
+      "loss": 0.3339,
+      "step": 6001
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.760400375572314e-06,
+      "loss": 0.3461,
+      "step": 6002
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75758210555248e-06,
+      "loss": 0.3502,
+      "step": 6003
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.754763854798227e-06,
+      "loss": 0.3439,
+      "step": 6004
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751945623533536e-06,
+      "loss": 0.4281,
+      "step": 6005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.749127411982379e-06,
+      "loss": 0.3571,
+      "step": 6006
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746309220368718e-06,
+      "loss": 0.3821,
+      "step": 6007
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.743491048916533e-06,
+      "loss": 0.3928,
+      "step": 6008
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.740672897849784e-06,
+      "loss": 0.3219,
+      "step": 6009
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.737854767392443e-06,
+      "loss": 0.3583,
+      "step": 6010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.735036657768468e-06,
+      "loss": 0.3483,
+      "step": 6011
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.732218569201821e-06,
+      "loss": 0.3268,
+      "step": 6012
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.729400501916471e-06,
+      "loss": 0.3269,
+      "step": 6013
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.726582456136372e-06,
+      "loss": 0.3485,
+      "step": 6014
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.723764432085482e-06,
+      "loss": 0.3466,
+      "step": 6015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.720946429987752e-06,
+      "loss": 0.3728,
+      "step": 6016
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.718128450067142e-06,
+      "loss": 0.3647,
+      "step": 6017
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715310492547608e-06,
+      "loss": 0.343,
+      "step": 6018
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.712492557653091e-06,
+      "loss": 0.3754,
+      "step": 6019
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.709674645607547e-06,
+      "loss": 0.3515,
+      "step": 6020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.706856756634914e-06,
+      "loss": 0.3638,
+      "step": 6021
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.704038890959147e-06,
+      "loss": 0.3889,
+      "step": 6022
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.701221048804185e-06,
+      "loss": 0.3425,
+      "step": 6023
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69840323039397e-06,
+      "loss": 0.3566,
+      "step": 6024
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69558543595244e-06,
+      "loss": 0.3568,
+      "step": 6025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69276766570353e-06,
+      "loss": 0.3472,
+      "step": 6026
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.689949919871185e-06,
+      "loss": 0.3726,
+      "step": 6027
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.687132198679328e-06,
+      "loss": 0.3627,
+      "step": 6028
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684314502351895e-06,
+      "loss": 0.3437,
+      "step": 6029
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.681496831112818e-06,
+      "loss": 0.3875,
+      "step": 6030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.67867918518602e-06,
+      "loss": 0.3242,
+      "step": 6031
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.675861564795433e-06,
+      "loss": 0.3916,
+      "step": 6032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.673043970164976e-06,
+      "loss": 0.3788,
+      "step": 6033
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.670226401518568e-06,
+      "loss": 0.3568,
+      "step": 6034
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.667408859080138e-06,
+      "loss": 0.3224,
+      "step": 6035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664591343073593e-06,
+      "loss": 0.361,
+      "step": 6036
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.661773853722855e-06,
+      "loss": 0.358,
+      "step": 6037
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658956391251835e-06,
+      "loss": 0.3327,
+      "step": 6038
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656138955884446e-06,
+      "loss": 0.3863,
+      "step": 6039
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.653321547844592e-06,
+      "loss": 0.3801,
+      "step": 6040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.650504167356185e-06,
+      "loss": 0.3769,
+      "step": 6041
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64768681464313e-06,
+      "loss": 0.4092,
+      "step": 6042
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644869489929322e-06,
+      "loss": 0.3421,
+      "step": 6043
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.642052193438672e-06,
+      "loss": 0.3339,
+      "step": 6044
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.639234925395066e-06,
+      "loss": 0.4047,
+      "step": 6045
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63641768602241e-06,
+      "loss": 0.3564,
+      "step": 6046
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633600475544594e-06,
+      "loss": 0.4274,
+      "step": 6047
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.630783294185506e-06,
+      "loss": 0.3524,
+      "step": 6048
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.627966142169041e-06,
+      "loss": 0.343,
+      "step": 6049
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62514901971908e-06,
+      "loss": 0.3881,
+      "step": 6050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62233192705951e-06,
+      "loss": 0.3204,
+      "step": 6051
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619514864414207e-06,
+      "loss": 0.3387,
+      "step": 6052
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.616697832007056e-06,
+      "loss": 0.3358,
+      "step": 6053
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.613880830061937e-06,
+      "loss": 0.4004,
+      "step": 6054
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611063858802718e-06,
+      "loss": 0.4101,
+      "step": 6055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.608246918453276e-06,
+      "loss": 0.3193,
+      "step": 6056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.605430009237474e-06,
+      "loss": 0.3653,
+      "step": 6057
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.602613131379183e-06,
+      "loss": 0.3425,
+      "step": 6058
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599796285102272e-06,
+      "loss": 0.3383,
+      "step": 6059
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596979470630597e-06,
+      "loss": 0.3777,
+      "step": 6060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594162688188022e-06,
+      "loss": 0.3615,
+      "step": 6061
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591345937998398e-06,
+      "loss": 0.3618,
+      "step": 6062
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588529220285586e-06,
+      "loss": 0.3532,
+      "step": 6063
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.585712535273435e-06,
+      "loss": 0.3888,
+      "step": 6064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582895883185791e-06,
+      "loss": 0.3557,
+      "step": 6065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58007926424651e-06,
+      "loss": 0.3311,
+      "step": 6066
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.577262678679426e-06,
+      "loss": 0.369,
+      "step": 6067
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.574446126708389e-06,
+      "loss": 0.4088,
+      "step": 6068
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57162960855723e-06,
+      "loss": 0.3659,
+      "step": 6069
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56881312444979e-06,
+      "loss": 0.3566,
+      "step": 6070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.565996674609903e-06,
+      "loss": 0.3791,
+      "step": 6071
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563180259261393e-06,
+      "loss": 0.3746,
+      "step": 6072
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.560363878628097e-06,
+      "loss": 0.3503,
+      "step": 6073
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557547532933835e-06,
+      "loss": 0.4131,
+      "step": 6074
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.554731222402427e-06,
+      "loss": 0.3617,
+      "step": 6075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551914947257701e-06,
+      "loss": 0.3557,
+      "step": 6076
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549098707723466e-06,
+      "loss": 0.3288,
+      "step": 6077
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54628250402354e-06,
+      "loss": 0.3482,
+      "step": 6078
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.543466336381728e-06,
+      "loss": 0.3531,
+      "step": 6079
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.540650205021848e-06,
+      "loss": 0.3615,
+      "step": 6080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537834110167697e-06,
+      "loss": 0.3691,
+      "step": 6081
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53501805204308e-06,
+      "loss": 0.3714,
+      "step": 6082
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.532202030871802e-06,
+      "loss": 0.3591,
+      "step": 6083
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.529386046877648e-06,
+      "loss": 0.3171,
+      "step": 6084
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526570100284423e-06,
+      "loss": 0.3758,
+      "step": 6085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523754191315907e-06,
+      "loss": 0.359,
+      "step": 6086
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520938320195899e-06,
+      "loss": 0.3428,
+      "step": 6087
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51812248714818e-06,
+      "loss": 0.3629,
+      "step": 6088
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515306692396524e-06,
+      "loss": 0.3811,
+      "step": 6089
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.512490936164722e-06,
+      "loss": 0.3494,
+      "step": 6090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509675218676539e-06,
+      "loss": 0.3404,
+      "step": 6091
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50685954015575e-06,
+      "loss": 0.3182,
+      "step": 6092
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50404390082613e-06,
+      "loss": 0.3391,
+      "step": 6093
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501228300911436e-06,
+      "loss": 0.3396,
+      "step": 6094
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.498412740635441e-06,
+      "loss": 0.4179,
+      "step": 6095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495597220221898e-06,
+      "loss": 0.3669,
+      "step": 6096
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492781739894568e-06,
+      "loss": 0.3557,
+      "step": 6097
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489966299877198e-06,
+      "loss": 0.3514,
+      "step": 6098
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487150900393546e-06,
+      "loss": 0.3266,
+      "step": 6099
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.484335541667357e-06,
+      "loss": 0.3574,
+      "step": 6100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48152022392237e-06,
+      "loss": 0.334,
+      "step": 6101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.478704947382335e-06,
+      "loss": 0.3411,
+      "step": 6102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47588971227098e-06,
+      "loss": 0.3529,
+      "step": 6103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.473074518812044e-06,
+      "loss": 0.4066,
+      "step": 6104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.470259367229258e-06,
+      "loss": 0.338,
+      "step": 6105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.467444257746346e-06,
+      "loss": 0.3439,
+      "step": 6106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46462919058704e-06,
+      "loss": 0.3758,
+      "step": 6107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46181416597505e-06,
+      "loss": 0.3366,
+      "step": 6108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458999184134101e-06,
+      "loss": 0.355,
+      "step": 6109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.456184245287904e-06,
+      "loss": 0.3855,
+      "step": 6110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45336934966017e-06,
+      "loss": 0.3292,
+      "step": 6111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.450554497474608e-06,
+      "loss": 0.3482,
+      "step": 6112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44773968895492e-06,
+      "loss": 0.3843,
+      "step": 6113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44492492432481e-06,
+      "loss": 0.3585,
+      "step": 6114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442110203807963e-06,
+      "loss": 0.369,
+      "step": 6115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439295527628083e-06,
+      "loss": 0.3233,
+      "step": 6116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436480896008859e-06,
+      "loss": 0.3276,
+      "step": 6117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.433666309173973e-06,
+      "loss": 0.3385,
+      "step": 6118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430851767347114e-06,
+      "loss": 0.3294,
+      "step": 6119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42803727075195e-06,
+      "loss": 0.3394,
+      "step": 6120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425222819612166e-06,
+      "loss": 0.3238,
+      "step": 6121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.422408414151433e-06,
+      "loss": 0.3547,
+      "step": 6122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.419594054593415e-06,
+      "loss": 0.372,
+      "step": 6123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.416779741161783e-06,
+      "loss": 0.3274,
+      "step": 6124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413965474080186e-06,
+      "loss": 0.3722,
+      "step": 6125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.411151253572296e-06,
+      "loss": 0.38,
+      "step": 6126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408337079861755e-06,
+      "loss": 0.3671,
+      "step": 6127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.405522953172215e-06,
+      "loss": 0.3348,
+      "step": 6128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40270887372733e-06,
+      "loss": 0.3627,
+      "step": 6129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.399894841750731e-06,
+      "loss": 0.3708,
+      "step": 6130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397080857466066e-06,
+      "loss": 0.3694,
+      "step": 6131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.394266921096962e-06,
+      "loss": 0.302,
+      "step": 6132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391453032867053e-06,
+      "loss": 0.3424,
+      "step": 6133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388639192999969e-06,
+      "loss": 0.3483,
+      "step": 6134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.385825401719328e-06,
+      "loss": 0.3923,
+      "step": 6135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383011659248756e-06,
+      "loss": 0.3693,
+      "step": 6136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.380197965811863e-06,
+      "loss": 0.3903,
+      "step": 6137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377384321632263e-06,
+      "loss": 0.3455,
+      "step": 6138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.374570726933558e-06,
+      "loss": 0.385,
+      "step": 6139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37175718193936e-06,
+      "loss": 0.3431,
+      "step": 6140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.368943686873266e-06,
+      "loss": 0.3417,
+      "step": 6141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.366130241958868e-06,
+      "loss": 0.3744,
+      "step": 6142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.363316847419766e-06,
+      "loss": 0.3508,
+      "step": 6143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.360503503479536e-06,
+      "loss": 0.4026,
+      "step": 6144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357690210361772e-06,
+      "loss": 0.3229,
+      "step": 6145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.354876968290054e-06,
+      "loss": 0.3588,
+      "step": 6146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.35206377748795e-06,
+      "loss": 0.3571,
+      "step": 6147
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.349250638179038e-06,
+      "loss": 0.3673,
+      "step": 6148
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.346437550586881e-06,
+      "loss": 0.3367,
+      "step": 6149
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.343624514935046e-06,
+      "loss": 0.3539,
+      "step": 6150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.340811531447091e-06,
+      "loss": 0.383,
+      "step": 6151
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.337998600346569e-06,
+      "loss": 0.3555,
+      "step": 6152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.335185721857037e-06,
+      "loss": 0.3231,
+      "step": 6153
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.332372896202038e-06,
+      "loss": 0.3452,
+      "step": 6154
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.329560123605114e-06,
+      "loss": 0.3027,
+      "step": 6155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326747404289802e-06,
+      "loss": 0.3507,
+      "step": 6156
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.323934738479636e-06,
+      "loss": 0.3626,
+      "step": 6157
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321122126398154e-06,
+      "loss": 0.316,
+      "step": 6158
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.318309568268875e-06,
+      "loss": 0.3744,
+      "step": 6159
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.315497064315323e-06,
+      "loss": 0.3777,
+      "step": 6160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.312684614761007e-06,
+      "loss": 0.3746,
+      "step": 6161
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.309872219829448e-06,
+      "loss": 0.3792,
+      "step": 6162
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.307059879744157e-06,
+      "loss": 0.3888,
+      "step": 6163
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.304247594728627e-06,
+      "loss": 0.3382,
+      "step": 6164
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.301435365006371e-06,
+      "loss": 0.3414,
+      "step": 6165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298623190800872e-06,
+      "loss": 0.3423,
+      "step": 6166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29581107233563e-06,
+      "loss": 0.3577,
+      "step": 6167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.292999009834127e-06,
+      "loss": 0.3791,
+      "step": 6168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.290187003519841e-06,
+      "loss": 0.3318,
+      "step": 6169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.287375053616259e-06,
+      "loss": 0.3918,
+      "step": 6170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.284563160346847e-06,
+      "loss": 0.3266,
+      "step": 6171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281751323935078e-06,
+      "loss": 0.348,
+      "step": 6172
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.278939544604413e-06,
+      "loss": 0.3468,
+      "step": 6173
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27612782257831e-06,
+      "loss": 0.3826,
+      "step": 6174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.273316158080228e-06,
+      "loss": 0.4052,
+      "step": 6175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.270504551333616e-06,
+      "loss": 0.3483,
+      "step": 6176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.267693002561921e-06,
+      "loss": 0.3568,
+      "step": 6177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.264881511988577e-06,
+      "loss": 0.3516,
+      "step": 6178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26207007983703e-06,
+      "loss": 0.3353,
+      "step": 6179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259258706330705e-06,
+      "loss": 0.3345,
+      "step": 6180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.256447391693033e-06,
+      "loss": 0.3487,
+      "step": 6181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253636136147438e-06,
+      "loss": 0.3336,
+      "step": 6182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25082493991733e-06,
+      "loss": 0.3518,
+      "step": 6183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.248013803226135e-06,
+      "loss": 0.3531,
+      "step": 6184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.245202726297248e-06,
+      "loss": 0.3761,
+      "step": 6185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24239170935408e-06,
+      "loss": 0.3635,
+      "step": 6186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239580752620032e-06,
+      "loss": 0.3071,
+      "step": 6187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.236769856318488e-06,
+      "loss": 0.3357,
+      "step": 6188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23395902067285e-06,
+      "loss": 0.3988,
+      "step": 6189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231148245906494e-06,
+      "loss": 0.3184,
+      "step": 6190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.228337532242799e-06,
+      "loss": 0.3375,
+      "step": 6191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225526879905149e-06,
+      "loss": 0.3839,
+      "step": 6192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.222716289116901e-06,
+      "loss": 0.3202,
+      "step": 6193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219905760101432e-06,
+      "loss": 0.3648,
+      "step": 6194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217095293082095e-06,
+      "loss": 0.351,
+      "step": 6195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214284888282248e-06,
+      "loss": 0.3243,
+      "step": 6196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.211474545925237e-06,
+      "loss": 0.3984,
+      "step": 6197
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.208664266234412e-06,
+      "loss": 0.3672,
+      "step": 6198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.205854049433112e-06,
+      "loss": 0.3701,
+      "step": 6199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203043895744673e-06,
+      "loss": 0.3527,
+      "step": 6200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.200233805392424e-06,
+      "loss": 0.3681,
+      "step": 6201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.197423778599687e-06,
+      "loss": 0.3757,
+      "step": 6202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19461381558979e-06,
+      "loss": 0.3809,
+      "step": 6203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.191803916586045e-06,
+      "loss": 0.3657,
+      "step": 6204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.188994081811756e-06,
+      "loss": 0.3742,
+      "step": 6205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.186184311490239e-06,
+      "loss": 0.3688,
+      "step": 6206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18337460584478e-06,
+      "loss": 0.3502,
+      "step": 6207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18056496509869e-06,
+      "loss": 0.3508,
+      "step": 6208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.177755389475245e-06,
+      "loss": 0.3793,
+      "step": 6209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.174945879197731e-06,
+      "loss": 0.3691,
+      "step": 6210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.172136434489436e-06,
+      "loss": 0.3546,
+      "step": 6211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.169327055573628e-06,
+      "loss": 0.341,
+      "step": 6212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166517742673575e-06,
+      "loss": 0.361,
+      "step": 6213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16370849601254e-06,
+      "loss": 0.3729,
+      "step": 6214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.160899315813781e-06,
+      "loss": 0.357,
+      "step": 6215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.158090202300557e-06,
+      "loss": 0.3373,
+      "step": 6216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15528115569611e-06,
+      "loss": 0.3564,
+      "step": 6217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152472176223684e-06,
+      "loss": 0.3405,
+      "step": 6218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.149663264106511e-06,
+      "loss": 0.3638,
+      "step": 6219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.146854419567834e-06,
+      "loss": 0.3346,
+      "step": 6220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.144045642830865e-06,
+      "loss": 0.3256,
+      "step": 6221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.141236934118837e-06,
+      "loss": 0.3331,
+      "step": 6222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.138428293654961e-06,
+      "loss": 0.3441,
+      "step": 6223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.135619721662443e-06,
+      "loss": 0.3653,
+      "step": 6224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132811218364494e-06,
+      "loss": 0.3745,
+      "step": 6225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13000278398431e-06,
+      "loss": 0.3436,
+      "step": 6226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127194418745083e-06,
+      "loss": 0.398,
+      "step": 6227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.124386122870007e-06,
+      "loss": 0.3958,
+      "step": 6228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121577896582255e-06,
+      "loss": 0.3403,
+      "step": 6229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118769740105014e-06,
+      "loss": 0.3785,
+      "step": 6230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11596165366145e-06,
+      "loss": 0.3604,
+      "step": 6231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.113153637474729e-06,
+      "loss": 0.3194,
+      "step": 6232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.110345691768016e-06,
+      "loss": 0.3567,
+      "step": 6233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107537816764459e-06,
+      "loss": 0.3638,
+      "step": 6234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104730012687213e-06,
+      "loss": 0.3397,
+      "step": 6235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10192227975942e-06,
+      "loss": 0.3403,
+      "step": 6236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.099114618204217e-06,
+      "loss": 0.3767,
+      "step": 6237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096307028244734e-06,
+      "loss": 0.3815,
+      "step": 6238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.093499510104102e-06,
+      "loss": 0.3966,
+      "step": 6239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090692064005442e-06,
+      "loss": 0.352,
+      "step": 6240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08788469017186e-06,
+      "loss": 0.3472,
+      "step": 6241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.085077388826479e-06,
+      "loss": 0.3457,
+      "step": 6242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.082270160192391e-06,
+      "loss": 0.3513,
+      "step": 6243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0794630044927e-06,
+      "loss": 0.3959,
+      "step": 6244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076655921950502e-06,
+      "loss": 0.3224,
+      "step": 6245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07384891278887e-06,
+      "loss": 0.3629,
+      "step": 6246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071041977230897e-06,
+      "loss": 0.3341,
+      "step": 6247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06823511549965e-06,
+      "loss": 0.3512,
+      "step": 6248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065428327818205e-06,
+      "loss": 0.3511,
+      "step": 6249
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.062621614409615e-06,
+      "loss": 0.3905,
+      "step": 6250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.059814975496943e-06,
+      "loss": 0.3222,
+      "step": 6251
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.057008411303241e-06,
+      "loss": 0.3295,
+      "step": 6252
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.054201922051552e-06,
+      "loss": 0.348,
+      "step": 6253
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.051395507964916e-06,
+      "loss": 0.3199,
+      "step": 6254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.048589169266361e-06,
+      "loss": 0.3294,
+      "step": 6255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04578290617892e-06,
+      "loss": 0.3551,
+      "step": 6256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.042976718925615e-06,
+      "loss": 0.3432,
+      "step": 6257
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.040170607729456e-06,
+      "loss": 0.3743,
+      "step": 6258
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.037364572813456e-06,
+      "loss": 0.3265,
+      "step": 6259
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.034558614400613e-06,
+      "loss": 0.3623,
+      "step": 6260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.031752732713929e-06,
+      "loss": 0.3459,
+      "step": 6261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.028946927976395e-06,
+      "loss": 0.3682,
+      "step": 6262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02614120041099e-06,
+      "loss": 0.3375,
+      "step": 6263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.023335550240701e-06,
+      "loss": 0.3353,
+      "step": 6264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020529977688491e-06,
+      "loss": 0.3429,
+      "step": 6265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.017724482977335e-06,
+      "loss": 0.336,
+      "step": 6266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.014919066330187e-06,
+      "loss": 0.343,
+      "step": 6267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01211372797e-06,
+      "loss": 0.3864,
+      "step": 6268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.009308468119728e-06,
+      "loss": 0.3525,
+      "step": 6269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006503287002304e-06,
+      "loss": 0.3243,
+      "step": 6270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.003698184840674e-06,
+      "loss": 0.3614,
+      "step": 6271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000893161857755e-06,
+      "loss": 0.3739,
+      "step": 6272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.998088218276475e-06,
+      "loss": 0.3167,
+      "step": 6273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995283354319755e-06,
+      "loss": 0.349,
+      "step": 6274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.992478570210497e-06,
+      "loss": 0.3594,
+      "step": 6275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98967386617161e-06,
+      "loss": 0.3713,
+      "step": 6276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.986869242425983e-06,
+      "loss": 0.3552,
+      "step": 6277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98406469919652e-06,
+      "loss": 0.3655,
+      "step": 6278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98126023670609e-06,
+      "loss": 0.3587,
+      "step": 6279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978455855177583e-06,
+      "loss": 0.3798,
+      "step": 6280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.975651554833869e-06,
+      "loss": 0.3569,
+      "step": 6281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.972847335897807e-06,
+      "loss": 0.3838,
+      "step": 6282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970043198592263e-06,
+      "loss": 0.3895,
+      "step": 6283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96723914314008e-06,
+      "loss": 0.3477,
+      "step": 6284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.964435169764113e-06,
+      "loss": 0.3517,
+      "step": 6285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961631278687198e-06,
+      "loss": 0.3361,
+      "step": 6286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.958827470132163e-06,
+      "loss": 0.357,
+      "step": 6287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.956023744321842e-06,
+      "loss": 0.392,
+      "step": 6288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.953220101479047e-06,
+      "loss": 0.3889,
+      "step": 6289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.950416541826599e-06,
+      "loss": 0.3616,
+      "step": 6290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.947613065587292e-06,
+      "loss": 0.4012,
+      "step": 6291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944809672983934e-06,
+      "loss": 0.3458,
+      "step": 6292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.942006364239322e-06,
+      "loss": 0.3741,
+      "step": 6293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939203139576234e-06,
+      "loss": 0.3505,
+      "step": 6294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.936399999217455e-06,
+      "loss": 0.3177,
+      "step": 6295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93359694338575e-06,
+      "loss": 0.3532,
+      "step": 6296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.930793972303892e-06,
+      "loss": 0.3131,
+      "step": 6297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.927991086194646e-06,
+      "loss": 0.3518,
+      "step": 6298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.925188285280752e-06,
+      "loss": 0.3666,
+      "step": 6299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922385569784966e-06,
+      "loss": 0.3493,
+      "step": 6300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.919582939930018e-06,
+      "loss": 0.3643,
+      "step": 6301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916780395938647e-06,
+      "loss": 0.3464,
+      "step": 6302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913977938033581e-06,
+      "loss": 0.3408,
+      "step": 6303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91117556643753e-06,
+      "loss": 0.3456,
+      "step": 6304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908373281373214e-06,
+      "loss": 0.3344,
+      "step": 6305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90557108306333e-06,
+      "loss": 0.3237,
+      "step": 6306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902768971730586e-06,
+      "loss": 0.3314,
+      "step": 6307
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.899966947597664e-06,
+      "loss": 0.358,
+      "step": 6308
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89716501088725e-06,
+      "loss": 0.3502,
+      "step": 6309
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.894363161822027e-06,
+      "loss": 0.353,
+      "step": 6310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891561400624657e-06,
+      "loss": 0.3905,
+      "step": 6311
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.888759727517813e-06,
+      "loss": 0.3414,
+      "step": 6312
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.885958142724143e-06,
+      "loss": 0.3164,
+      "step": 6313
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883156646466295e-06,
+      "loss": 0.3468,
+      "step": 6314
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880355238966923e-06,
+      "loss": 0.3316,
+      "step": 6315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87755392044865e-06,
+      "loss": 0.3579,
+      "step": 6316
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.874752691134113e-06,
+      "loss": 0.3703,
+      "step": 6317
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.871951551245923e-06,
+      "loss": 0.4091,
+      "step": 6318
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869150501006705e-06,
+      "loss": 0.4133,
+      "step": 6319
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.866349540639056e-06,
+      "loss": 0.3637,
+      "step": 6320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.863548670365583e-06,
+      "loss": 0.3735,
+      "step": 6321
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.860747890408876e-06,
+      "loss": 0.3313,
+      "step": 6322
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.857947200991517e-06,
+      "loss": 0.3719,
+      "step": 6323
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.855146602336091e-06,
+      "loss": 0.3533,
+      "step": 6324
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852346094665164e-06,
+      "loss": 0.3515,
+      "step": 6325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.849545678201297e-06,
+      "loss": 0.3219,
+      "step": 6326
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.846745353167054e-06,
+      "loss": 0.3373,
+      "step": 6327
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.843945119784978e-06,
+      "loss": 0.3734,
+      "step": 6328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.841144978277615e-06,
+      "loss": 0.3696,
+      "step": 6329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.838344928867499e-06,
+      "loss": 0.3033,
+      "step": 6330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.835544971777152e-06,
+      "loss": 0.3567,
+      "step": 6331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832745107229103e-06,
+      "loss": 0.4265,
+      "step": 6332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829945335445855e-06,
+      "loss": 0.3516,
+      "step": 6333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.827145656649921e-06,
+      "loss": 0.3909,
+      "step": 6334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.824346071063795e-06,
+      "loss": 0.376,
+      "step": 6335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82154657890997e-06,
+      "loss": 0.3476,
+      "step": 6336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81874718041092e-06,
+      "loss": 0.3678,
+      "step": 6337
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.815947875789133e-06,
+      "loss": 0.368,
+      "step": 6338
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.813148665267073e-06,
+      "loss": 0.3729,
+      "step": 6339
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810349549067194e-06,
+      "loss": 0.3728,
+      "step": 6340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80755052741196e-06,
+      "loss": 0.3296,
+      "step": 6341
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804751600523805e-06,
+      "loss": 0.3251,
+      "step": 6342
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801952768625176e-06,
+      "loss": 0.3559,
+      "step": 6343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799154031938503e-06,
+      "loss": 0.3379,
+      "step": 6344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796355390686201e-06,
+      "loss": 0.3364,
+      "step": 6345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.793556845090695e-06,
+      "loss": 0.3436,
+      "step": 6346
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.790758395374384e-06,
+      "loss": 0.344,
+      "step": 6347
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787960041759678e-06,
+      "loss": 0.3468,
+      "step": 6348
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.785161784468961e-06,
+      "loss": 0.3854,
+      "step": 6349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78236362372462e-06,
+      "loss": 0.35,
+      "step": 6350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.779565559749037e-06,
+      "loss": 0.3614,
+      "step": 6351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.776767592764575e-06,
+      "loss": 0.3895,
+      "step": 6352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773969722993601e-06,
+      "loss": 0.3765,
+      "step": 6353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77117195065846e-06,
+      "loss": 0.325,
+      "step": 6354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.768374275981508e-06,
+      "loss": 0.367,
+      "step": 6355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765576699185083e-06,
+      "loss": 0.3237,
+      "step": 6356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76277922049151e-06,
+      "loss": 0.3505,
+      "step": 6357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.759981840123119e-06,
+      "loss": 0.3524,
+      "step": 6358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.757184558302215e-06,
+      "loss": 0.3952,
+      "step": 6359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.754387375251113e-06,
+      "loss": 0.3449,
+      "step": 6360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.751590291192115e-06,
+      "loss": 0.3744,
+      "step": 6361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.748793306347505e-06,
+      "loss": 0.4083,
+      "step": 6362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745996420939574e-06,
+      "loss": 0.3425,
+      "step": 6363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.743199635190588e-06,
+      "loss": 0.359,
+      "step": 6364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.740402949322827e-06,
+      "loss": 0.3515,
+      "step": 6365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.737606363558542e-06,
+      "loss": 0.3457,
+      "step": 6366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.734809878119986e-06,
+      "loss": 0.4089,
+      "step": 6367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.732013493229409e-06,
+      "loss": 0.3288,
+      "step": 6368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.729217209109037e-06,
+      "loss": 0.3674,
+      "step": 6369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.726421025981112e-06,
+      "loss": 0.361,
+      "step": 6370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.723624944067842e-06,
+      "loss": 0.3521,
+      "step": 6371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72082896359144e-06,
+      "loss": 0.303,
+      "step": 6372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718033084774117e-06,
+      "loss": 0.3721,
+      "step": 6373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715237307838062e-06,
+      "loss": 0.3568,
+      "step": 6374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712441633005467e-06,
+      "loss": 0.3531,
+      "step": 6375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.70964606049851e-06,
+      "loss": 0.3463,
+      "step": 6376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706850590539364e-06,
+      "loss": 0.3995,
+      "step": 6377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.704055223350186e-06,
+      "loss": 0.3524,
+      "step": 6378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.701259959153139e-06,
+      "loss": 0.3407,
+      "step": 6379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698464798170369e-06,
+      "loss": 0.3761,
+      "step": 6380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.695669740624009e-06,
+      "loss": 0.36,
+      "step": 6381
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692874786736197e-06,
+      "loss": 0.3832,
+      "step": 6382
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.690079936729046e-06,
+      "loss": 0.3446,
+      "step": 6383
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.687285190824678e-06,
+      "loss": 0.3306,
+      "step": 6384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6844905492452e-06,
+      "loss": 0.3798,
+      "step": 6385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6816960122127e-06,
+      "loss": 0.3325,
+      "step": 6386
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.678901579949278e-06,
+      "loss": 0.3819,
+      "step": 6387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676107252677008e-06,
+      "loss": 0.3698,
+      "step": 6388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.673313030617965e-06,
+      "loss": 0.3458,
+      "step": 6389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67051891399421e-06,
+      "loss": 0.3375,
+      "step": 6390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.667724903027802e-06,
+      "loss": 0.3186,
+      "step": 6391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66493099794079e-06,
+      "loss": 0.3109,
+      "step": 6392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.662137198955211e-06,
+      "loss": 0.3081,
+      "step": 6393
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659343506293095e-06,
+      "loss": 0.3584,
+      "step": 6394
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656549920176461e-06,
+      "loss": 0.3552,
+      "step": 6395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.653756440827326e-06,
+      "loss": 0.3591,
+      "step": 6396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650963068467699e-06,
+      "loss": 0.34,
+      "step": 6397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64816980331957e-06,
+      "loss": 0.3448,
+      "step": 6398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645376645604932e-06,
+      "loss": 0.3197,
+      "step": 6399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64258359554576e-06,
+      "loss": 0.3753,
+      "step": 6400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.639790653364027e-06,
+      "loss": 0.3181,
+      "step": 6401
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.636997819281699e-06,
+      "loss": 0.3694,
+      "step": 6402
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634205093520723e-06,
+      "loss": 0.3908,
+      "step": 6403
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63141247630305e-06,
+      "loss": 0.3227,
+      "step": 6404
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62861996785061e-06,
+      "loss": 0.3774,
+      "step": 6405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.625827568385342e-06,
+      "loss": 0.347,
+      "step": 6406
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623035278129156e-06,
+      "loss": 0.3363,
+      "step": 6407
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620243097303962e-06,
+      "loss": 0.3765,
+      "step": 6408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61745102613167e-06,
+      "loss": 0.367,
+      "step": 6409
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.614659064834165e-06,
+      "loss": 0.3252,
+      "step": 6410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.611867213633338e-06,
+      "loss": 0.3353,
+      "step": 6411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.609075472751062e-06,
+      "loss": 0.3432,
+      "step": 6412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.606283842409199e-06,
+      "loss": 0.3735,
+      "step": 6413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.603492322829616e-06,
+      "loss": 0.348,
+      "step": 6414
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600700914234158e-06,
+      "loss": 0.3567,
+      "step": 6415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597909616844667e-06,
+      "loss": 0.3504,
+      "step": 6416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59511843088297e-06,
+      "loss": 0.4194,
+      "step": 6417
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.592327356570896e-06,
+      "loss": 0.3227,
+      "step": 6418
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589536394130253e-06,
+      "loss": 0.3186,
+      "step": 6419
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58674554378285e-06,
+      "loss": 0.3541,
+      "step": 6420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583954805750488e-06,
+      "loss": 0.3226,
+      "step": 6421
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.581164180254942e-06,
+      "loss": 0.389,
+      "step": 6422
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.578373667518002e-06,
+      "loss": 0.3435,
+      "step": 6423
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.575583267761428e-06,
+      "loss": 0.3386,
+      "step": 6424
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.572792981206988e-06,
+      "loss": 0.3221,
+      "step": 6425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570002808076431e-06,
+      "loss": 0.3388,
+      "step": 6426
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567212748591496e-06,
+      "loss": 0.3268,
+      "step": 6427
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.564422802973923e-06,
+      "loss": 0.3339,
+      "step": 6428
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56163297144543e-06,
+      "loss": 0.3412,
+      "step": 6429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558843254227733e-06,
+      "loss": 0.3125,
+      "step": 6430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556053651542543e-06,
+      "loss": 0.3366,
+      "step": 6431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55326416361155e-06,
+      "loss": 0.3356,
+      "step": 6432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55047479065645e-06,
+      "loss": 0.3586,
+      "step": 6433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547685532898915e-06,
+      "loss": 0.3454,
+      "step": 6434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54489639056062e-06,
+      "loss": 0.3754,
+      "step": 6435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.542107363863216e-06,
+      "loss": 0.357,
+      "step": 6436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539318453028366e-06,
+      "loss": 0.3343,
+      "step": 6437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.536529658277704e-06,
+      "loss": 0.3526,
+      "step": 6438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533740979832867e-06,
+      "loss": 0.3306,
+      "step": 6439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53095241791548e-06,
+      "loss": 0.3565,
+      "step": 6440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52816397274715e-06,
+      "loss": 0.3322,
+      "step": 6441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.525375644549489e-06,
+      "loss": 0.3538,
+      "step": 6442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522587433544092e-06,
+      "loss": 0.3213,
+      "step": 6443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.51979933995254e-06,
+      "loss": 0.3545,
+      "step": 6444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.517011363996417e-06,
+      "loss": 0.3503,
+      "step": 6445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.514223505897284e-06,
+      "loss": 0.3398,
+      "step": 6446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511435765876708e-06,
+      "loss": 0.4071,
+      "step": 6447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50864814415623e-06,
+      "loss": 0.3314,
+      "step": 6448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50586064095739e-06,
+      "loss": 0.3366,
+      "step": 6449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.503073256501727e-06,
+      "loss": 0.3716,
+      "step": 6450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.500285991010752e-06,
+      "loss": 0.3338,
+      "step": 6451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497498844705982e-06,
+      "loss": 0.3454,
+      "step": 6452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.494711817808913e-06,
+      "loss": 0.3429,
+      "step": 6453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491924910541041e-06,
+      "loss": 0.3555,
+      "step": 6454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489138123123854e-06,
+      "loss": 0.3487,
+      "step": 6455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486351455778817e-06,
+      "loss": 0.3096,
+      "step": 6456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.483564908727398e-06,
+      "loss": 0.3213,
+      "step": 6457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480778482191047e-06,
+      "loss": 0.3536,
+      "step": 6458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.477992176391217e-06,
+      "loss": 0.3474,
+      "step": 6459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.475205991549331e-06,
+      "loss": 0.3465,
+      "step": 6460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472419927886825e-06,
+      "loss": 0.3179,
+      "step": 6461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.469633985625115e-06,
+      "loss": 0.3499,
+      "step": 6462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.466848164985595e-06,
+      "loss": 0.3364,
+      "step": 6463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.464062466189677e-06,
+      "loss": 0.3803,
+      "step": 6464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.461276889458736e-06,
+      "loss": 0.3347,
+      "step": 6465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.458491435014152e-06,
+      "loss": 0.3432,
+      "step": 6466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.455706103077299e-06,
+      "loss": 0.3309,
+      "step": 6467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.452920893869524e-06,
+      "loss": 0.35,
+      "step": 6468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.450135807612184e-06,
+      "loss": 0.3626,
+      "step": 6469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.447350844526612e-06,
+      "loss": 0.3438,
+      "step": 6470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.444566004834136e-06,
+      "loss": 0.3272,
+      "step": 6471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44178128875608e-06,
+      "loss": 0.3147,
+      "step": 6472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438996696513745e-06,
+      "loss": 0.359,
+      "step": 6473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.436212228328438e-06,
+      "loss": 0.3265,
+      "step": 6474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43342788442144e-06,
+      "loss": 0.3433,
+      "step": 6475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430643665014038e-06,
+      "loss": 0.3658,
+      "step": 6476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.427859570327494e-06,
+      "loss": 0.3417,
+      "step": 6477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42507560058307e-06,
+      "loss": 0.3782,
+      "step": 6478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42229175600202e-06,
+      "loss": 0.337,
+      "step": 6479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.419508036805572e-06,
+      "loss": 0.3234,
+      "step": 6480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41672444321497e-06,
+      "loss": 0.3776,
+      "step": 6481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41394097545142e-06,
+      "loss": 0.3701,
+      "step": 6482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.411157633736139e-06,
+      "loss": 0.3424,
+      "step": 6483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.408374418290329e-06,
+      "loss": 0.4048,
+      "step": 6484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40559132933517e-06,
+      "loss": 0.3309,
+      "step": 6485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40280836709185e-06,
+      "loss": 0.3242,
+      "step": 6486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.400025531781532e-06,
+      "loss": 0.3241,
+      "step": 6487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.397242823625385e-06,
+      "loss": 0.3487,
+      "step": 6488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394460242844547e-06,
+      "loss": 0.3723,
+      "step": 6489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.391677789660158e-06,
+      "loss": 0.3683,
+      "step": 6490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.388895464293358e-06,
+      "loss": 0.3324,
+      "step": 6491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.386113266965254e-06,
+      "loss": 0.3523,
+      "step": 6492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38333119789696e-06,
+      "loss": 0.3406,
+      "step": 6493
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38054925730957e-06,
+      "loss": 0.3287,
+      "step": 6494
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.377767445424174e-06,
+      "loss": 0.3509,
+      "step": 6495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374985762461853e-06,
+      "loss": 0.3832,
+      "step": 6496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.372204208643673e-06,
+      "loss": 0.3534,
+      "step": 6497
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369422784190692e-06,
+      "loss": 0.3699,
+      "step": 6498
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.366641489323953e-06,
+      "loss": 0.3163,
+      "step": 6499
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.363860324264497e-06,
+      "loss": 0.3239,
+      "step": 6500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.361079289233352e-06,
+      "loss": 0.3766,
+      "step": 6501
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358298384451526e-06,
+      "loss": 0.3477,
+      "step": 6502
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355517610140034e-06,
+      "loss": 0.3982,
+      "step": 6503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.352736966519866e-06,
+      "loss": 0.3291,
+      "step": 6504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349956453812009e-06,
+      "loss": 0.3505,
+      "step": 6505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.347176072237438e-06,
+      "loss": 0.3376,
+      "step": 6506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344395822017112e-06,
+      "loss": 0.3542,
+      "step": 6507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.341615703371996e-06,
+      "loss": 0.3459,
+      "step": 6508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33883571652302e-06,
+      "loss": 0.3959,
+      "step": 6509
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.336055861691128e-06,
+      "loss": 0.3774,
+      "step": 6510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.333276139097238e-06,
+      "loss": 0.3643,
+      "step": 6511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.330496548962256e-06,
+      "loss": 0.3936,
+      "step": 6512
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.327717091507095e-06,
+      "loss": 0.3605,
+      "step": 6513
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.324937766952638e-06,
+      "loss": 0.3638,
+      "step": 6514
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.322158575519768e-06,
+      "loss": 0.3528,
+      "step": 6515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319379517429349e-06,
+      "loss": 0.3335,
+      "step": 6516
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.316600592902249e-06,
+      "loss": 0.3526,
+      "step": 6517
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.313821802159307e-06,
+      "loss": 0.3777,
+      "step": 6518
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.311043145421369e-06,
+      "loss": 0.3579,
+      "step": 6519
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30826462290926e-06,
+      "loss": 0.3987,
+      "step": 6520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305486234843792e-06,
+      "loss": 0.3373,
+      "step": 6521
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.302707981445777e-06,
+      "loss": 0.301,
+      "step": 6522
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299929862936004e-06,
+      "loss": 0.3207,
+      "step": 6523
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.297151879535262e-06,
+      "loss": 0.3553,
+      "step": 6524
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294374031464325e-06,
+      "loss": 0.3492,
+      "step": 6525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29159631894395e-06,
+      "loss": 0.3706,
+      "step": 6526
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.288818742194899e-06,
+      "loss": 0.322,
+      "step": 6527
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286041301437903e-06,
+      "loss": 0.3658,
+      "step": 6528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283263996893695e-06,
+      "loss": 0.3675,
+      "step": 6529
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280486828783002e-06,
+      "loss": 0.3595,
+      "step": 6530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277709797326523e-06,
+      "loss": 0.3963,
+      "step": 6531
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274932902744962e-06,
+      "loss": 0.3215,
+      "step": 6532
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272156145259006e-06,
+      "loss": 0.3707,
+      "step": 6533
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26937952508933e-06,
+      "loss": 0.3667,
+      "step": 6534
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266603042456596e-06,
+      "loss": 0.3576,
+      "step": 6535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.263826697581458e-06,
+      "loss": 0.3842,
+      "step": 6536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26105049068457e-06,
+      "loss": 0.3877,
+      "step": 6537
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258274421986553e-06,
+      "loss": 0.341,
+      "step": 6538
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.255498491708037e-06,
+      "loss": 0.3645,
+      "step": 6539
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.252722700069622e-06,
+      "loss": 0.351,
+      "step": 6540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.249947047291917e-06,
+      "loss": 0.3733,
+      "step": 6541
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247171533595509e-06,
+      "loss": 0.3625,
+      "step": 6542
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244396159200967e-06,
+      "loss": 0.3578,
+      "step": 6543
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.241620924328868e-06,
+      "loss": 0.3686,
+      "step": 6544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238845829199759e-06,
+      "loss": 0.3609,
+      "step": 6545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.236070874034193e-06,
+      "loss": 0.3457,
+      "step": 6546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.233296059052695e-06,
+      "loss": 0.3381,
+      "step": 6547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.230521384475788e-06,
+      "loss": 0.3425,
+      "step": 6548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.227746850523987e-06,
+      "loss": 0.3807,
+      "step": 6549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.224972457417785e-06,
+      "loss": 0.3672,
+      "step": 6550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.222198205377678e-06,
+      "loss": 0.3575,
+      "step": 6551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.219424094624136e-06,
+      "loss": 0.3793,
+      "step": 6552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216650125377627e-06,
+      "loss": 0.3705,
+      "step": 6553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213876297858611e-06,
+      "loss": 0.4105,
+      "step": 6554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.211102612287524e-06,
+      "loss": 0.3556,
+      "step": 6555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208329068884805e-06,
+      "loss": 0.3501,
+      "step": 6556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.205555667870864e-06,
+      "loss": 0.3808,
+      "step": 6557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202782409466123e-06,
+      "loss": 0.3507,
+      "step": 6558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.200009293890969e-06,
+      "loss": 0.3625,
+      "step": 6559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.197236321365796e-06,
+      "loss": 0.3405,
+      "step": 6560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194463492110982e-06,
+      "loss": 0.3596,
+      "step": 6561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19169080634688e-06,
+      "loss": 0.3419,
+      "step": 6562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188918264293852e-06,
+      "loss": 0.3194,
+      "step": 6563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186145866172235e-06,
+      "loss": 0.3942,
+      "step": 6564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18337361220236e-06,
+      "loss": 0.3599,
+      "step": 6565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180601502604547e-06,
+      "loss": 0.3745,
+      "step": 6566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.177829537599099e-06,
+      "loss": 0.3603,
+      "step": 6567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.175057717406315e-06,
+      "loss": 0.3493,
+      "step": 6568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.172286042246476e-06,
+      "loss": 0.3646,
+      "step": 6569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.169514512339854e-06,
+      "loss": 0.3334,
+      "step": 6570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.166743127906715e-06,
+      "loss": 0.4143,
+      "step": 6571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1639718891673e-06,
+      "loss": 0.3644,
+      "step": 6572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161200796341855e-06,
+      "loss": 0.3338,
+      "step": 6573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1584298496506e-06,
+      "loss": 0.3848,
+      "step": 6574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.155659049313755e-06,
+      "loss": 0.3629,
+      "step": 6575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152888395551513e-06,
+      "loss": 0.3357,
+      "step": 6576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150117888584074e-06,
+      "loss": 0.4131,
+      "step": 6577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147347528631617e-06,
+      "loss": 0.3524,
+      "step": 6578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.144577315914304e-06,
+      "loss": 0.3585,
+      "step": 6579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.141807250652299e-06,
+      "loss": 0.3309,
+      "step": 6580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139037333065736e-06,
+      "loss": 0.3803,
+      "step": 6581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136267563374756e-06,
+      "loss": 0.3219,
+      "step": 6582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.133497941799479e-06,
+      "loss": 0.3453,
+      "step": 6583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13072846856001e-06,
+      "loss": 0.347,
+      "step": 6584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.127959143876453e-06,
+      "loss": 0.3123,
+      "step": 6585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125189967968884e-06,
+      "loss": 0.4021,
+      "step": 6586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122420941057386e-06,
+      "loss": 0.3851,
+      "step": 6587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.119652063362015e-06,
+      "loss": 0.3336,
+      "step": 6588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.116883335102821e-06,
+      "loss": 0.3749,
+      "step": 6589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.114114756499847e-06,
+      "loss": 0.3545,
+      "step": 6590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.111346327773115e-06,
+      "loss": 0.361,
+      "step": 6591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108578049142643e-06,
+      "loss": 0.3696,
+      "step": 6592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.105809920828425e-06,
+      "loss": 0.3227,
+      "step": 6593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103041943050456e-06,
+      "loss": 0.3637,
+      "step": 6594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100274116028723e-06,
+      "loss": 0.3508,
+      "step": 6595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09750643998318e-06,
+      "loss": 0.3377,
+      "step": 6596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094738915133791e-06,
+      "loss": 0.2956,
+      "step": 6597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.091971541700488e-06,
+      "loss": 0.3677,
+      "step": 6598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.089204319903207e-06,
+      "loss": 0.3159,
+      "step": 6599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.086437249961871e-06,
+      "loss": 0.349,
+      "step": 6600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.083670332096382e-06,
+      "loss": 0.3467,
+      "step": 6601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.080903566526634e-06,
+      "loss": 0.3527,
+      "step": 6602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.078136953472507e-06,
+      "loss": 0.3362,
+      "step": 6603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.075370493153876e-06,
+      "loss": 0.325,
+      "step": 6604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.072604185790594e-06,
+      "loss": 0.3371,
+      "step": 6605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.069838031602509e-06,
+      "loss": 0.3456,
+      "step": 6606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.067072030809456e-06,
+      "loss": 0.3667,
+      "step": 6607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06430618363125e-06,
+      "loss": 0.3381,
+      "step": 6608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.061540490287712e-06,
+      "loss": 0.3405,
+      "step": 6609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.058774950998627e-06,
+      "loss": 0.3033,
+      "step": 6610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.056009565983783e-06,
+      "loss": 0.3224,
+      "step": 6611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05324433546296e-06,
+      "loss": 0.3507,
+      "step": 6612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.050479259655908e-06,
+      "loss": 0.3047,
+      "step": 6613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.047714338782378e-06,
+      "loss": 0.3415,
+      "step": 6614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.044949573062108e-06,
+      "loss": 0.3073,
+      "step": 6615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042184962714819e-06,
+      "loss": 0.3793,
+      "step": 6616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03942050796022e-06,
+      "loss": 0.3373,
+      "step": 6617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036656209018014e-06,
+      "loss": 0.3506,
+      "step": 6618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033892066107887e-06,
+      "loss": 0.3867,
+      "step": 6619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.031128079449505e-06,
+      "loss": 0.3235,
+      "step": 6620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.028364249262539e-06,
+      "loss": 0.3667,
+      "step": 6621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02560057576663e-06,
+      "loss": 0.3703,
+      "step": 6622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.022837059181421e-06,
+      "loss": 0.3272,
+      "step": 6623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.020073699726534e-06,
+      "loss": 0.3289,
+      "step": 6624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.017310497621575e-06,
+      "loss": 0.3385,
+      "step": 6625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014547453086151e-06,
+      "loss": 0.3478,
+      "step": 6626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.011784566339843e-06,
+      "loss": 0.3879,
+      "step": 6627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.009021837602228e-06,
+      "loss": 0.3495,
+      "step": 6628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.006259267092862e-06,
+      "loss": 0.3243,
+      "step": 6629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.003496855031298e-06,
+      "loss": 0.3465,
+      "step": 6630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.000734601637075e-06,
+      "loss": 0.3088,
+      "step": 6631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.99797250712971e-06,
+      "loss": 0.3158,
+      "step": 6632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.99521057172872e-06,
+      "loss": 0.3702,
+      "step": 6633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992448795653597e-06,
+      "loss": 0.3623,
+      "step": 6634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989687179123828e-06,
+      "loss": 0.3707,
+      "step": 6635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986925722358894e-06,
+      "loss": 0.3225,
+      "step": 6636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984164425578245e-06,
+      "loss": 0.348,
+      "step": 6637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981403289001336e-06,
+      "loss": 0.3362,
+      "step": 6638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978642312847592e-06,
+      "loss": 0.3347,
+      "step": 6639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.975881497336444e-06,
+      "loss": 0.3366,
+      "step": 6640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9731208426873e-06,
+      "loss": 0.3708,
+      "step": 6641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.970360349119552e-06,
+      "loss": 0.3883,
+      "step": 6642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967600016852591e-06,
+      "loss": 0.3391,
+      "step": 6643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.964839846105779e-06,
+      "loss": 0.3656,
+      "step": 6644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962079837098481e-06,
+      "loss": 0.3705,
+      "step": 6645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.959319990050038e-06,
+      "loss": 0.3139,
+      "step": 6646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.956560305179782e-06,
+      "loss": 0.3194,
+      "step": 6647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95380078270704e-06,
+      "loss": 0.3767,
+      "step": 6648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.951041422851105e-06,
+      "loss": 0.3036,
+      "step": 6649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948282225831285e-06,
+      "loss": 0.3543,
+      "step": 6650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.945523191866852e-06,
+      "loss": 0.3855,
+      "step": 6651
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.942764321177071e-06,
+      "loss": 0.347,
+      "step": 6652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940005613981207e-06,
+      "loss": 0.3359,
+      "step": 6653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.937247070498496e-06,
+      "loss": 0.3338,
+      "step": 6654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.934488690948166e-06,
+      "loss": 0.3665,
+      "step": 6655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.931730475549429e-06,
+      "loss": 0.3258,
+      "step": 6656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.928972424521497e-06,
+      "loss": 0.3355,
+      "step": 6657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926214538083548e-06,
+      "loss": 0.3415,
+      "step": 6658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.923456816454769e-06,
+      "loss": 0.3455,
+      "step": 6659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92069925985432e-06,
+      "loss": 0.3422,
+      "step": 6660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91794186850135e-06,
+      "loss": 0.3472,
+      "step": 6661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.915184642614998e-06,
+      "loss": 0.3769,
+      "step": 6662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91242758241438e-06,
+      "loss": 0.3682,
+      "step": 6663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.909670688118619e-06,
+      "loss": 0.3589,
+      "step": 6664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.906913959946807e-06,
+      "loss": 0.3488,
+      "step": 6665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.904157398118023e-06,
+      "loss": 0.3642,
+      "step": 6666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.901401002851351e-06,
+      "loss": 0.3405,
+      "step": 6667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.898644774365838e-06,
+      "loss": 0.3214,
+      "step": 6668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89588871288053e-06,
+      "loss": 0.3386,
+      "step": 6669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.893132818614467e-06,
+      "loss": 0.3283,
+      "step": 6670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.890377091786652e-06,
+      "loss": 0.3583,
+      "step": 6671
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887621532616106e-06,
+      "loss": 0.3496,
+      "step": 6672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884866141321811e-06,
+      "loss": 0.3385,
+      "step": 6673
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88211091812275e-06,
+      "loss": 0.402,
+      "step": 6674
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.879355863237878e-06,
+      "loss": 0.3789,
+      "step": 6675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876600976886158e-06,
+      "loss": 0.3736,
+      "step": 6676
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.873846259286524e-06,
+      "loss": 0.36,
+      "step": 6677
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871091710657899e-06,
+      "loss": 0.3633,
+      "step": 6678
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868337331219197e-06,
+      "loss": 0.3667,
+      "step": 6679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86558312118931e-06,
+      "loss": 0.3775,
+      "step": 6680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86282908078713e-06,
+      "loss": 0.3198,
+      "step": 6681
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.860075210231525e-06,
+      "loss": 0.3565,
+      "step": 6682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.857321509741349e-06,
+      "loss": 0.3241,
+      "step": 6683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85456797953545e-06,
+      "loss": 0.3687,
+      "step": 6684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.851814619832654e-06,
+      "loss": 0.3878,
+      "step": 6685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.849061430851785e-06,
+      "loss": 0.3744,
+      "step": 6686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.846308412811639e-06,
+      "loss": 0.3703,
+      "step": 6687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.843555565931005e-06,
+      "loss": 0.3338,
+      "step": 6688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840802890428668e-06,
+      "loss": 0.3321,
+      "step": 6689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83805038652338e-06,
+      "loss": 0.3551,
+      "step": 6690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.835298054433897e-06,
+      "loss": 0.3944,
+      "step": 6691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.832545894378948e-06,
+      "loss": 0.4049,
+      "step": 6692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829793906577256e-06,
+      "loss": 0.3397,
+      "step": 6693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.827042091247535e-06,
+      "loss": 0.3831,
+      "step": 6694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.824290448608473e-06,
+      "loss": 0.3732,
+      "step": 6695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82153897887875e-06,
+      "loss": 0.3274,
+      "step": 6696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.818787682277034e-06,
+      "loss": 0.3402,
+      "step": 6697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81603655902198e-06,
+      "loss": 0.3315,
+      "step": 6698
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81328560933222e-06,
+      "loss": 0.3658,
+      "step": 6699
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.810534833426387e-06,
+      "loss": 0.3206,
+      "step": 6700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80778423152309e-06,
+      "loss": 0.3689,
+      "step": 6701
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80503380384092e-06,
+      "loss": 0.3605,
+      "step": 6702
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.802283550598473e-06,
+      "loss": 0.3415,
+      "step": 6703
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79953347201431e-06,
+      "loss": 0.3602,
+      "step": 6704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796783568306988e-06,
+      "loss": 0.3892,
+      "step": 6705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.794033839695052e-06,
+      "loss": 0.3461,
+      "step": 6706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.791284286397026e-06,
+      "loss": 0.3986,
+      "step": 6707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78853490863143e-06,
+      "loss": 0.3752,
+      "step": 6708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.785785706616758e-06,
+      "loss": 0.3533,
+      "step": 6709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.783036680571498e-06,
+      "loss": 0.3077,
+      "step": 6710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.780287830714128e-06,
+      "loss": 0.3464,
+      "step": 6711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777539157263096e-06,
+      "loss": 0.3462,
+      "step": 6712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.361,
+      "step": 6713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772042340453835e-06,
+      "loss": 0.3745,
+      "step": 6714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76929419753245e-06,
+      "loss": 0.3598,
+      "step": 6715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766546231891094e-06,
+      "loss": 0.374,
+      "step": 6716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.763798443748167e-06,
+      "loss": 0.3407,
+      "step": 6717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761050833322041e-06,
+      "loss": 0.3414,
+      "step": 6718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75830340083107e-06,
+      "loss": 0.373,
+      "step": 6719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.755556146493605e-06,
+      "loss": 0.4047,
+      "step": 6720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75280907052797e-06,
+      "loss": 0.3886,
+      "step": 6721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.750062173152494e-06,
+      "loss": 0.3486,
+      "step": 6722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747315454585475e-06,
+      "loss": 0.3594,
+      "step": 6723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.744568915045195e-06,
+      "loss": 0.3474,
+      "step": 6724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.741822554749939e-06,
+      "loss": 0.338,
+      "step": 6725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.739076373917959e-06,
+      "loss": 0.3478,
+      "step": 6726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73633037276751e-06,
+      "loss": 0.3935,
+      "step": 6727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733584551516817e-06,
+      "loss": 0.3831,
+      "step": 6728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.730838910384098e-06,
+      "loss": 0.3629,
+      "step": 6729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.728093449587562e-06,
+      "loss": 0.3457,
+      "step": 6730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725348169345391e-06,
+      "loss": 0.3813,
+      "step": 6731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.722603069875766e-06,
+      "loss": 0.3792,
+      "step": 6732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.719858151396838e-06,
+      "loss": 0.331,
+      "step": 6733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71711341412676e-06,
+      "loss": 0.3497,
+      "step": 6734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.714368858283667e-06,
+      "loss": 0.3561,
+      "step": 6735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.711624484085668e-06,
+      "loss": 0.353,
+      "step": 6736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.708880291750872e-06,
+      "loss": 0.3934,
+      "step": 6737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706136281497359e-06,
+      "loss": 0.3162,
+      "step": 6738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.703392453543208e-06,
+      "loss": 0.3775,
+      "step": 6739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.700648808106483e-06,
+      "loss": 0.3739,
+      "step": 6740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697905345405218e-06,
+      "loss": 0.3335,
+      "step": 6741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695162065657454e-06,
+      "loss": 0.3429,
+      "step": 6742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692418969081195e-06,
+      "loss": 0.3422,
+      "step": 6743
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.689676055894454e-06,
+      "loss": 0.323,
+      "step": 6744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.686933326315209e-06,
+      "loss": 0.3328,
+      "step": 6745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.684190780561433e-06,
+      "loss": 0.3081,
+      "step": 6746
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681448418851091e-06,
+      "loss": 0.3104,
+      "step": 6747
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.678706241402114e-06,
+      "loss": 0.3793,
+      "step": 6748
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675964248432441e-06,
+      "loss": 0.386,
+      "step": 6749
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.673222440159978e-06,
+      "loss": 0.3679,
+      "step": 6750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670480816802623e-06,
+      "loss": 0.3686,
+      "step": 6751
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667739378578268e-06,
+      "loss": 0.3279,
+      "step": 6752
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664998125704775e-06,
+      "loss": 0.3826,
+      "step": 6753
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662257058400003e-06,
+      "loss": 0.3823,
+      "step": 6754
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659516176881785e-06,
+      "loss": 0.3315,
+      "step": 6755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.656775481367954e-06,
+      "loss": 0.3406,
+      "step": 6756
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.654034972076314e-06,
+      "loss": 0.3517,
+      "step": 6757
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651294649224665e-06,
+      "loss": 0.3656,
+      "step": 6758
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.648554513030788e-06,
+      "loss": 0.3418,
+      "step": 6759
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.645814563712441e-06,
+      "loss": 0.3702,
+      "step": 6760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643074801487386e-06,
+      "loss": 0.3474,
+      "step": 6761
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64033522657335e-06,
+      "loss": 0.3564,
+      "step": 6762
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637595839188061e-06,
+      "loss": 0.3706,
+      "step": 6763
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634856639549223e-06,
+      "loss": 0.2974,
+      "step": 6764
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632117627874522e-06,
+      "loss": 0.3695,
+      "step": 6765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.629378804381645e-06,
+      "loss": 0.3404,
+      "step": 6766
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.626640169288244e-06,
+      "loss": 0.3901,
+      "step": 6767
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623901722811968e-06,
+      "loss": 0.3562,
+      "step": 6768
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6211634651704534e-06,
+      "loss": 0.3644,
+      "step": 6769
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6184253965813085e-06,
+      "loss": 0.3775,
+      "step": 6770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.615687517262143e-06,
+      "loss": 0.3632,
+      "step": 6771
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.612949827430538e-06,
+      "loss": 0.3509,
+      "step": 6772
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6102123273040695e-06,
+      "loss": 0.3558,
+      "step": 6773
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607475017100285e-06,
+      "loss": 0.3525,
+      "step": 6774
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6047378970367315e-06,
+      "loss": 0.3837,
+      "step": 6775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.60200096733094e-06,
+      "loss": 0.3598,
+      "step": 6776
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.599264228200414e-06,
+      "loss": 0.3664,
+      "step": 6777
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.596527679862655e-06,
+      "loss": 0.3568,
+      "step": 6778
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.593791322535136e-06,
+      "loss": 0.3575,
+      "step": 6779
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591055156435327e-06,
+      "loss": 0.3453,
+      "step": 6780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.588319181780682e-06,
+      "loss": 0.3459,
+      "step": 6781
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585583398788628e-06,
+      "loss": 0.3511,
+      "step": 6782
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582847807676595e-06,
+      "loss": 0.333,
+      "step": 6783
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.580112408661975e-06,
+      "loss": 0.3419,
+      "step": 6784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.57737720196217e-06,
+      "loss": 0.3631,
+      "step": 6785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.574642187794545e-06,
+      "loss": 0.3379,
+      "step": 6786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5719073663764605e-06,
+      "loss": 0.36,
+      "step": 6787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.569172737925267e-06,
+      "loss": 0.3685,
+      "step": 6788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.566438302658283e-06,
+      "loss": 0.3687,
+      "step": 6789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.563704060792826e-06,
+      "loss": 0.3472,
+      "step": 6790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.560970012546192e-06,
+      "loss": 0.3815,
+      "step": 6791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.558236158135661e-06,
+      "loss": 0.3359,
+      "step": 6792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.555502497778507e-06,
+      "loss": 0.3671,
+      "step": 6793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.552769031691976e-06,
+      "loss": 0.3769,
+      "step": 6794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.550035760093304e-06,
+      "loss": 0.35,
+      "step": 6795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547302683199707e-06,
+      "loss": 0.3368,
+      "step": 6796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.544569801228399e-06,
+      "loss": 0.3391,
+      "step": 6797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541837114396559e-06,
+      "loss": 0.3567,
+      "step": 6798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.539104622921368e-06,
+      "loss": 0.3577,
+      "step": 6799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.536372327019986e-06,
+      "loss": 0.3746,
+      "step": 6800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.533640226909547e-06,
+      "loss": 0.3744,
+      "step": 6801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530908322807187e-06,
+      "loss": 0.3609,
+      "step": 6802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.528176614930014e-06,
+      "loss": 0.3295,
+      "step": 6803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525445103495119e-06,
+      "loss": 0.3379,
+      "step": 6804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.522713788719593e-06,
+      "loss": 0.3749,
+      "step": 6805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.51998267082049e-06,
+      "loss": 0.3524,
+      "step": 6806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.517251750014869e-06,
+      "loss": 0.3605,
+      "step": 6807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.514521026519755e-06,
+      "loss": 0.3714,
+      "step": 6808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5117905005521686e-06,
+      "loss": 0.3584,
+      "step": 6809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.509060172329115e-06,
+      "loss": 0.3384,
+      "step": 6810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.506330042067575e-06,
+      "loss": 0.3324,
+      "step": 6811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.503600109984527e-06,
+      "loss": 0.342,
+      "step": 6812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.500870376296918e-06,
+      "loss": 0.3458,
+      "step": 6813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498140841221693e-06,
+      "loss": 0.3755,
+      "step": 6814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.495411504975767e-06,
+      "loss": 0.3244,
+      "step": 6815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.492682367776055e-06,
+      "loss": 0.3391,
+      "step": 6816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489953429839449e-06,
+      "loss": 0.331,
+      "step": 6817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48722469138282e-06,
+      "loss": 0.3596,
+      "step": 6818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484496152623033e-06,
+      "loss": 0.3222,
+      "step": 6819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481767813776926e-06,
+      "loss": 0.3464,
+      "step": 6820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.479039675061333e-06,
+      "loss": 0.3517,
+      "step": 6821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.476311736693066e-06,
+      "loss": 0.3509,
+      "step": 6822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473583998888916e-06,
+      "loss": 0.3222,
+      "step": 6823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.47085646186567e-06,
+      "loss": 0.3641,
+      "step": 6824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.468129125840087e-06,
+      "loss": 0.3612,
+      "step": 6825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465401991028921e-06,
+      "loss": 0.3618,
+      "step": 6826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4626750576489e-06,
+      "loss": 0.3292,
+      "step": 6827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.459948325916741e-06,
+      "loss": 0.3631,
+      "step": 6828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4572217960491485e-06,
+      "loss": 0.3353,
+      "step": 6829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4544954682628034e-06,
+      "loss": 0.3709,
+      "step": 6830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4517693427743765e-06,
+      "loss": 0.3384,
+      "step": 6831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449043419800515e-06,
+      "loss": 0.3186,
+      "step": 6832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446317699557857e-06,
+      "loss": 0.3676,
+      "step": 6833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.443592182263031e-06,
+      "loss": 0.3332,
+      "step": 6834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.44086686813263e-06,
+      "loss": 0.3634,
+      "step": 6835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.438141757383249e-06,
+      "loss": 0.3812,
+      "step": 6836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435416850231454e-06,
+      "loss": 0.3714,
+      "step": 6837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.432692146893804e-06,
+      "loss": 0.3549,
+      "step": 6838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.429967647586838e-06,
+      "loss": 0.3371,
+      "step": 6839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.427243352527079e-06,
+      "loss": 0.346,
+      "step": 6840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.424519261931036e-06,
+      "loss": 0.3426,
+      "step": 6841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.421795376015193e-06,
+      "loss": 0.3622,
+      "step": 6842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.419071694996033e-06,
+      "loss": 0.3426,
+      "step": 6843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.416348219090007e-06,
+      "loss": 0.3407,
+      "step": 6844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.413624948513558e-06,
+      "loss": 0.3549,
+      "step": 6845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.410901883483117e-06,
+      "loss": 0.3503,
+      "step": 6846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408179024215085e-06,
+      "loss": 0.3442,
+      "step": 6847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.405456370925863e-06,
+      "loss": 0.3832,
+      "step": 6848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402733923831822e-06,
+      "loss": 0.3501,
+      "step": 6849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40001168314932e-06,
+      "loss": 0.3202,
+      "step": 6850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.397289649094708e-06,
+      "loss": 0.3388,
+      "step": 6851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394567821884306e-06,
+      "loss": 0.3982,
+      "step": 6852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3918462017344295e-06,
+      "loss": 0.3372,
+      "step": 6853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3891247888613705e-06,
+      "loss": 0.3891,
+      "step": 6854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.38640358348141e-06,
+      "loss": 0.3203,
+      "step": 6855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383682585810802e-06,
+      "loss": 0.3416,
+      "step": 6856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3809617960657976e-06,
+      "loss": 0.329,
+      "step": 6857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.378241214462627e-06,
+      "loss": 0.318,
+      "step": 6858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.375520841217493e-06,
+      "loss": 0.3424,
+      "step": 6859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372800676546601e-06,
+      "loss": 0.3205,
+      "step": 6860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37008072066612e-06,
+      "loss": 0.3568,
+      "step": 6861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3673609737922195e-06,
+      "loss": 0.3728,
+      "step": 6862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.364641436141044e-06,
+      "loss": 0.388,
+      "step": 6863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.361922107928716e-06,
+      "loss": 0.3596,
+      "step": 6864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.359202989371358e-06,
+      "loss": 0.3278,
+      "step": 6865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356484080685055e-06,
+      "loss": 0.3308,
+      "step": 6866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.353765382085896e-06,
+      "loss": 0.3438,
+      "step": 6867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35104689378993e-06,
+      "loss": 0.3138,
+      "step": 6868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.348328616013213e-06,
+      "loss": 0.3143,
+      "step": 6869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.345610548971773e-06,
+      "loss": 0.3628,
+      "step": 6870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.342892692881618e-06,
+      "loss": 0.4071,
+      "step": 6871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.340175047958747e-06,
+      "loss": 0.3584,
+      "step": 6872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337457614419131e-06,
+      "loss": 0.3189,
+      "step": 6873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.334740392478737e-06,
+      "loss": 0.3668,
+      "step": 6874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332023382353515e-06,
+      "loss": 0.3647,
+      "step": 6875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.329306584259384e-06,
+      "loss": 0.3442,
+      "step": 6876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326589998412259e-06,
+      "loss": 0.3551,
+      "step": 6877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.323873625028031e-06,
+      "loss": 0.3746,
+      "step": 6878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3211574643225825e-06,
+      "loss": 0.3613,
+      "step": 6879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.318441516511772e-06,
+      "loss": 0.3771,
+      "step": 6880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.315725781811439e-06,
+      "loss": 0.3525,
+      "step": 6881
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.313010260437417e-06,
+      "loss": 0.3273,
+      "step": 6882
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310294952605508e-06,
+      "loss": 0.3137,
+      "step": 6883
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.307579858531513e-06,
+      "loss": 0.339,
+      "step": 6884
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.304864978431201e-06,
+      "loss": 0.3513,
+      "step": 6885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302150312520331e-06,
+      "loss": 0.3261,
+      "step": 6886
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29943586101465e-06,
+      "loss": 0.3607,
+      "step": 6887
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.296721624129874e-06,
+      "loss": 0.3502,
+      "step": 6888
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.294007602081721e-06,
+      "loss": 0.3103,
+      "step": 6889
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.291293795085874e-06,
+      "loss": 0.3514,
+      "step": 6890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2885802033580055e-06,
+      "loss": 0.3545,
+      "step": 6891
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285866827113779e-06,
+      "loss": 0.3415,
+      "step": 6892
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.283153666568828e-06,
+      "loss": 0.3693,
+      "step": 6893
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280440721938777e-06,
+      "loss": 0.3604,
+      "step": 6894
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2777279934392256e-06,
+      "loss": 0.3116,
+      "step": 6895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.275015481285769e-06,
+      "loss": 0.3329,
+      "step": 6896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27230318569397e-06,
+      "loss": 0.3351,
+      "step": 6897
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2695911068793874e-06,
+      "loss": 0.3389,
+      "step": 6898
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.266879245057559e-06,
+      "loss": 0.3816,
+      "step": 6899
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.264167600443995e-06,
+      "loss": 0.3823,
+      "step": 6900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2614561732542064e-06,
+      "loss": 0.3344,
+      "step": 6901
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25874496370367e-06,
+      "loss": 0.3535,
+      "step": 6902
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2560339720078576e-06,
+      "loss": 0.3778,
+      "step": 6903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.253323198382219e-06,
+      "loss": 0.3618,
+      "step": 6904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25061264304218e-06,
+      "loss": 0.3389,
+      "step": 6905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.247902306203165e-06,
+      "loss": 0.349,
+      "step": 6906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.245192188080563e-06,
+      "loss": 0.355,
+      "step": 6907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242482288889758e-06,
+      "loss": 0.3473,
+      "step": 6908
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.239772608846116e-06,
+      "loss": 0.3196,
+      "step": 6909
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.237063148164974e-06,
+      "loss": 0.378,
+      "step": 6910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.234353907061671e-06,
+      "loss": 0.366,
+      "step": 6911
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2316448857515076e-06,
+      "loss": 0.3374,
+      "step": 6912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.228936084449783e-06,
+      "loss": 0.3377,
+      "step": 6913
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.226227503371768e-06,
+      "loss": 0.3602,
+      "step": 6914
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.223519142732722e-06,
+      "loss": 0.3408,
+      "step": 6915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.220811002747891e-06,
+      "loss": 0.4171,
+      "step": 6916
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2181030836324895e-06,
+      "loss": 0.375,
+      "step": 6917
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2153953856017314e-06,
+      "loss": 0.3399,
+      "step": 6918
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.212687908870795e-06,
+      "loss": 0.3531,
+      "step": 6919
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.209980653654858e-06,
+      "loss": 0.3361,
+      "step": 6920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207273620169073e-06,
+      "loss": 0.3851,
+      "step": 6921
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204566808628568e-06,
+      "loss": 0.3241,
+      "step": 6922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20186021924847e-06,
+      "loss": 0.3768,
+      "step": 6923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.19915385224387e-06,
+      "loss": 0.3508,
+      "step": 6924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1964477078298574e-06,
+      "loss": 0.3803,
+      "step": 6925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.193741786221493e-06,
+      "loss": 0.3355,
+      "step": 6926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.19103608763382e-06,
+      "loss": 0.3387,
+      "step": 6927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.188330612281876e-06,
+      "loss": 0.357,
+      "step": 6928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.185625360380666e-06,
+      "loss": 0.3488,
+      "step": 6929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182920332145185e-06,
+      "loss": 0.3559,
+      "step": 6930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.180215527790406e-06,
+      "loss": 0.327,
+      "step": 6931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177510947531287e-06,
+      "loss": 0.3323,
+      "step": 6932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174806591582777e-06,
+      "loss": 0.3458,
+      "step": 6933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.172102460159789e-06,
+      "loss": 0.3225,
+      "step": 6934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.169398553477234e-06,
+      "loss": 0.3083,
+      "step": 6935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.166694871749988e-06,
+      "loss": 0.3375,
+      "step": 6936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1639914151929255e-06,
+      "loss": 0.368,
+      "step": 6937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.161288184020905e-06,
+      "loss": 0.3188,
+      "step": 6938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158585178448749e-06,
+      "loss": 0.38,
+      "step": 6939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.155882398691278e-06,
+      "loss": 0.3908,
+      "step": 6940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153179844963283e-06,
+      "loss": 0.373,
+      "step": 6941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.15047751747955e-06,
+      "loss": 0.3241,
+      "step": 6942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.147775416454837e-06,
+      "loss": 0.3554,
+      "step": 6943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.145073542103884e-06,
+      "loss": 0.3254,
+      "step": 6944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.142371894641424e-06,
+      "loss": 0.3321,
+      "step": 6945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.139670474282153e-06,
+      "loss": 0.3362,
+      "step": 6946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.136969281240771e-06,
+      "loss": 0.3722,
+      "step": 6947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.134268315731942e-06,
+      "loss": 0.346,
+      "step": 6948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1315675779703195e-06,
+      "loss": 0.3379,
+      "step": 6949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.128867068170543e-06,
+      "loss": 0.3636,
+      "step": 6950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1261667865472215e-06,
+      "loss": 0.3487,
+      "step": 6951
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.123466733314961e-06,
+      "loss": 0.343,
+      "step": 6952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1207669086883366e-06,
+      "loss": 0.3631,
+      "step": 6953
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.118067312881914e-06,
+      "loss": 0.3416,
+      "step": 6954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.115367946110232e-06,
+      "loss": 0.3293,
+      "step": 6955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112668808587821e-06,
+      "loss": 0.338,
+      "step": 6956
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.109969900529189e-06,
+      "loss": 0.3669,
+      "step": 6957
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.107271222148819e-06,
+      "loss": 0.343,
+      "step": 6958
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.104572773661191e-06,
+      "loss": 0.3497,
+      "step": 6959
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10187455528075e-06,
+      "loss": 0.3522,
+      "step": 6960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099176567221936e-06,
+      "loss": 0.3309,
+      "step": 6961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.096478809699164e-06,
+      "loss": 0.3784,
+      "step": 6962
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.093781282926827e-06,
+      "loss": 0.3404,
+      "step": 6963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0910839871193124e-06,
+      "loss": 0.33,
+      "step": 6964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088386922490975e-06,
+      "loss": 0.3206,
+      "step": 6965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.085690089256161e-06,
+      "loss": 0.3733,
+      "step": 6966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0829934876291926e-06,
+      "loss": 0.343,
+      "step": 6967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.080297117824375e-06,
+      "loss": 0.3822,
+      "step": 6968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.077600980056003e-06,
+      "loss": 0.3446,
+      "step": 6969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.074905074538337e-06,
+      "loss": 0.3931,
+      "step": 6970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.072209401485633e-06,
+      "loss": 0.3161,
+      "step": 6971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0695139611121175e-06,
+      "loss": 0.354,
+      "step": 6972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.06681875363201e-06,
+      "loss": 0.3415,
+      "step": 6973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.064123779259505e-06,
+      "loss": 0.3243,
+      "step": 6974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.061429038208778e-06,
+      "loss": 0.3784,
+      "step": 6975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.058734530693987e-06,
+      "loss": 0.3398,
+      "step": 6976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.056040256929268e-06,
+      "loss": 0.3844,
+      "step": 6977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053346217128749e-06,
+      "loss": 0.3564,
+      "step": 6978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.05065241150653e-06,
+      "loss": 0.3635,
+      "step": 6979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.04795884027669e-06,
+      "loss": 0.3596,
+      "step": 6980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045265503653302e-06,
+      "loss": 0.3428,
+      "step": 6981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0425724018504036e-06,
+      "loss": 0.3959,
+      "step": 6982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.039879535082033e-06,
+      "loss": 0.3453,
+      "step": 6983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037186903562191e-06,
+      "loss": 0.3647,
+      "step": 6984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0344945075048696e-06,
+      "loss": 0.3142,
+      "step": 6985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.031802347124046e-06,
+      "loss": 0.3352,
+      "step": 6986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.029110422633664e-06,
+      "loss": 0.3566,
+      "step": 6987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.026418734247669e-06,
+      "loss": 0.3692,
+      "step": 6988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.023727282179966e-06,
+      "loss": 0.3594,
+      "step": 6989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.021036066644455e-06,
+      "loss": 0.3277,
+      "step": 6990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.018345087855022e-06,
+      "loss": 0.3522,
+      "step": 6991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.015654346025516e-06,
+      "loss": 0.3537,
+      "step": 6992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0129638413697824e-06,
+      "loss": 0.3468,
+      "step": 6993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.010273574101636e-06,
+      "loss": 0.3236,
+      "step": 6994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0075835444348905e-06,
+      "loss": 0.3498,
+      "step": 6995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.004893752583318e-06,
+      "loss": 0.3218,
+      "step": 6996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.002204198760691e-06,
+      "loss": 0.3396,
+      "step": 6997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999514883180757e-06,
+      "loss": 0.3869,
+      "step": 6998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9968258060572335e-06,
+      "loss": 0.3526,
+      "step": 6999
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.994136967603838e-06,
+      "loss": 0.3377,
+      "step": 7000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.991448368034253e-06,
+      "loss": 0.3741,
+      "step": 7001
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988760007562152e-06,
+      "loss": 0.3431,
+      "step": 7002
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.986071886401189e-06,
+      "loss": 0.321,
+      "step": 7003
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983384004764987e-06,
+      "loss": 0.3353,
+      "step": 7004
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.980696362867169e-06,
+      "loss": 0.3353,
+      "step": 7005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.978008960921322e-06,
+      "loss": 0.3117,
+      "step": 7006
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.975321799141022e-06,
+      "loss": 0.3205,
+      "step": 7007
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9726348777398275e-06,
+      "loss": 0.3678,
+      "step": 7008
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.969948196931272e-06,
+      "loss": 0.3558,
+      "step": 7009
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967261756928878e-06,
+      "loss": 0.3553,
+      "step": 7010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9645755579461396e-06,
+      "loss": 0.3589,
+      "step": 7011
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9618896001965385e-06,
+      "loss": 0.3314,
+      "step": 7012
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959203883893528e-06,
+      "loss": 0.3405,
+      "step": 7013
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956518409250559e-06,
+      "loss": 0.3561,
+      "step": 7014
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.953833176481047e-06,
+      "loss": 0.2999,
+      "step": 7015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951148185798397e-06,
+      "loss": 0.3502,
+      "step": 7016
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.948463437415993e-06,
+      "loss": 0.3615,
+      "step": 7017
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.945778931547194e-06,
+      "loss": 0.3261,
+      "step": 7018
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.943094668405348e-06,
+      "loss": 0.3789,
+      "step": 7019
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9404106482037844e-06,
+      "loss": 0.3564,
+      "step": 7020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.937726871155801e-06,
+      "loss": 0.3633,
+      "step": 7021
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.935043337474694e-06,
+      "loss": 0.3215,
+      "step": 7022
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.932360047373721e-06,
+      "loss": 0.4019,
+      "step": 7023
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929677001066139e-06,
+      "loss": 0.364,
+      "step": 7024
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.926994198765171e-06,
+      "loss": 0.3406,
+      "step": 7025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.924311640684025e-06,
+      "loss": 0.3396,
+      "step": 7026
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921629327035899e-06,
+      "loss": 0.3482,
+      "step": 7027
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.918947258033957e-06,
+      "loss": 0.3823,
+      "step": 7028
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.91626543389135e-06,
+      "loss": 0.3383,
+      "step": 7029
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.913583854821212e-06,
+      "loss": 0.3652,
+      "step": 7030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.910902521036652e-06,
+      "loss": 0.3515,
+      "step": 7031
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.908221432750769e-06,
+      "loss": 0.3342,
+      "step": 7032
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.90554059017663e-06,
+      "loss": 0.3231,
+      "step": 7033
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.902859993527293e-06,
+      "loss": 0.3341,
+      "step": 7034
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.900179643015785e-06,
+      "loss": 0.3566,
+      "step": 7035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897499538855129e-06,
+      "loss": 0.3381,
+      "step": 7036
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894819681258312e-06,
+      "loss": 0.4068,
+      "step": 7037
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.892140070438315e-06,
+      "loss": 0.3927,
+      "step": 7038
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.889460706608095e-06,
+      "loss": 0.3638,
+      "step": 7039
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88678158998058e-06,
+      "loss": 0.3436,
+      "step": 7040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.884102720768697e-06,
+      "loss": 0.3588,
+      "step": 7041
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.881424099185334e-06,
+      "loss": 0.3339,
+      "step": 7042
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.87874572544337e-06,
+      "loss": 0.3361,
+      "step": 7043
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876067599755667e-06,
+      "loss": 0.3642,
+      "step": 7044
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.873389722335056e-06,
+      "loss": 0.381,
+      "step": 7045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.870712093394361e-06,
+      "loss": 0.3235,
+      "step": 7046
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868034713146377e-06,
+      "loss": 0.3684,
+      "step": 7047
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86535758180388e-06,
+      "loss": 0.3603,
+      "step": 7048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.862680699579636e-06,
+      "loss": 0.3298,
+      "step": 7049
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.860004066686374e-06,
+      "loss": 0.3331,
+      "step": 7050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8573276833368245e-06,
+      "loss": 0.3531,
+      "step": 7051
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.854651549743677e-06,
+      "loss": 0.3566,
+      "step": 7052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8519756661196165e-06,
+      "loss": 0.3528,
+      "step": 7053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849300032677298e-06,
+      "loss": 0.3372,
+      "step": 7054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8466246496293656e-06,
+      "loss": 0.3602,
+      "step": 7055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8439495171884375e-06,
+      "loss": 0.3201,
+      "step": 7056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8412746355671085e-06,
+      "loss": 0.3106,
+      "step": 7057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.838600004977969e-06,
+      "loss": 0.3432,
+      "step": 7058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835925625633567e-06,
+      "loss": 0.3558,
+      "step": 7059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.833251497746451e-06,
+      "loss": 0.3499,
+      "step": 7060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.83057762152914e-06,
+      "loss": 0.335,
+      "step": 7061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.827903997194128e-06,
+      "loss": 0.3531,
+      "step": 7062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.825230624953904e-06,
+      "loss": 0.3393,
+      "step": 7063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.822557505020918e-06,
+      "loss": 0.3293,
+      "step": 7064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.819884637607619e-06,
+      "loss": 0.3315,
+      "step": 7065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.817212022926421e-06,
+      "loss": 0.3848,
+      "step": 7066
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.814539661189723e-06,
+      "loss": 0.3292,
+      "step": 7067
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811867552609912e-06,
+      "loss": 0.3467,
+      "step": 7068
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.809195697399338e-06,
+      "loss": 0.3188,
+      "step": 7069
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.806524095770348e-06,
+      "loss": 0.3287,
+      "step": 7070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.803852747935255e-06,
+      "loss": 0.3356,
+      "step": 7071
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.801181654106359e-06,
+      "loss": 0.3239,
+      "step": 7072
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.798510814495945e-06,
+      "loss": 0.3814,
+      "step": 7073
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.795840229316266e-06,
+      "loss": 0.3332,
+      "step": 7074
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.793169898779564e-06,
+      "loss": 0.3734,
+      "step": 7075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790499823098051e-06,
+      "loss": 0.3086,
+      "step": 7076
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.787830002483931e-06,
+      "loss": 0.352,
+      "step": 7077
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.785160437149379e-06,
+      "loss": 0.3383,
+      "step": 7078
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782491127306552e-06,
+      "loss": 0.332,
+      "step": 7079
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.77982207316759e-06,
+      "loss": 0.3488,
+      "step": 7080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.777153274944603e-06,
+      "loss": 0.3709,
+      "step": 7081
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.774484732849698e-06,
+      "loss": 0.3465,
+      "step": 7082
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.771816447094942e-06,
+      "loss": 0.3222,
+      "step": 7083
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7691484178923915e-06,
+      "loss": 0.3647,
+      "step": 7084
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.766480645454089e-06,
+      "loss": 0.3465,
+      "step": 7085
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.763813129992038e-06,
+      "loss": 0.3597,
+      "step": 7086
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.761145871718246e-06,
+      "loss": 0.304,
+      "step": 7087
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758478870844677e-06,
+      "loss": 0.3587,
+      "step": 7088
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.755812127583286e-06,
+      "loss": 0.3805,
+      "step": 7089
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7531456421460115e-06,
+      "loss": 0.3355,
+      "step": 7090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750479414744761e-06,
+      "loss": 0.3339,
+      "step": 7091
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.74781344559143e-06,
+      "loss": 0.3772,
+      "step": 7092
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7451477348978835e-06,
+      "loss": 0.347,
+      "step": 7093
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.742482282875982e-06,
+      "loss": 0.3122,
+      "step": 7094
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.739817089737547e-06,
+      "loss": 0.3661,
+      "step": 7095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737152155694394e-06,
+      "loss": 0.3546,
+      "step": 7096
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7344874809583136e-06,
+      "loss": 0.3847,
+      "step": 7097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.731823065741066e-06,
+      "loss": 0.3372,
+      "step": 7098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729158910254411e-06,
+      "loss": 0.3266,
+      "step": 7099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.726495014710064e-06,
+      "loss": 0.3485,
+      "step": 7100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723831379319741e-06,
+      "loss": 0.3784,
+      "step": 7101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.721168004295126e-06,
+      "loss": 0.2992,
+      "step": 7102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.71850488984788e-06,
+      "loss": 0.3321,
+      "step": 7103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.715842036189655e-06,
+      "loss": 0.3391,
+      "step": 7104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.713179443532068e-06,
+      "loss": 0.3423,
+      "step": 7105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7105171120867275e-06,
+      "loss": 0.33,
+      "step": 7106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.707855042065209e-06,
+      "loss": 0.3655,
+      "step": 7107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705193233679079e-06,
+      "loss": 0.336,
+      "step": 7108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.702531687139882e-06,
+      "loss": 0.345,
+      "step": 7109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699870402659132e-06,
+      "loss": 0.3029,
+      "step": 7110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697209380448333e-06,
+      "loss": 0.3596,
+      "step": 7111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6945486207189554e-06,
+      "loss": 0.3273,
+      "step": 7112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.691888123682464e-06,
+      "loss": 0.3345,
+      "step": 7113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.689227889550296e-06,
+      "loss": 0.366,
+      "step": 7114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.686567918533864e-06,
+      "loss": 0.3557,
+      "step": 7115
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.683908210844565e-06,
+      "loss": 0.3576,
+      "step": 7116
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.681248766693769e-06,
+      "loss": 0.3695,
+      "step": 7117
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.678589586292832e-06,
+      "loss": 0.3464,
+      "step": 7118
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.675930669853088e-06,
+      "loss": 0.3212,
+      "step": 7119
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.673272017585842e-06,
+      "loss": 0.3436,
+      "step": 7120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670613629702391e-06,
+      "loss": 0.341,
+      "step": 7121
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.667955506413996e-06,
+      "loss": 0.3377,
+      "step": 7122
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.665297647931917e-06,
+      "loss": 0.3386,
+      "step": 7123
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66264005446737e-06,
+      "loss": 0.329,
+      "step": 7124
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.659982726231562e-06,
+      "loss": 0.3514,
+      "step": 7125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.657325663435686e-06,
+      "loss": 0.3092,
+      "step": 7126
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.654668866290897e-06,
+      "loss": 0.3463,
+      "step": 7127
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652012335008342e-06,
+      "loss": 0.3844,
+      "step": 7128
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6493560697991426e-06,
+      "loss": 0.3689,
+      "step": 7129
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6467000708743945e-06,
+      "loss": 0.3445,
+      "step": 7130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6440443384451835e-06,
+      "loss": 0.3586,
+      "step": 7131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.641388872722564e-06,
+      "loss": 0.3387,
+      "step": 7132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638733673917575e-06,
+      "loss": 0.3128,
+      "step": 7133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6360787422412266e-06,
+      "loss": 0.3702,
+      "step": 7134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.63342407790452e-06,
+      "loss": 0.3592,
+      "step": 7135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.630769681118423e-06,
+      "loss": 0.3499,
+      "step": 7136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.628115552093891e-06,
+      "loss": 0.3226,
+      "step": 7137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.625461691041856e-06,
+      "loss": 0.3669,
+      "step": 7138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.622808098173219e-06,
+      "loss": 0.3683,
+      "step": 7139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620154773698879e-06,
+      "loss": 0.3323,
+      "step": 7140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.617501717829695e-06,
+      "loss": 0.3385,
+      "step": 7141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614848930776514e-06,
+      "loss": 0.3395,
+      "step": 7142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.612196412750164e-06,
+      "loss": 0.3524,
+      "step": 7143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6095441639614385e-06,
+      "loss": 0.3202,
+      "step": 7144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60689218462113e-06,
+      "loss": 0.3447,
+      "step": 7145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60424047493999e-06,
+      "loss": 0.3336,
+      "step": 7146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601589035128757e-06,
+      "loss": 0.3238,
+      "step": 7147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.598937865398156e-06,
+      "loss": 0.3631,
+      "step": 7148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596286965958872e-06,
+      "loss": 0.3602,
+      "step": 7149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5936363370215865e-06,
+      "loss": 0.3673,
+      "step": 7150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.590985978796947e-06,
+      "loss": 0.3618,
+      "step": 7151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.58833589149559e-06,
+      "loss": 0.3219,
+      "step": 7152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.585686075328115e-06,
+      "loss": 0.3497,
+      "step": 7153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.58303653050512e-06,
+      "loss": 0.3443,
+      "step": 7154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.580387257237169e-06,
+      "loss": 0.3268,
+      "step": 7155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5777382557348e-06,
+      "loss": 0.3899,
+      "step": 7156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5750895262085466e-06,
+      "loss": 0.3161,
+      "step": 7157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5724410688689e-06,
+      "loss": 0.3723,
+      "step": 7158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569792883926348e-06,
+      "loss": 0.3029,
+      "step": 7159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.567144971591348e-06,
+      "loss": 0.3268,
+      "step": 7160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5644973320743285e-06,
+      "loss": 0.3349,
+      "step": 7161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561849965585716e-06,
+      "loss": 0.3435,
+      "step": 7162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.559202872335893e-06,
+      "loss": 0.405,
+      "step": 7163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.556556052535239e-06,
+      "loss": 0.3494,
+      "step": 7164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.553909506394099e-06,
+      "loss": 0.3424,
+      "step": 7165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5512632341228e-06,
+      "loss": 0.3639,
+      "step": 7166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.548617235931654e-06,
+      "loss": 0.37,
+      "step": 7167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545971512030942e-06,
+      "loss": 0.3228,
+      "step": 7168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.543326062630928e-06,
+      "loss": 0.3385,
+      "step": 7169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.540680887941846e-06,
+      "loss": 0.3202,
+      "step": 7170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.538035988173921e-06,
+      "loss": 0.3071,
+      "step": 7171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535391363537354e-06,
+      "loss": 0.3485,
+      "step": 7172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532747014242313e-06,
+      "loss": 0.3545,
+      "step": 7173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.530102940498957e-06,
+      "loss": 0.3416,
+      "step": 7174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.527459142517409e-06,
+      "loss": 0.3532,
+      "step": 7175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.524815620507785e-06,
+      "loss": 0.3379,
+      "step": 7176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.522172374680177e-06,
+      "loss": 0.3466,
+      "step": 7177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5195294052446424e-06,
+      "loss": 0.3544,
+      "step": 7178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.51688671241123e-06,
+      "loss": 0.347,
+      "step": 7179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5142442963899545e-06,
+      "loss": 0.3725,
+      "step": 7180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.511602157390826e-06,
+      "loss": 0.3337,
+      "step": 7181
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.508960295623814e-06,
+      "loss": 0.3647,
+      "step": 7182
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506318711298874e-06,
+      "loss": 0.3448,
+      "step": 7183
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503677404625948e-06,
+      "loss": 0.351,
+      "step": 7184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.501036375814938e-06,
+      "loss": 0.3404,
+      "step": 7185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498395625075742e-06,
+      "loss": 0.3556,
+      "step": 7186
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.495755152618219e-06,
+      "loss": 0.349,
+      "step": 7187
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.493114958652218e-06,
+      "loss": 0.3502,
+      "step": 7188
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.490475043387568e-06,
+      "loss": 0.3401,
+      "step": 7189
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.48783540703406e-06,
+      "loss": 0.3349,
+      "step": 7190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.485196049801477e-06,
+      "loss": 0.3551,
+      "step": 7191
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482556971899576e-06,
+      "loss": 0.3279,
+      "step": 7192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479918173538096e-06,
+      "loss": 0.3201,
+      "step": 7193
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.477279654926738e-06,
+      "loss": 0.3502,
+      "step": 7194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4746414162752e-06,
+      "loss": 0.3163,
+      "step": 7195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4720034577931525e-06,
+      "loss": 0.4047,
+      "step": 7196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.469365779690231e-06,
+      "loss": 0.3569,
+      "step": 7197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.466728382176069e-06,
+      "loss": 0.3121,
+      "step": 7198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.464091265460257e-06,
+      "loss": 0.3356,
+      "step": 7199
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.461454429752384e-06,
+      "loss": 0.377,
+      "step": 7200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458817875262002e-06,
+      "loss": 0.3231,
+      "step": 7201
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.456181602198641e-06,
+      "loss": 0.3194,
+      "step": 7202
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453545610771821e-06,
+      "loss": 0.336,
+      "step": 7203
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4509099011910235e-06,
+      "loss": 0.3408,
+      "step": 7204
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448274473665718e-06,
+      "loss": 0.3365,
+      "step": 7205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4456393284053486e-06,
+      "loss": 0.3498,
+      "step": 7206
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.443004465619336e-06,
+      "loss": 0.3468,
+      "step": 7207
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.440369885517084e-06,
+      "loss": 0.3426,
+      "step": 7208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.437735588307965e-06,
+      "loss": 0.3535,
+      "step": 7209
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.435101574201337e-06,
+      "loss": 0.3743,
+      "step": 7210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432467843406527e-06,
+      "loss": 0.3818,
+      "step": 7211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429834396132846e-06,
+      "loss": 0.3392,
+      "step": 7212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.427201232589589e-06,
+      "loss": 0.3543,
+      "step": 7213
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.424568352986009e-06,
+      "loss": 0.3266,
+      "step": 7214
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.421935757531357e-06,
+      "loss": 0.3348,
+      "step": 7215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4193034464348435e-06,
+      "loss": 0.3317,
+      "step": 7216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.416671419905671e-06,
+      "loss": 0.3722,
+      "step": 7217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.414039678153016e-06,
+      "loss": 0.3717,
+      "step": 7218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411408221386021e-06,
+      "loss": 0.3197,
+      "step": 7219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.408777049813824e-06,
+      "loss": 0.328,
+      "step": 7220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406146163645523e-06,
+      "loss": 0.386,
+      "step": 7221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.40351556309021e-06,
+      "loss": 0.3335,
+      "step": 7222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400885248356939e-06,
+      "loss": 0.3447,
+      "step": 7223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.398255219654748e-06,
+      "loss": 0.3515,
+      "step": 7224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3956254771926595e-06,
+      "loss": 0.3514,
+      "step": 7225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.392996021179656e-06,
+      "loss": 0.3539,
+      "step": 7226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.390366851824716e-06,
+      "loss": 0.3799,
+      "step": 7227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.387737969336781e-06,
+      "loss": 0.3522,
+      "step": 7228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385109373924776e-06,
+      "loss": 0.3082,
+      "step": 7229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382481065797606e-06,
+      "loss": 0.3757,
+      "step": 7230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.379853045164145e-06,
+      "loss": 0.3229,
+      "step": 7231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3772253122332526e-06,
+      "loss": 0.3569,
+      "step": 7232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.374597867213756e-06,
+      "loss": 0.3625,
+      "step": 7233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.371970710314472e-06,
+      "loss": 0.3257,
+      "step": 7234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.369343841744182e-06,
+      "loss": 0.3428,
+      "step": 7235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.366717261711653e-06,
+      "loss": 0.3445,
+      "step": 7236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.364090970425628e-06,
+      "loss": 0.3259,
+      "step": 7237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.361464968094819e-06,
+      "loss": 0.342,
+      "step": 7238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358839254927929e-06,
+      "loss": 0.3689,
+      "step": 7239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.356213831133622e-06,
+      "loss": 0.4135,
+      "step": 7240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3535886969205555e-06,
+      "loss": 0.3467,
+      "step": 7241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.350963852497353e-06,
+      "loss": 0.3373,
+      "step": 7242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348339298072612e-06,
+      "loss": 0.3475,
+      "step": 7243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.345715033854922e-06,
+      "loss": 0.3459,
+      "step": 7244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.343091060052835e-06,
+      "loss": 0.3919,
+      "step": 7245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340467376874883e-06,
+      "loss": 0.3398,
+      "step": 7246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3378439845295845e-06,
+      "loss": 0.3348,
+      "step": 7247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.335220883225418e-06,
+      "loss": 0.3755,
+      "step": 7248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.332598073170857e-06,
+      "loss": 0.3479,
+      "step": 7249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.329975554574337e-06,
+      "loss": 0.3561,
+      "step": 7250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.327353327644281e-06,
+      "loss": 0.3328,
+      "step": 7251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.324731392589076e-06,
+      "loss": 0.3529,
+      "step": 7252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.322109749617105e-06,
+      "loss": 0.3442,
+      "step": 7253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.319488398936708e-06,
+      "loss": 0.3494,
+      "step": 7254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.316867340756217e-06,
+      "loss": 0.3642,
+      "step": 7255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.314246575283931e-06,
+      "loss": 0.301,
+      "step": 7256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311626102728128e-06,
+      "loss": 0.3437,
+      "step": 7257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.309005923297064e-06,
+      "loss": 0.3614,
+      "step": 7258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.306386037198979e-06,
+      "loss": 0.3177,
+      "step": 7259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30376644464207e-06,
+      "loss": 0.3274,
+      "step": 7260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.301147145834534e-06,
+      "loss": 0.3664,
+      "step": 7261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.298528140984523e-06,
+      "loss": 0.3828,
+      "step": 7262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295909430300189e-06,
+      "loss": 0.3374,
+      "step": 7263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.293291013989635e-06,
+      "loss": 0.3585,
+      "step": 7264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290672892260959e-06,
+      "loss": 0.3491,
+      "step": 7265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.288055065322234e-06,
+      "loss": 0.3583,
+      "step": 7266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2854375333815e-06,
+      "loss": 0.3446,
+      "step": 7267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.282820296646781e-06,
+      "loss": 0.3718,
+      "step": 7268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.280203355326073e-06,
+      "loss": 0.4515,
+      "step": 7269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277586709627352e-06,
+      "loss": 0.3906,
+      "step": 7270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.274970359758576e-06,
+      "loss": 0.3622,
+      "step": 7271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.272354305927667e-06,
+      "loss": 0.3607,
+      "step": 7272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2697385483425324e-06,
+      "loss": 0.3836,
+      "step": 7273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2671230872110485e-06,
+      "loss": 0.3383,
+      "step": 7274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26450792274108e-06,
+      "loss": 0.3375,
+      "step": 7275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.261893055140453e-06,
+      "loss": 0.3421,
+      "step": 7276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259278484616984e-06,
+      "loss": 0.3438,
+      "step": 7277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.256664211378461e-06,
+      "loss": 0.3378,
+      "step": 7278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2540502356326396e-06,
+      "loss": 0.3371,
+      "step": 7279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.251436557587267e-06,
+      "loss": 0.348,
+      "step": 7280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.248823177450054e-06,
+      "loss": 0.3401,
+      "step": 7281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.246210095428693e-06,
+      "loss": 0.3563,
+      "step": 7282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243597311730858e-06,
+      "loss": 0.3702,
+      "step": 7283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240984826564185e-06,
+      "loss": 0.3579,
+      "step": 7284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238372640136305e-06,
+      "loss": 0.312,
+      "step": 7285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235760752654807e-06,
+      "loss": 0.2992,
+      "step": 7286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.233149164327265e-06,
+      "loss": 0.3539,
+      "step": 7287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230537875361238e-06,
+      "loss": 0.3594,
+      "step": 7288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2279268859642396e-06,
+      "loss": 0.3699,
+      "step": 7289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.225316196343782e-06,
+      "loss": 0.3774,
+      "step": 7290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2227058067073376e-06,
+      "loss": 0.3416,
+      "step": 7291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.220095717262365e-06,
+      "loss": 0.3672,
+      "step": 7292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217485928216287e-06,
+      "loss": 0.3359,
+      "step": 7293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214876439776519e-06,
+      "loss": 0.371,
+      "step": 7294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.212267252150443e-06,
+      "loss": 0.3403,
+      "step": 7295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209658365545411e-06,
+      "loss": 0.3287,
+      "step": 7296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.207049780168768e-06,
+      "loss": 0.3519,
+      "step": 7297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204441496227815e-06,
+      "loss": 0.3385,
+      "step": 7298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.201833513929844e-06,
+      "loss": 0.3385,
+      "step": 7299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.199225833482123e-06,
+      "loss": 0.3445,
+      "step": 7300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1966184550918826e-06,
+      "loss": 0.3705,
+      "step": 7301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.194011378966345e-06,
+      "loss": 0.3437,
+      "step": 7302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191404605312694e-06,
+      "loss": 0.387,
+      "step": 7303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.188798134338105e-06,
+      "loss": 0.3328,
+      "step": 7304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186191966249715e-06,
+      "loss": 0.36,
+      "step": 7305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183586101254644e-06,
+      "loss": 0.3522,
+      "step": 7306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.180980539559991e-06,
+      "loss": 0.332,
+      "step": 7307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.178375281372821e-06,
+      "loss": 0.3529,
+      "step": 7308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.175770326900188e-06,
+      "loss": 0.3454,
+      "step": 7309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.3354,
+      "step": 7310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.170561329926573e-06,
+      "loss": 0.3296,
+      "step": 7311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167957287839575e-06,
+      "loss": 0.3826,
+      "step": 7312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.165353550295053e-06,
+      "loss": 0.3473,
+      "step": 7313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.162750117499936e-06,
+      "loss": 0.3272,
+      "step": 7314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.16014698966112e-06,
+      "loss": 0.3067,
+      "step": 7315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1575441669854875e-06,
+      "loss": 0.3461,
+      "step": 7316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.154941649679894e-06,
+      "loss": 0.3899,
+      "step": 7317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.152339437951161e-06,
+      "loss": 0.3659,
+      "step": 7318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.149737532006101e-06,
+      "loss": 0.3714,
+      "step": 7319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.147135932051486e-06,
+      "loss": 0.3908,
+      "step": 7320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144534638294082e-06,
+      "loss": 0.3686,
+      "step": 7321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141933650940613e-06,
+      "loss": 0.3447,
+      "step": 7322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1393329701977845e-06,
+      "loss": 0.3567,
+      "step": 7323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136732596272288e-06,
+      "loss": 0.373,
+      "step": 7324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.134132529370774e-06,
+      "loss": 0.3658,
+      "step": 7325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.131532769699884e-06,
+      "loss": 0.3612,
+      "step": 7326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.128933317466222e-06,
+      "loss": 0.3599,
+      "step": 7327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.126334172876372e-06,
+      "loss": 0.336,
+      "step": 7328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1237353361369e-06,
+      "loss": 0.3544,
+      "step": 7329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12113680745434e-06,
+      "loss": 0.3645,
+      "step": 7330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118538587035206e-06,
+      "loss": 0.3958,
+      "step": 7331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115940675085979e-06,
+      "loss": 0.3249,
+      "step": 7332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113343071813132e-06,
+      "loss": 0.3181,
+      "step": 7333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.110745777423091e-06,
+      "loss": 0.362,
+      "step": 7334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.10814879212228e-06,
+      "loss": 0.319,
+      "step": 7335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105552116117086e-06,
+      "loss": 0.3237,
+      "step": 7336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1029557496138695e-06,
+      "loss": 0.346,
+      "step": 7337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100359692818976e-06,
+      "loss": 0.3621,
+      "step": 7338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.097763945938715e-06,
+      "loss": 0.3138,
+      "step": 7339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.095168509179381e-06,
+      "loss": 0.3851,
+      "step": 7340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092573382747244e-06,
+      "loss": 0.3327,
+      "step": 7341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.089978566848535e-06,
+      "loss": 0.3156,
+      "step": 7342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.087384061689483e-06,
+      "loss": 0.3489,
+      "step": 7343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.084789867476271e-06,
+      "loss": 0.3632,
+      "step": 7344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.082195984415069e-06,
+      "loss": 0.3038,
+      "step": 7345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.079602412712023e-06,
+      "loss": 0.3512,
+      "step": 7346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.077009152573244e-06,
+      "loss": 0.352,
+      "step": 7347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0744162042048335e-06,
+      "loss": 0.3127,
+      "step": 7348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.071823567812855e-06,
+      "loss": 0.3341,
+      "step": 7349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.069231243603354e-06,
+      "loss": 0.36,
+      "step": 7350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.066639231782344e-06,
+      "loss": 0.3526,
+      "step": 7351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.064047532555824e-06,
+      "loss": 0.3457,
+      "step": 7352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.061456146129765e-06,
+      "loss": 0.3589,
+      "step": 7353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.058865072710108e-06,
+      "loss": 0.3285,
+      "step": 7354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056274312502774e-06,
+      "loss": 0.3456,
+      "step": 7355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.053683865713654e-06,
+      "loss": 0.3246,
+      "step": 7356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.05109373254862e-06,
+      "loss": 0.3425,
+      "step": 7357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.048503913213519e-06,
+      "loss": 0.352,
+      "step": 7358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0459144079141665e-06,
+      "loss": 0.3341,
+      "step": 7359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0433252168563615e-06,
+      "loss": 0.3122,
+      "step": 7360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040736340245867e-06,
+      "loss": 0.3125,
+      "step": 7361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.038147778288438e-06,
+      "loss": 0.3701,
+      "step": 7362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0355595311897855e-06,
+      "loss": 0.3267,
+      "step": 7363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0329715991556055e-06,
+      "loss": 0.3987,
+      "step": 7364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0303839823915745e-06,
+      "loss": 0.385,
+      "step": 7365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.027796681103331e-06,
+      "loss": 0.3431,
+      "step": 7366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.025209695496495e-06,
+      "loss": 0.3791,
+      "step": 7367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022623025776661e-06,
+      "loss": 0.3283,
+      "step": 7368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.020036672149399e-06,
+      "loss": 0.3242,
+      "step": 7369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.017450634820256e-06,
+      "loss": 0.3478,
+      "step": 7370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0148649139947494e-06,
+      "loss": 0.396,
+      "step": 7371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0122795098783724e-06,
+      "loss": 0.3499,
+      "step": 7372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.009694422676591e-06,
+      "loss": 0.3261,
+      "step": 7373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0071096525948555e-06,
+      "loss": 0.3166,
+      "step": 7374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.004525199838578e-06,
+      "loss": 0.3562,
+      "step": 7375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.001941064613157e-06,
+      "loss": 0.3296,
+      "step": 7376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.999357247123958e-06,
+      "loss": 0.346,
+      "step": 7377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996773747576321e-06,
+      "loss": 0.3418,
+      "step": 7378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9941905661755705e-06,
+      "loss": 0.3558,
+      "step": 7379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991607703126992e-06,
+      "loss": 0.3514,
+      "step": 7380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.989025158635854e-06,
+      "loss": 0.3477,
+      "step": 7381
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.986442932907401e-06,
+      "loss": 0.3553,
+      "step": 7382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983861026146845e-06,
+      "loss": 0.3652,
+      "step": 7383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.981279438559383e-06,
+      "loss": 0.3502,
+      "step": 7384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.978698170350174e-06,
+      "loss": 0.3339,
+      "step": 7385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9761172217243605e-06,
+      "loss": 0.3633,
+      "step": 7386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.973536592887059e-06,
+      "loss": 0.3508,
+      "step": 7387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.970956284043354e-06,
+      "loss": 0.3621,
+      "step": 7388
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9683762953983185e-06,
+      "loss": 0.3181,
+      "step": 7389
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.965796627156981e-06,
+      "loss": 0.3493,
+      "step": 7390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.963217279524363e-06,
+      "loss": 0.3527,
+      "step": 7391
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960638252705442e-06,
+      "loss": 0.4061,
+      "step": 7392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.958059546905188e-06,
+      "loss": 0.3761,
+      "step": 7393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955481162328538e-06,
+      "loss": 0.3424,
+      "step": 7394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.952903099180397e-06,
+      "loss": 0.373,
+      "step": 7395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.950325357665656e-06,
+      "loss": 0.3471,
+      "step": 7396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94774793798917e-06,
+      "loss": 0.3341,
+      "step": 7397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.945170840355777e-06,
+      "loss": 0.3285,
+      "step": 7398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.942594064970287e-06,
+      "loss": 0.3594,
+      "step": 7399
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.940017612037477e-06,
+      "loss": 0.3423,
+      "step": 7400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.937441481762112e-06,
+      "loss": 0.3366,
+      "step": 7401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.934865674348915e-06,
+      "loss": 0.3243,
+      "step": 7402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932290190002603e-06,
+      "loss": 0.3453,
+      "step": 7403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.929715028927847e-06,
+      "loss": 0.3536,
+      "step": 7404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.927140191329304e-06,
+      "loss": 0.3183,
+      "step": 7405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924565677411609e-06,
+      "loss": 0.3496,
+      "step": 7406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.921991487379359e-06,
+      "loss": 0.3191,
+      "step": 7407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919417621437135e-06,
+      "loss": 0.3552,
+      "step": 7408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.916844079789483e-06,
+      "loss": 0.3694,
+      "step": 7409
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9142708626409336e-06,
+      "loss": 0.3213,
+      "step": 7410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.911697970195992e-06,
+      "loss": 0.379,
+      "step": 7411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.909125402659125e-06,
+      "loss": 0.3331,
+      "step": 7412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9065531602347855e-06,
+      "loss": 0.3304,
+      "step": 7413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9039812431273925e-06,
+      "loss": 0.3358,
+      "step": 7414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.901409651541346e-06,
+      "loss": 0.3215,
+      "step": 7415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898838385681016e-06,
+      "loss": 0.3388,
+      "step": 7416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89626744575075e-06,
+      "loss": 0.355,
+      "step": 7417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.893696831954866e-06,
+      "loss": 0.3318,
+      "step": 7418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.891126544497654e-06,
+      "loss": 0.3397,
+      "step": 7419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888556583583387e-06,
+      "loss": 0.3637,
+      "step": 7420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.885986949416303e-06,
+      "loss": 0.402,
+      "step": 7421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883417642200617e-06,
+      "loss": 0.3226,
+      "step": 7422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.880848662140524e-06,
+      "loss": 0.3483,
+      "step": 7423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878280009440178e-06,
+      "loss": 0.3228,
+      "step": 7424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.875711684303729e-06,
+      "loss": 0.3375,
+      "step": 7425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.873143686935278e-06,
+      "loss": 0.3273,
+      "step": 7426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.870576017538912e-06,
+      "loss": 0.3157,
+      "step": 7427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868008676318697e-06,
+      "loss": 0.3205,
+      "step": 7428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.865441663478661e-06,
+      "loss": 0.333,
+      "step": 7429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8628749792228105e-06,
+      "loss": 0.3733,
+      "step": 7430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.86030862375513e-06,
+      "loss": 0.3426,
+      "step": 7431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.857742597279574e-06,
+      "loss": 0.3497,
+      "step": 7432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.855176900000066e-06,
+      "loss": 0.3837,
+      "step": 7433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.852611532120515e-06,
+      "loss": 0.343,
+      "step": 7434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.850046493844798e-06,
+      "loss": 0.3433,
+      "step": 7435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.847481785376758e-06,
+      "loss": 0.3368,
+      "step": 7436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.84491740692023e-06,
+      "loss": 0.3347,
+      "step": 7437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.842353358678999e-06,
+      "loss": 0.3493,
+      "step": 7438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.839789640856847e-06,
+      "loss": 0.3575,
+      "step": 7439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.837226253657515e-06,
+      "loss": 0.4012,
+      "step": 7440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.834663197284722e-06,
+      "loss": 0.3266,
+      "step": 7441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8321004719421685e-06,
+      "loss": 0.3534,
+      "step": 7442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.829538077833503e-06,
+      "loss": 0.3285,
+      "step": 7443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.826976015162387e-06,
+      "loss": 0.3398,
+      "step": 7444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8244142841324135e-06,
+      "loss": 0.2973,
+      "step": 7445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.821852884947188e-06,
+      "loss": 0.3083,
+      "step": 7446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8192918178102656e-06,
+      "loss": 0.3515,
+      "step": 7447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816731082925172e-06,
+      "loss": 0.3487,
+      "step": 7448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.814170680495433e-06,
+      "loss": 0.3342,
+      "step": 7449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.811610610724514e-06,
+      "loss": 0.3424,
+      "step": 7450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.809050873815875e-06,
+      "loss": 0.3551,
+      "step": 7451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806491469972955e-06,
+      "loss": 0.352,
+      "step": 7452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.80393239939914e-06,
+      "loss": 0.327,
+      "step": 7453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801373662297822e-06,
+      "loss": 0.344,
+      "step": 7454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.798815258872341e-06,
+      "loss": 0.3482,
+      "step": 7455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7962571893260165e-06,
+      "loss": 0.3137,
+      "step": 7456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.793699453862161e-06,
+      "loss": 0.349,
+      "step": 7457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.791142052684023e-06,
+      "loss": 0.3511,
+      "step": 7458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.788584985994867e-06,
+      "loss": 0.3321,
+      "step": 7459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7860282539978955e-06,
+      "loss": 0.3216,
+      "step": 7460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.783471856896303e-06,
+      "loss": 0.3227,
+      "step": 7461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780915794893253e-06,
+      "loss": 0.3542,
+      "step": 7462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7783600681918835e-06,
+      "loss": 0.322,
+      "step": 7463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.775804676995304e-06,
+      "loss": 0.3131,
+      "step": 7464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7732496215065955e-06,
+      "loss": 0.3456,
+      "step": 7465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.77069490192882e-06,
+      "loss": 0.3649,
+      "step": 7466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768140518465004e-06,
+      "loss": 0.3376,
+      "step": 7467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.765586471318153e-06,
+      "loss": 0.3211,
+      "step": 7468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.763032760691243e-06,
+      "loss": 0.3683,
+      "step": 7469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.760479386787224e-06,
+      "loss": 0.3442,
+      "step": 7470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.75792634980902e-06,
+      "loss": 0.3392,
+      "step": 7471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755373649959526e-06,
+      "loss": 0.3338,
+      "step": 7472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7528212874416125e-06,
+      "loss": 0.3696,
+      "step": 7473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.750269262458123e-06,
+      "loss": 0.3262,
+      "step": 7474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.747717575211874e-06,
+      "loss": 0.344,
+      "step": 7475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.745166225905654e-06,
+      "loss": 0.3219,
+      "step": 7476
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742615214742225e-06,
+      "loss": 0.3657,
+      "step": 7477
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740064541924328e-06,
+      "loss": 0.305,
+      "step": 7478
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7375142076546575e-06,
+      "loss": 0.3723,
+      "step": 7479
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.734964212135909e-06,
+      "loss": 0.3522,
+      "step": 7480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.732414555570733e-06,
+      "loss": 0.338,
+      "step": 7481
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.729865238161756e-06,
+      "loss": 0.3553,
+      "step": 7482
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727316260111586e-06,
+      "loss": 0.3442,
+      "step": 7483
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724767621622782e-06,
+      "loss": 0.335,
+      "step": 7484
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.722219322897904e-06,
+      "loss": 0.4238,
+      "step": 7485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719671364139468e-06,
+      "loss": 0.3553,
+      "step": 7486
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7171237455499666e-06,
+      "loss": 0.3756,
+      "step": 7487
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7145764673318715e-06,
+      "loss": 0.3736,
+      "step": 7488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.712029529687606e-06,
+      "loss": 0.3658,
+      "step": 7489
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709482932819602e-06,
+      "loss": 0.3374,
+      "step": 7490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.70693667693023e-06,
+      "loss": 0.3708,
+      "step": 7491
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.704390762221847e-06,
+      "loss": 0.3519,
+      "step": 7492
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701845188896797e-06,
+      "loss": 0.3472,
+      "step": 7493
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699299957157368e-06,
+      "loss": 0.3668,
+      "step": 7494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.69675506720585e-06,
+      "loss": 0.3783,
+      "step": 7495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.694210519244481e-06,
+      "loss": 0.3502,
+      "step": 7496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.691666313475484e-06,
+      "loss": 0.3327,
+      "step": 7497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689122450101063e-06,
+      "loss": 0.3747,
+      "step": 7498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.686578929323378e-06,
+      "loss": 0.3544,
+      "step": 7499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.684035751344568e-06,
+      "loss": 0.332,
+      "step": 7500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68149291636675e-06,
+      "loss": 0.3469,
+      "step": 7501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6789504245920065e-06,
+      "loss": 0.3422,
+      "step": 7502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.676408276222401e-06,
+      "loss": 0.3448,
+      "step": 7503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.673866471459959e-06,
+      "loss": 0.3188,
+      "step": 7504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.671325010506688e-06,
+      "loss": 0.3227,
+      "step": 7505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6687838935645625e-06,
+      "loss": 0.3231,
+      "step": 7506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6662431208355326e-06,
+      "loss": 0.3434,
+      "step": 7507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6637026925215196e-06,
+      "loss": 0.3276,
+      "step": 7508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.316,
+      "step": 7509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.658622869946099e-06,
+      "loss": 0.3535,
+      "step": 7510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.656083476088396e-06,
+      "loss": 0.3505,
+      "step": 7511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653544427453126e-06,
+      "loss": 0.3247,
+      "step": 7512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651005724242072e-06,
+      "loss": 0.3434,
+      "step": 7513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.648467366656989e-06,
+      "loss": 0.3367,
+      "step": 7514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.645929354899611e-06,
+      "loss": 0.3082,
+      "step": 7515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6433916891716375e-06,
+      "loss": 0.338,
+      "step": 7516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.640854369674746e-06,
+      "loss": 0.3265,
+      "step": 7517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638317396610582e-06,
+      "loss": 0.359,
+      "step": 7518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63578077018077e-06,
+      "loss": 0.3301,
+      "step": 7519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.633244490586891e-06,
+      "loss": 0.366,
+      "step": 7520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630708558030521e-06,
+      "loss": 0.3946,
+      "step": 7521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.628172972713193e-06,
+      "loss": 0.3406,
+      "step": 7522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.625637734836417e-06,
+      "loss": 0.324,
+      "step": 7523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623102844601679e-06,
+      "loss": 0.3219,
+      "step": 7524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62056830221042e-06,
+      "loss": 0.319,
+      "step": 7525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6180341078640824e-06,
+      "loss": 0.3359,
+      "step": 7526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.615500261764062e-06,
+      "loss": 0.3419,
+      "step": 7527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612966764111718e-06,
+      "loss": 0.3714,
+      "step": 7528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610433615108413e-06,
+      "loss": 0.3566,
+      "step": 7529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607900814955445e-06,
+      "loss": 0.3249,
+      "step": 7530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605368363854118e-06,
+      "loss": 0.3375,
+      "step": 7531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60283626200568e-06,
+      "loss": 0.3684,
+      "step": 7532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.600304509611365e-06,
+      "loss": 0.3674,
+      "step": 7533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.59777310687239e-06,
+      "loss": 0.3496,
+      "step": 7534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5952420539899174e-06,
+      "loss": 0.305,
+      "step": 7535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.59271135116511e-06,
+      "loss": 0.3446,
+      "step": 7536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5901809985990795e-06,
+      "loss": 0.3523,
+      "step": 7537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587650996492917e-06,
+      "loss": 0.3383,
+      "step": 7538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.585121345047704e-06,
+      "loss": 0.3249,
+      "step": 7539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.582592044464463e-06,
+      "loss": 0.3529,
+      "step": 7540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.58006309494421e-06,
+      "loss": 0.3127,
+      "step": 7541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577534496687929e-06,
+      "loss": 0.3221,
+      "step": 7542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.575006249896571e-06,
+      "loss": 0.3536,
+      "step": 7543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572478354771062e-06,
+      "loss": 0.3453,
+      "step": 7544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.569950811512305e-06,
+      "loss": 0.328,
+      "step": 7545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567423620321166e-06,
+      "loss": 0.3598,
+      "step": 7546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.56489678139849e-06,
+      "loss": 0.3772,
+      "step": 7547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562370294945091e-06,
+      "loss": 0.3877,
+      "step": 7548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.559844161161755e-06,
+      "loss": 0.331,
+      "step": 7549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.557318380249241e-06,
+      "loss": 0.3669,
+      "step": 7550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.554792952408278e-06,
+      "loss": 0.3416,
+      "step": 7551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55226787783957e-06,
+      "loss": 0.3211,
+      "step": 7552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549743156743791e-06,
+      "loss": 0.3288,
+      "step": 7553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5472187893215865e-06,
+      "loss": 0.3535,
+      "step": 7554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5446947757735755e-06,
+      "loss": 0.3595,
+      "step": 7555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.542171116300348e-06,
+      "loss": 0.3302,
+      "step": 7556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539647811102464e-06,
+      "loss": 0.3713,
+      "step": 7557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.537124860380461e-06,
+      "loss": 0.3267,
+      "step": 7558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.534602264334841e-06,
+      "loss": 0.3229,
+      "step": 7559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532080023166085e-06,
+      "loss": 0.3321,
+      "step": 7560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529558137074633e-06,
+      "loss": 0.3481,
+      "step": 7561
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.527036606260916e-06,
+      "loss": 0.3212,
+      "step": 7562
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524515430925327e-06,
+      "loss": 0.3352,
+      "step": 7563
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521994611268218e-06,
+      "loss": 0.342,
+      "step": 7564
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.519474147489943e-06,
+      "loss": 0.3189,
+      "step": 7565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.516954039790791e-06,
+      "loss": 0.3597,
+      "step": 7566
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5144342883710554e-06,
+      "loss": 0.3478,
+      "step": 7567
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5119148934309876e-06,
+      "loss": 0.3745,
+      "step": 7568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.509395855170798e-06,
+      "loss": 0.3695,
+      "step": 7569
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506877173790697e-06,
+      "loss": 0.3216,
+      "step": 7570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.504358849490835e-06,
+      "loss": 0.348,
+      "step": 7571
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.501840882471366e-06,
+      "loss": 0.3216,
+      "step": 7572
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4993232729323865e-06,
+      "loss": 0.3384,
+      "step": 7573
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.496806021073979e-06,
+      "loss": 0.3382,
+      "step": 7574
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494289127096209e-06,
+      "loss": 0.3431,
+      "step": 7575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491772591199085e-06,
+      "loss": 0.3979,
+      "step": 7576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489256413582609e-06,
+      "loss": 0.3441,
+      "step": 7577
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486740594446748e-06,
+      "loss": 0.3609,
+      "step": 7578
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.484225133991436e-06,
+      "loss": 0.3152,
+      "step": 7579
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.481710032416597e-06,
+      "loss": 0.3189,
+      "step": 7580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4791952899221006e-06,
+      "loss": 0.3137,
+      "step": 7581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.476680906707801e-06,
+      "loss": 0.3847,
+      "step": 7582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.474166882973526e-06,
+      "loss": 0.3571,
+      "step": 7583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471653218919066e-06,
+      "loss": 0.386,
+      "step": 7584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.469139914744203e-06,
+      "loss": 0.3584,
+      "step": 7585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.466626970648661e-06,
+      "loss": 0.3461,
+      "step": 7586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.464114386832153e-06,
+      "loss": 0.3183,
+      "step": 7587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.461602163494365e-06,
+      "loss": 0.2974,
+      "step": 7588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.459090300834947e-06,
+      "loss": 0.3666,
+      "step": 7589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456578799053524e-06,
+      "loss": 0.3234,
+      "step": 7590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454067658349692e-06,
+      "loss": 0.3662,
+      "step": 7591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.451556878923018e-06,
+      "loss": 0.3309,
+      "step": 7592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.449046460973039e-06,
+      "loss": 0.3539,
+      "step": 7593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.446536404699265e-06,
+      "loss": 0.3497,
+      "step": 7594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.444026710301178e-06,
+      "loss": 0.3786,
+      "step": 7595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.441517377978229e-06,
+      "loss": 0.3253,
+      "step": 7596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43900840792984e-06,
+      "loss": 0.3356,
+      "step": 7597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.436499800355408e-06,
+      "loss": 0.3758,
+      "step": 7598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433991555454298e-06,
+      "loss": 0.3302,
+      "step": 7599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.431483673425845e-06,
+      "loss": 0.3411,
+      "step": 7600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.428976154469362e-06,
+      "loss": 0.3614,
+      "step": 7601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426468998784115e-06,
+      "loss": 0.3518,
+      "step": 7602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423962206569371e-06,
+      "loss": 0.3754,
+      "step": 7603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4214557780243446e-06,
+      "loss": 0.3503,
+      "step": 7604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4189497133482205e-06,
+      "loss": 0.3257,
+      "step": 7605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4164440127401775e-06,
+      "loss": 0.3722,
+      "step": 7606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.413938676399333e-06,
+      "loss": 0.3645,
+      "step": 7607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.411433704524807e-06,
+      "loss": 0.3463,
+      "step": 7608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.408929097315671e-06,
+      "loss": 0.3438,
+      "step": 7609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4064248549709685e-06,
+      "loss": 0.3442,
+      "step": 7610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.403920977689728e-06,
+      "loss": 0.3066,
+      "step": 7611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.40141746567093e-06,
+      "loss": 0.3387,
+      "step": 7612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398914319113537e-06,
+      "loss": 0.3467,
+      "step": 7613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396411538216484e-06,
+      "loss": 0.3413,
+      "step": 7614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.393909123178667e-06,
+      "loss": 0.3358,
+      "step": 7615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.391407074198971e-06,
+      "loss": 0.3773,
+      "step": 7616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3889053914762315e-06,
+      "loss": 0.3611,
+      "step": 7617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.386404075209264e-06,
+      "loss": 0.3236,
+      "step": 7618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.383903125596856e-06,
+      "loss": 0.3041,
+      "step": 7619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.381402542837762e-06,
+      "loss": 0.3417,
+      "step": 7620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378902327130722e-06,
+      "loss": 0.3646,
+      "step": 7621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376402478674419e-06,
+      "loss": 0.3526,
+      "step": 7622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373902997667531e-06,
+      "loss": 0.3475,
+      "step": 7623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3714038843086944e-06,
+      "loss": 0.3302,
+      "step": 7624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.368905138796523e-06,
+      "loss": 0.3483,
+      "step": 7625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3664067613295986e-06,
+      "loss": 0.3639,
+      "step": 7626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.363908752106473e-06,
+      "loss": 0.3035,
+      "step": 7627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.361411111325668e-06,
+      "loss": 0.3368,
+      "step": 7628
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35891383918568e-06,
+      "loss": 0.2974,
+      "step": 7629
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3564169358849735e-06,
+      "loss": 0.3421,
+      "step": 7630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.353920401621983e-06,
+      "loss": 0.3683,
+      "step": 7631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3514242365951156e-06,
+      "loss": 0.3751,
+      "step": 7632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.348928441002748e-06,
+      "loss": 0.3477,
+      "step": 7633
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.346433015043227e-06,
+      "loss": 0.3385,
+      "step": 7634
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.343937958914869e-06,
+      "loss": 0.3477,
+      "step": 7635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.341443272815966e-06,
+      "loss": 0.3429,
+      "step": 7636
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338948956944775e-06,
+      "loss": 0.3233,
+      "step": 7637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3364550114995275e-06,
+      "loss": 0.3699,
+      "step": 7638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333961436678422e-06,
+      "loss": 0.3739,
+      "step": 7639
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.331468232679635e-06,
+      "loss": 0.33,
+      "step": 7640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.328975399701295e-06,
+      "loss": 0.3493,
+      "step": 7641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32648293794153e-06,
+      "loss": 0.354,
+      "step": 7642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.323990847598406e-06,
+      "loss": 0.3295,
+      "step": 7643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32149912886999e-06,
+      "loss": 0.3669,
+      "step": 7644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.319007781954304e-06,
+      "loss": 0.2965,
+      "step": 7645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.31651680704933e-06,
+      "loss": 0.3572,
+      "step": 7646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.314026204353048e-06,
+      "loss": 0.3274,
+      "step": 7647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311535974063377e-06,
+      "loss": 0.3132,
+      "step": 7648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3090461163782336e-06,
+      "loss": 0.3401,
+      "step": 7649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.306556631495494e-06,
+      "loss": 0.334,
+      "step": 7650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.304067519612992e-06,
+      "loss": 0.3509,
+      "step": 7651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.30157878092856e-06,
+      "loss": 0.3776,
+      "step": 7652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299090415639973e-06,
+      "loss": 0.3538,
+      "step": 7653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.296602423944987e-06,
+      "loss": 0.3326,
+      "step": 7654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29411480604134e-06,
+      "loss": 0.3439,
+      "step": 7655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291627562126716e-06,
+      "loss": 0.3501,
+      "step": 7656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.289140692398797e-06,
+      "loss": 0.3575,
+      "step": 7657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2866541970552125e-06,
+      "loss": 0.3116,
+      "step": 7658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28416807629357e-06,
+      "loss": 0.3684,
+      "step": 7659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2816823303114505e-06,
+      "loss": 0.3687,
+      "step": 7660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.279196959306403e-06,
+      "loss": 0.3532,
+      "step": 7661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2767119634759445e-06,
+      "loss": 0.3475,
+      "step": 7662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274227343017566e-06,
+      "loss": 0.3348,
+      "step": 7663
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.271743098128728e-06,
+      "loss": 0.3628,
+      "step": 7664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269259229006856e-06,
+      "loss": 0.3386,
+      "step": 7665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.266775735849352e-06,
+      "loss": 0.3764,
+      "step": 7666
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.264292618853587e-06,
+      "loss": 0.3145,
+      "step": 7667
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2618098782168995e-06,
+      "loss": 0.3615,
+      "step": 7668
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.259327514136599e-06,
+      "loss": 0.3738,
+      "step": 7669
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256845526809964e-06,
+      "loss": 0.3092,
+      "step": 7670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.254363916434248e-06,
+      "loss": 0.3198,
+      "step": 7671
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.251882683206668e-06,
+      "loss": 0.3369,
+      "step": 7672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249401827324417e-06,
+      "loss": 0.3416,
+      "step": 7673
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2469213489846526e-06,
+      "loss": 0.3678,
+      "step": 7674
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244441248384506e-06,
+      "loss": 0.3351,
+      "step": 7675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.241961525721082e-06,
+      "loss": 0.3503,
+      "step": 7676
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239482181191438e-06,
+      "loss": 0.3258,
+      "step": 7677
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.237003214992626e-06,
+      "loss": 0.363,
+      "step": 7678
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.234524627321651e-06,
+      "loss": 0.362,
+      "step": 7679
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.232046418375494e-06,
+      "loss": 0.3687,
+      "step": 7680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2295685883511086e-06,
+      "loss": 0.3946,
+      "step": 7681
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.227091137445401e-06,
+      "loss": 0.3464,
+      "step": 7682
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224614065855279e-06,
+      "loss": 0.3697,
+      "step": 7683
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.222137373777585e-06,
+      "loss": 0.335,
+      "step": 7684
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.21966106140916e-06,
+      "loss": 0.3565,
+      "step": 7685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217185128946804e-06,
+      "loss": 0.323,
+      "step": 7686
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.214709576587273e-06,
+      "loss": 0.3091,
+      "step": 7687
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.212234404527321e-06,
+      "loss": 0.3197,
+      "step": 7688
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209759612963645e-06,
+      "loss": 0.3528,
+      "step": 7689
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.207285202092923e-06,
+      "loss": 0.3423,
+      "step": 7690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.204811172111817e-06,
+      "loss": 0.3365,
+      "step": 7691
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202337523216925e-06,
+      "loss": 0.3262,
+      "step": 7692
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.199864255604851e-06,
+      "loss": 0.3721,
+      "step": 7693
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1973913694721446e-06,
+      "loss": 0.3353,
+      "step": 7694
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.194918865015328e-06,
+      "loss": 0.3656,
+      "step": 7695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.19244674243091e-06,
+      "loss": 0.3694,
+      "step": 7696
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.189975001915342e-06,
+      "loss": 0.3245,
+      "step": 7697
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.187503643665075e-06,
+      "loss": 0.3578,
+      "step": 7698
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.185032667876502e-06,
+      "loss": 0.3053,
+      "step": 7699
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.182562074746005e-06,
+      "loss": 0.3609,
+      "step": 7700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.180091864469923e-06,
+      "loss": 0.3321,
+      "step": 7701
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.177622037244574e-06,
+      "loss": 0.384,
+      "step": 7702
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1751525932662406e-06,
+      "loss": 0.3636,
+      "step": 7703
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.172683532731175e-06,
+      "loss": 0.3536,
+      "step": 7704
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1702148558356025e-06,
+      "loss": 0.3459,
+      "step": 7705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.167746562775714e-06,
+      "loss": 0.3213,
+      "step": 7706
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16527865374767e-06,
+      "loss": 0.3339,
+      "step": 7707
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1628111289476025e-06,
+      "loss": 0.377,
+      "step": 7708
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.160343988571614e-06,
+      "loss": 0.3253,
+      "step": 7709
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157877232815771e-06,
+      "loss": 0.3958,
+      "step": 7710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.155410861876116e-06,
+      "loss": 0.3295,
+      "step": 7711
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152944875948658e-06,
+      "loss": 0.3795,
+      "step": 7712
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.150479275229372e-06,
+      "loss": 0.356,
+      "step": 7713
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148014059914209e-06,
+      "loss": 0.3387,
+      "step": 7714
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.145549230199085e-06,
+      "loss": 0.3539,
+      "step": 7715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.143084786279888e-06,
+      "loss": 0.3259,
+      "step": 7716
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.140620728352476e-06,
+      "loss": 0.3126,
+      "step": 7717
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138157056612662e-06,
+      "loss": 0.3544,
+      "step": 7718
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135693771256254e-06,
+      "loss": 0.3707,
+      "step": 7719
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133230872479009e-06,
+      "loss": 0.3629,
+      "step": 7720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130768360476663e-06,
+      "loss": 0.368,
+      "step": 7721
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.128306235444922e-06,
+      "loss": 0.3489,
+      "step": 7722
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.125844497579444e-06,
+      "loss": 0.3284,
+      "step": 7723
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.123383147075882e-06,
+      "loss": 0.372,
+      "step": 7724
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.120922184129846e-06,
+      "loss": 0.3684,
+      "step": 7725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.118461608936905e-06,
+      "loss": 0.3138,
+      "step": 7726
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.116001421692621e-06,
+      "loss": 0.3536,
+      "step": 7727
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113541622592497e-06,
+      "loss": 0.3473,
+      "step": 7728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1110822118320346e-06,
+      "loss": 0.3281,
+      "step": 7729
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.108623189606679e-06,
+      "loss": 0.3407,
+      "step": 7730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.106164556111853e-06,
+      "loss": 0.3556,
+      "step": 7731
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103706311542964e-06,
+      "loss": 0.3024,
+      "step": 7732
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.10124845609536e-06,
+      "loss": 0.3143,
+      "step": 7733
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.098790989964388e-06,
+      "loss": 0.3418,
+      "step": 7734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.096333913345339e-06,
+      "loss": 0.3254,
+      "step": 7735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.09387722643348e-06,
+      "loss": 0.3533,
+      "step": 7736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091420929424065e-06,
+      "loss": 0.3521,
+      "step": 7737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08896502251229e-06,
+      "loss": 0.3222,
+      "step": 7738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0865095058933356e-06,
+      "loss": 0.377,
+      "step": 7739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084054379762349e-06,
+      "loss": 0.3352,
+      "step": 7740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.081599644314443e-06,
+      "loss": 0.3669,
+      "step": 7741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.079145299744706e-06,
+      "loss": 0.3372,
+      "step": 7742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.076691346248188e-06,
+      "loss": 0.3373,
+      "step": 7743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.074237784019913e-06,
+      "loss": 0.3446,
+      "step": 7744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.071784613254871e-06,
+      "loss": 0.3467,
+      "step": 7745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0693318341480215e-06,
+      "loss": 0.361,
+      "step": 7746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.066879446894295e-06,
+      "loss": 0.3567,
+      "step": 7747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064427451688587e-06,
+      "loss": 0.3527,
+      "step": 7748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.061975848725765e-06,
+      "loss": 0.3063,
+      "step": 7749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059524638200665e-06,
+      "loss": 0.3271,
+      "step": 7750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.057073820308089e-06,
+      "loss": 0.3313,
+      "step": 7751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054623395242814e-06,
+      "loss": 0.372,
+      "step": 7752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.052173363199578e-06,
+      "loss": 0.3287,
+      "step": 7753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.049723724373092e-06,
+      "loss": 0.3606,
+      "step": 7754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.047274478958035e-06,
+      "loss": 0.3287,
+      "step": 7755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.044825627149057e-06,
+      "loss": 0.3749,
+      "step": 7756
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.042377169140774e-06,
+      "loss": 0.3447,
+      "step": 7757
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.039929105127775e-06,
+      "loss": 0.3373,
+      "step": 7758
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0374814353046e-06,
+      "loss": 0.3483,
+      "step": 7759
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.035034159865786e-06,
+      "loss": 0.3453,
+      "step": 7760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032587279005821e-06,
+      "loss": 0.3267,
+      "step": 7761
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.030140792919163e-06,
+      "loss": 0.351,
+      "step": 7762
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0276947018002465e-06,
+      "loss": 0.3195,
+      "step": 7763
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0252490058434555e-06,
+      "loss": 0.3224,
+      "step": 7764
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0228037052431685e-06,
+      "loss": 0.3272,
+      "step": 7765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.020358800193719e-06,
+      "loss": 0.3116,
+      "step": 7766
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0179142908894005e-06,
+      "loss": 0.3409,
+      "step": 7767
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.015470177524497e-06,
+      "loss": 0.3469,
+      "step": 7768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.013026460293235e-06,
+      "loss": 0.3402,
+      "step": 7769
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.01058313938984e-06,
+      "loss": 0.3325,
+      "step": 7770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.008140215008473e-06,
+      "loss": 0.3425,
+      "step": 7771
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.005697687343284e-06,
+      "loss": 0.3555,
+      "step": 7772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003255556588397e-06,
+      "loss": 0.3251,
+      "step": 7773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000813822937882e-06,
+      "loss": 0.345,
+      "step": 7774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998372486585796e-06,
+      "loss": 0.3135,
+      "step": 7775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.995931547726157e-06,
+      "loss": 0.3199,
+      "step": 7776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.993491006552949e-06,
+      "loss": 0.3163,
+      "step": 7777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9910508632601404e-06,
+      "loss": 0.3401,
+      "step": 7778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.988611118041644e-06,
+      "loss": 0.3471,
+      "step": 7779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.986171771091356e-06,
+      "loss": 0.3206,
+      "step": 7780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9837328226031375e-06,
+      "loss": 0.3501,
+      "step": 7781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.981294272770819e-06,
+      "loss": 0.3325,
+      "step": 7782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.978856121788198e-06,
+      "loss": 0.332,
+      "step": 7783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976418369849041e-06,
+      "loss": 0.3223,
+      "step": 7784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973981017147082e-06,
+      "loss": 0.3826,
+      "step": 7785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.971544063876026e-06,
+      "loss": 0.3745,
+      "step": 7786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969107510229541e-06,
+      "loss": 0.3693,
+      "step": 7787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966671356401268e-06,
+      "loss": 0.4165,
+      "step": 7788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.964235602584814e-06,
+      "loss": 0.3514,
+      "step": 7789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.961800248973754e-06,
+      "loss": 0.3844,
+      "step": 7790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.959365295761634e-06,
+      "loss": 0.323,
+      "step": 7791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956930743141965e-06,
+      "loss": 0.372,
+      "step": 7792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954496591308227e-06,
+      "loss": 0.3442,
+      "step": 7793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.952062840453868e-06,
+      "loss": 0.3374,
+      "step": 7794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.949629490772305e-06,
+      "loss": 0.3262,
+      "step": 7795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947196542456924e-06,
+      "loss": 0.3331,
+      "step": 7796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.944763995701075e-06,
+      "loss": 0.326,
+      "step": 7797
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9423318506980824e-06,
+      "loss": 0.3763,
+      "step": 7798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.939900107641236e-06,
+      "loss": 0.3997,
+      "step": 7799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.937468766723782e-06,
+      "loss": 0.3836,
+      "step": 7800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935037828138959e-06,
+      "loss": 0.3164,
+      "step": 7801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.932607292079956e-06,
+      "loss": 0.3299,
+      "step": 7802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9301771587399275e-06,
+      "loss": 0.3221,
+      "step": 7803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.927747428312013e-06,
+      "loss": 0.3558,
+      "step": 7804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.925318100989298e-06,
+      "loss": 0.3428,
+      "step": 7805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.922889176964858e-06,
+      "loss": 0.3158,
+      "step": 7806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9204606564317235e-06,
+      "loss": 0.3267,
+      "step": 7807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.918032539582888e-06,
+      "loss": 0.3448,
+      "step": 7808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.915604826611332e-06,
+      "loss": 0.369,
+      "step": 7809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913177517709979e-06,
+      "loss": 0.3449,
+      "step": 7810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.910750613071749e-06,
+      "loss": 0.312,
+      "step": 7811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.908324112889502e-06,
+      "loss": 0.3427,
+      "step": 7812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.905898017356079e-06,
+      "loss": 0.3235,
+      "step": 7813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903472326664298e-06,
+      "loss": 0.3449,
+      "step": 7814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.901047041006925e-06,
+      "loss": 0.3144,
+      "step": 7815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.898622160576709e-06,
+      "loss": 0.3587,
+      "step": 7816
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.89619768556636e-06,
+      "loss": 0.3473,
+      "step": 7817
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.893773616168552e-06,
+      "loss": 0.3036,
+      "step": 7818
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891349952575947e-06,
+      "loss": 0.3512,
+      "step": 7819
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.888926694981146e-06,
+      "loss": 0.3275,
+      "step": 7820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886503843576736e-06,
+      "loss": 0.3234,
+      "step": 7821
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.884081398555267e-06,
+      "loss": 0.3434,
+      "step": 7822
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881659360109257e-06,
+      "loss": 0.3206,
+      "step": 7823
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879237728431194e-06,
+      "loss": 0.3275,
+      "step": 7824
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8768165037135286e-06,
+      "loss": 0.3706,
+      "step": 7825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874395686148684e-06,
+      "loss": 0.3646,
+      "step": 7826
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871975275929047e-06,
+      "loss": 0.3932,
+      "step": 7827
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.869555273246975e-06,
+      "loss": 0.3459,
+      "step": 7828
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8671356782947924e-06,
+      "loss": 0.3565,
+      "step": 7829
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.864716491264789e-06,
+      "loss": 0.3253,
+      "step": 7830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.862297712349227e-06,
+      "loss": 0.3356,
+      "step": 7831
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.859879341740329e-06,
+      "loss": 0.3062,
+      "step": 7832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857461379630293e-06,
+      "loss": 0.3605,
+      "step": 7833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8550438262112784e-06,
+      "loss": 0.3319,
+      "step": 7834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852626681675415e-06,
+      "loss": 0.3908,
+      "step": 7835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.850209946214801e-06,
+      "loss": 0.3222,
+      "step": 7836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847793620021499e-06,
+      "loss": 0.3535,
+      "step": 7837
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.845377703287542e-06,
+      "loss": 0.3745,
+      "step": 7838
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8429621962049275e-06,
+      "loss": 0.3546,
+      "step": 7839
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840547098965628e-06,
+      "loss": 0.3653,
+      "step": 7840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.838132411761565e-06,
+      "loss": 0.3599,
+      "step": 7841
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.835718134784652e-06,
+      "loss": 0.337,
+      "step": 7842
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833304268226757e-06,
+      "loss": 0.3337,
+      "step": 7843
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830890812279707e-06,
+      "loss": 0.3096,
+      "step": 7844
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8284777671353185e-06,
+      "loss": 0.339,
+      "step": 7845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826065132985348e-06,
+      "loss": 0.3632,
+      "step": 7846
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8236529100215466e-06,
+      "loss": 0.357,
+      "step": 7847
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821241098435618e-06,
+      "loss": 0.3523,
+      "step": 7848
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.818829698419225e-06,
+      "loss": 0.3375,
+      "step": 7849
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8164187101640225e-06,
+      "loss": 0.3437,
+      "step": 7850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.814008133861608e-06,
+      "loss": 0.3318,
+      "step": 7851
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811597969703558e-06,
+      "loss": 0.3369,
+      "step": 7852
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.809188217881415e-06,
+      "loss": 0.3512,
+      "step": 7853
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8067788785866865e-06,
+      "loss": 0.3699,
+      "step": 7854
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.804369952010858e-06,
+      "loss": 0.3631,
+      "step": 7855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8019614383453625e-06,
+      "loss": 0.3737,
+      "step": 7856
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.799553337781616e-06,
+      "loss": 0.336,
+      "step": 7857
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.797145650510996e-06,
+      "loss": 0.3727,
+      "step": 7858
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794738376724842e-06,
+      "loss": 0.3079,
+      "step": 7859
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.792331516614481e-06,
+      "loss": 0.3233,
+      "step": 7860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7899250703711785e-06,
+      "loss": 0.3438,
+      "step": 7861
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.787519038186186e-06,
+      "loss": 0.349,
+      "step": 7862
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.785113420250715e-06,
+      "loss": 0.325,
+      "step": 7863
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78270821675595e-06,
+      "loss": 0.2913,
+      "step": 7864
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7803034278930375e-06,
+      "loss": 0.3361,
+      "step": 7865
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777899053853091e-06,
+      "loss": 0.3638,
+      "step": 7866
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7754950948271925e-06,
+      "loss": 0.3197,
+      "step": 7867
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773091551006394e-06,
+      "loss": 0.3462,
+      "step": 7868
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.770688422581707e-06,
+      "loss": 0.3324,
+      "step": 7869
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.768285709744117e-06,
+      "loss": 0.341,
+      "step": 7870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7658834126845745e-06,
+      "loss": 0.3962,
+      "step": 7871
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.763481531593994e-06,
+      "loss": 0.3222,
+      "step": 7872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.76108006666326e-06,
+      "loss": 0.3052,
+      "step": 7873
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758679018083225e-06,
+      "loss": 0.3395,
+      "step": 7874
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756278386044706e-06,
+      "loss": 0.3212,
+      "step": 7875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.753878170738486e-06,
+      "loss": 0.3955,
+      "step": 7876
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751478372355317e-06,
+      "loss": 0.3395,
+      "step": 7877
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7490789910859185e-06,
+      "loss": 0.3048,
+      "step": 7878
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.746680027120978e-06,
+      "loss": 0.3581,
+      "step": 7879
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7442814806511365e-06,
+      "loss": 0.3118,
+      "step": 7880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.741883351867027e-06,
+      "loss": 0.3207,
+      "step": 7881
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739485640959223e-06,
+      "loss": 0.3483,
+      "step": 7882
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737088348118285e-06,
+      "loss": 0.3551,
+      "step": 7883
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.734691473534732e-06,
+      "loss": 0.3381,
+      "step": 7884
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732295017399042e-06,
+      "loss": 0.3429,
+      "step": 7885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.72989897990168e-06,
+      "loss": 0.3649,
+      "step": 7886
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.727503361233051e-06,
+      "loss": 0.3171,
+      "step": 7887
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.725108161583552e-06,
+      "loss": 0.3301,
+      "step": 7888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722713381143536e-06,
+      "loss": 0.3061,
+      "step": 7889
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.720319020103312e-06,
+      "loss": 0.3975,
+      "step": 7890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.717925078653179e-06,
+      "loss": 0.3431,
+      "step": 7891
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.715531556983381e-06,
+      "loss": 0.3415,
+      "step": 7892
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713138455284137e-06,
+      "loss": 0.3304,
+      "step": 7893
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.710745773745644e-06,
+      "loss": 0.3604,
+      "step": 7894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7083535125580385e-06,
+      "loss": 0.3593,
+      "step": 7895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705961671911456e-06,
+      "loss": 0.327,
+      "step": 7896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.70357025199597e-06,
+      "loss": 0.3338,
+      "step": 7897
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.701179253001639e-06,
+      "loss": 0.3865,
+      "step": 7898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.69878867511848e-06,
+      "loss": 0.3224,
+      "step": 7899
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696398518536478e-06,
+      "loss": 0.3385,
+      "step": 7900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.694008783445586e-06,
+      "loss": 0.3414,
+      "step": 7901
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691619470035723e-06,
+      "loss": 0.3819,
+      "step": 7902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689230578496774e-06,
+      "loss": 0.3626,
+      "step": 7903
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686842109018589e-06,
+      "loss": 0.3908,
+      "step": 7904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684454061790987e-06,
+      "loss": 0.3131,
+      "step": 7905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.682066437003751e-06,
+      "loss": 0.3678,
+      "step": 7906
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.679679234846636e-06,
+      "loss": 0.3453,
+      "step": 7907
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6772924555093535e-06,
+      "loss": 0.3635,
+      "step": 7908
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674906099181591e-06,
+      "loss": 0.3303,
+      "step": 7909
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.672520166052997e-06,
+      "loss": 0.346,
+      "step": 7910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.670134656313189e-06,
+      "loss": 0.3262,
+      "step": 7911
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667749570151749e-06,
+      "loss": 0.3436,
+      "step": 7912
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665364907758224e-06,
+      "loss": 0.3234,
+      "step": 7913
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.662980669322133e-06,
+      "loss": 0.3404,
+      "step": 7914
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660596855032958e-06,
+      "loss": 0.3235,
+      "step": 7915
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658213465080139e-06,
+      "loss": 0.3443,
+      "step": 7916
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6558304996531e-06,
+      "loss": 0.3154,
+      "step": 7917
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653447958941217e-06,
+      "loss": 0.3382,
+      "step": 7918
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.651065843133837e-06,
+      "loss": 0.3511,
+      "step": 7919
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6486841524202784e-06,
+      "loss": 0.3772,
+      "step": 7920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6463028869898075e-06,
+      "loss": 0.3565,
+      "step": 7921
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.643922047031681e-06,
+      "loss": 0.3264,
+      "step": 7922
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.641541632735108e-06,
+      "loss": 0.3305,
+      "step": 7923
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.639161644289265e-06,
+      "loss": 0.2978,
+      "step": 7924
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.636782081883301e-06,
+      "loss": 0.3844,
+      "step": 7925
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.634402945706312e-06,
+      "loss": 0.3515,
+      "step": 7926
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.632024235947392e-06,
+      "loss": 0.3606,
+      "step": 7927
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629645952795572e-06,
+      "loss": 0.3251,
+      "step": 7928
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.627268096439858e-06,
+      "loss": 0.3441,
+      "step": 7929
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.624890667069239e-06,
+      "loss": 0.343,
+      "step": 7930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.622513664872637e-06,
+      "loss": 0.2968,
+      "step": 7931
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.620137090038977e-06,
+      "loss": 0.3376,
+      "step": 7932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617760942757117e-06,
+      "loss": 0.3358,
+      "step": 7933
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.615385223215898e-06,
+      "loss": 0.2997,
+      "step": 7934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6130099316041355e-06,
+      "loss": 0.3387,
+      "step": 7935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610635068110585e-06,
+      "loss": 0.3072,
+      "step": 7936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6082606329239985e-06,
+      "loss": 0.3844,
+      "step": 7937
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.605886626233066e-06,
+      "loss": 0.341,
+      "step": 7938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603513048226463e-06,
+      "loss": 0.3334,
+      "step": 7939
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6011398990928194e-06,
+      "loss": 0.357,
+      "step": 7940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5987671790207375e-06,
+      "loss": 0.3761,
+      "step": 7941
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596394888198785e-06,
+      "loss": 0.3516,
+      "step": 7942
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5940230268154935e-06,
+      "loss": 0.3208,
+      "step": 7943
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5916515950593595e-06,
+      "loss": 0.3757,
+      "step": 7944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5892805931188485e-06,
+      "loss": 0.3208,
+      "step": 7945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.586910021182391e-06,
+      "loss": 0.3591,
+      "step": 7946
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584539879438379e-06,
+      "loss": 0.348,
+      "step": 7947
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582170168075178e-06,
+      "loss": 0.3434,
+      "step": 7948
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579800887281112e-06,
+      "loss": 0.3397,
+      "step": 7949
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.577432037244476e-06,
+      "loss": 0.3353,
+      "step": 7950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.575063618153527e-06,
+      "loss": 0.3749,
+      "step": 7951
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.572695630196491e-06,
+      "loss": 0.3399,
+      "step": 7952
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570328073561559e-06,
+      "loss": 0.3123,
+      "step": 7953
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567960948436884e-06,
+      "loss": 0.3813,
+      "step": 7954
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.56559425501059e-06,
+      "loss": 0.3493,
+      "step": 7955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.563227993470766e-06,
+      "loss": 0.3285,
+      "step": 7956
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560862164005456e-06,
+      "loss": 0.3426,
+      "step": 7957
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558496766802689e-06,
+      "loss": 0.3249,
+      "step": 7958
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556131802050446e-06,
+      "loss": 0.3277,
+      "step": 7959
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.553767269936676e-06,
+      "loss": 0.3314,
+      "step": 7960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551403170649299e-06,
+      "loss": 0.355,
+      "step": 7961
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.549039504376184e-06,
+      "loss": 0.3707,
+      "step": 7962
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.546676271305192e-06,
+      "loss": 0.3672,
+      "step": 7963
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.544313471624131e-06,
+      "loss": 0.3319,
+      "step": 7964
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.54195110552077e-06,
+      "loss": 0.3703,
+      "step": 7965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.539589173182867e-06,
+      "loss": 0.3841,
+      "step": 7966
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.537227674798117e-06,
+      "loss": 0.3439,
+      "step": 7967
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.534866610554207e-06,
+      "loss": 0.3685,
+      "step": 7968
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532505980638768e-06,
+      "loss": 0.3324,
+      "step": 7969
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530145785239406e-06,
+      "loss": 0.3785,
+      "step": 7970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5277860245437e-06,
+      "loss": 0.3165,
+      "step": 7971
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5254266987391734e-06,
+      "loss": 0.3415,
+      "step": 7972
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.523067808013344e-06,
+      "loss": 0.3858,
+      "step": 7973
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.520709352553666e-06,
+      "loss": 0.3224,
+      "step": 7974
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518351332547573e-06,
+      "loss": 0.3306,
+      "step": 7975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.515993748182473e-06,
+      "loss": 0.33,
+      "step": 7976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.51363659964572e-06,
+      "loss": 0.3487,
+      "step": 7977
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.511279887124646e-06,
+      "loss": 0.3142,
+      "step": 7978
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508923610806544e-06,
+      "loss": 0.3576,
+      "step": 7979
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.506567770878674e-06,
+      "loss": 0.3339,
+      "step": 7980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504212367528261e-06,
+      "loss": 0.3335,
+      "step": 7981
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501857400942495e-06,
+      "loss": 0.3695,
+      "step": 7982
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.499502871308532e-06,
+      "loss": 0.3571,
+      "step": 7983
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4971487788134915e-06,
+      "loss": 0.3645,
+      "step": 7984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.494795123644461e-06,
+      "loss": 0.328,
+      "step": 7985
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.492441905988489e-06,
+      "loss": 0.3363,
+      "step": 7986
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.490089126032595e-06,
+      "loss": 0.3325,
+      "step": 7987
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.487736783963758e-06,
+      "loss": 0.3256,
+      "step": 7988
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485384879968926e-06,
+      "loss": 0.3923,
+      "step": 7989
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4830334142350105e-06,
+      "loss": 0.3372,
+      "step": 7990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48068238694889e-06,
+      "loss": 0.3521,
+      "step": 7991
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4783317982974055e-06,
+      "loss": 0.3676,
+      "step": 7992
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475981648467364e-06,
+      "loss": 0.3068,
+      "step": 7993
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47363193764554e-06,
+      "loss": 0.3456,
+      "step": 7994
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.471282666018669e-06,
+      "loss": 0.3169,
+      "step": 7995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4689338337734545e-06,
+      "loss": 0.3389,
+      "step": 7996
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466585441096569e-06,
+      "loss": 0.3394,
+      "step": 7997
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.464237488174633e-06,
+      "loss": 0.3421,
+      "step": 7998
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461889975194257e-06,
+      "loss": 0.3272,
+      "step": 7999
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.459542902341999e-06,
+      "loss": 0.3162,
+      "step": 8000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457196269804387e-06,
+      "loss": 0.3739,
+      "step": 8001
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454850077767918e-06,
+      "loss": 0.3367,
+      "step": 8002
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.452504326419041e-06,
+      "loss": 0.3836,
+      "step": 8003
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.450159015944186e-06,
+      "loss": 0.3207,
+      "step": 8004
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.447814146529746e-06,
+      "loss": 0.3463,
+      "step": 8005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.445469718362058e-06,
+      "loss": 0.377,
+      "step": 8006
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443125731627457e-06,
+      "loss": 0.3415,
+      "step": 8007
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4407821865122104e-06,
+      "loss": 0.36,
+      "step": 8008
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.438439083202582e-06,
+      "loss": 0.3456,
+      "step": 8009
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436096421884771e-06,
+      "loss": 0.3509,
+      "step": 8010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.433754202744957e-06,
+      "loss": 0.3546,
+      "step": 8011
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.431412425969293e-06,
+      "loss": 0.3212,
+      "step": 8012
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.429071091743874e-06,
+      "loss": 0.3235,
+      "step": 8013
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426730200254775e-06,
+      "loss": 0.3629,
+      "step": 8014
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.424389751688035e-06,
+      "loss": 0.3116,
+      "step": 8015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.422049746229651e-06,
+      "loss": 0.3182,
+      "step": 8016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4197101840656e-06,
+      "loss": 0.3435,
+      "step": 8017
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4173710653818024e-06,
+      "loss": 0.3319,
+      "step": 8018
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.415032390364158e-06,
+      "loss": 0.411,
+      "step": 8019
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4126941591985265e-06,
+      "loss": 0.3611,
+      "step": 8020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.410356372070734e-06,
+      "loss": 0.2947,
+      "step": 8021
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408019029166571e-06,
+      "loss": 0.3122,
+      "step": 8022
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.405682130671791e-06,
+      "loss": 0.3402,
+      "step": 8023
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403345676772115e-06,
+      "loss": 0.3167,
+      "step": 8024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.401009667653226e-06,
+      "loss": 0.3319,
+      "step": 8025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.398674103500773e-06,
+      "loss": 0.3408,
+      "step": 8026
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39633898450037e-06,
+      "loss": 0.3362,
+      "step": 8027
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394004310837594e-06,
+      "loss": 0.3701,
+      "step": 8028
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3916700826979885e-06,
+      "loss": 0.3236,
+      "step": 8029
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38933630026706e-06,
+      "loss": 0.3284,
+      "step": 8030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387002963730282e-06,
+      "loss": 0.315,
+      "step": 8031
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3846700732730875e-06,
+      "loss": 0.2896,
+      "step": 8032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38233762908088e-06,
+      "loss": 0.3244,
+      "step": 8033
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.380005631339025e-06,
+      "loss": 0.3413,
+      "step": 8034
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.377674080232851e-06,
+      "loss": 0.3212,
+      "step": 8035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.375342975947655e-06,
+      "loss": 0.3671,
+      "step": 8036
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.373012318668694e-06,
+      "loss": 0.3182,
+      "step": 8037
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.370682108581194e-06,
+      "loss": 0.3268,
+      "step": 8038
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368352345870334e-06,
+      "loss": 0.3198,
+      "step": 8039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366023030721277e-06,
+      "loss": 0.3435,
+      "step": 8040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3636941633191395e-06,
+      "loss": 0.3293,
+      "step": 8041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361365743848992e-06,
+      "loss": 0.3392,
+      "step": 8042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359037772495893e-06,
+      "loss": 0.3232,
+      "step": 8043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356710249444841e-06,
+      "loss": 0.381,
+      "step": 8044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.354383174880819e-06,
+      "loss": 0.3233,
+      "step": 8045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.352056548988766e-06,
+      "loss": 0.3454,
+      "step": 8046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349730371953575e-06,
+      "loss": 0.3393,
+      "step": 8047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347404643960127e-06,
+      "loss": 0.3428,
+      "step": 8048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.345079365193238e-06,
+      "loss": 0.3385,
+      "step": 8049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342754535837723e-06,
+      "loss": 0.3502,
+      "step": 8050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3404301560783265e-06,
+      "loss": 0.3694,
+      "step": 8051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.338106226099777e-06,
+      "loss": 0.3455,
+      "step": 8052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.33578274608677e-06,
+      "loss": 0.328,
+      "step": 8053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333459716223951e-06,
+      "loss": 0.3679,
+      "step": 8054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.33113713669594e-06,
+      "loss": 0.3482,
+      "step": 8055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328815007687317e-06,
+      "loss": 0.3426,
+      "step": 8056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.326493329382626e-06,
+      "loss": 0.3109,
+      "step": 8057
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324172101966387e-06,
+      "loss": 0.3438,
+      "step": 8058
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3218513256230635e-06,
+      "loss": 0.3351,
+      "step": 8059
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319531000537097e-06,
+      "loss": 0.3697,
+      "step": 8060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3172111268928895e-06,
+      "loss": 0.3561,
+      "step": 8061
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3148917048748075e-06,
+      "loss": 0.3227,
+      "step": 8062
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312572734667182e-06,
+      "loss": 0.3088,
+      "step": 8063
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310254216454308e-06,
+      "loss": 0.3176,
+      "step": 8064
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307936150420443e-06,
+      "loss": 0.3298,
+      "step": 8065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305618536749811e-06,
+      "loss": 0.3261,
+      "step": 8066
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.303301375626597e-06,
+      "loss": 0.3284,
+      "step": 8067
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.300984667234955e-06,
+      "loss": 0.362,
+      "step": 8068
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2986684117589965e-06,
+      "loss": 0.3331,
+      "step": 8069
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2963526093828026e-06,
+      "loss": 0.3379,
+      "step": 8070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.294037260290415e-06,
+      "loss": 0.3556,
+      "step": 8071
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.291722364665841e-06,
+      "loss": 0.3427,
+      "step": 8072
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.289407922693053e-06,
+      "loss": 0.356,
+      "step": 8073
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2870939345559825e-06,
+      "loss": 0.3248,
+      "step": 8074
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284780400438531e-06,
+      "loss": 0.36,
+      "step": 8075
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282467320524559e-06,
+      "loss": 0.3103,
+      "step": 8076
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.280154694997899e-06,
+      "loss": 0.3359,
+      "step": 8077
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277842524042328e-06,
+      "loss": 0.3989,
+      "step": 8078
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275530807841618e-06,
+      "loss": 0.3702,
+      "step": 8079
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.273219546579469e-06,
+      "loss": 0.3798,
+      "step": 8080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.270908740439575e-06,
+      "loss": 0.357,
+      "step": 8081
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268598389605584e-06,
+      "loss": 0.3444,
+      "step": 8082
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.266288494261093e-06,
+      "loss": 0.3286,
+      "step": 8083
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.26397905458969e-06,
+      "loss": 0.3512,
+      "step": 8084
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.261670070774897e-06,
+      "loss": 0.3436,
+      "step": 8085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259361543000226e-06,
+      "loss": 0.35,
+      "step": 8086
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257053471449144e-06,
+      "loss": 0.3851,
+      "step": 8087
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254745856305067e-06,
+      "loss": 0.3438,
+      "step": 8088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252438697751401e-06,
+      "loss": 0.4055,
+      "step": 8089
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250131995971491e-06,
+      "loss": 0.3154,
+      "step": 8090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247825751148663e-06,
+      "loss": 0.3411,
+      "step": 8091
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.245519963466198e-06,
+      "loss": 0.3471,
+      "step": 8092
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.243214633107339e-06,
+      "loss": 0.3298,
+      "step": 8093
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240909760255309e-06,
+      "loss": 0.3438,
+      "step": 8094
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.238605345093271e-06,
+      "loss": 0.3389,
+      "step": 8095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.236301387804365e-06,
+      "loss": 0.3554,
+      "step": 8096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2339978885716934e-06,
+      "loss": 0.3386,
+      "step": 8097
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2316948475783195e-06,
+      "loss": 0.3276,
+      "step": 8098
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.229392265007282e-06,
+      "loss": 0.3336,
+      "step": 8099
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.22709014104156e-06,
+      "loss": 0.3955,
+      "step": 8100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2247884758641155e-06,
+      "loss": 0.3353,
+      "step": 8101
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222487269657867e-06,
+      "loss": 0.3599,
+      "step": 8102
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220186522605697e-06,
+      "loss": 0.3344,
+      "step": 8103
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.217886234890452e-06,
+      "loss": 0.3461,
+      "step": 8104
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.215586406694944e-06,
+      "loss": 0.3631,
+      "step": 8105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.213287038201943e-06,
+      "loss": 0.3312,
+      "step": 8106
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210988129594189e-06,
+      "loss": 0.3525,
+      "step": 8107
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.20868968105438e-06,
+      "loss": 0.3262,
+      "step": 8108
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.206391692765179e-06,
+      "loss": 0.3232,
+      "step": 8109
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.204094164909217e-06,
+      "loss": 0.3489,
+      "step": 8110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.20179709766908e-06,
+      "loss": 0.3517,
+      "step": 8111
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.199500491227326e-06,
+      "loss": 0.3375,
+      "step": 8112
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197204345766469e-06,
+      "loss": 0.3451,
+      "step": 8113
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.194908661468993e-06,
+      "loss": 0.3594,
+      "step": 8114
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192613438517338e-06,
+      "loss": 0.35,
+      "step": 8115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190318677093915e-06,
+      "loss": 0.3917,
+      "step": 8116
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188024377381092e-06,
+      "loss": 0.3257,
+      "step": 8117
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1857305395612105e-06,
+      "loss": 0.3647,
+      "step": 8118
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1834371638165515e-06,
+      "loss": 0.379,
+      "step": 8119
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181144250329395e-06,
+      "loss": 0.371,
+      "step": 8120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.178851799281948e-06,
+      "loss": 0.3317,
+      "step": 8121
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1765598108564076e-06,
+      "loss": 0.3217,
+      "step": 8122
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1742682852349255e-06,
+      "loss": 0.4021,
+      "step": 8123
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171977222599605e-06,
+      "loss": 0.3343,
+      "step": 8124
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1696866231325365e-06,
+      "loss": 0.3344,
+      "step": 8125
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.167396487015748e-06,
+      "loss": 0.3584,
+      "step": 8126
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1651068144312445e-06,
+      "loss": 0.3401,
+      "step": 8127
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.162817605561002e-06,
+      "loss": 0.3649,
+      "step": 8128
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1605288605869365e-06,
+      "loss": 0.4216,
+      "step": 8129
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.158240579690954e-06,
+      "loss": 0.3616,
+      "step": 8130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1559527630549e-06,
+      "loss": 0.3469,
+      "step": 8131
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1536654108605925e-06,
+      "loss": 0.3516,
+      "step": 8132
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151378523289824e-06,
+      "loss": 0.3423,
+      "step": 8133
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1490921005243265e-06,
+      "loss": 0.3316,
+      "step": 8134
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.146806142745822e-06,
+      "loss": 0.3453,
+      "step": 8135
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144520650135969e-06,
+      "loss": 0.3229,
+      "step": 8136
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142235622876407e-06,
+      "loss": 0.3428,
+      "step": 8137
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1399510611487335e-06,
+      "loss": 0.3236,
+      "step": 8138
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137666965134507e-06,
+      "loss": 0.3655,
+      "step": 8139
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135383335015251e-06,
+      "loss": 0.3336,
+      "step": 8140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.133100170972454e-06,
+      "loss": 0.3344,
+      "step": 8141
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130817473187561e-06,
+      "loss": 0.3286,
+      "step": 8142
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.128535241841987e-06,
+      "loss": 0.3713,
+      "step": 8143
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126253477117105e-06,
+      "loss": 0.3665,
+      "step": 8144
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.123972179194253e-06,
+      "loss": 0.3575,
+      "step": 8145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1216913482547335e-06,
+      "loss": 0.3456,
+      "step": 8146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119410984479809e-06,
+      "loss": 0.3038,
+      "step": 8147
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117131088050705e-06,
+      "loss": 0.3678,
+      "step": 8148
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.114851659148613e-06,
+      "loss": 0.3467,
+      "step": 8149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112572697954683e-06,
+      "loss": 0.3395,
+      "step": 8150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.110294204650032e-06,
+      "loss": 0.3349,
+      "step": 8151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.108016179415737e-06,
+      "loss": 0.3344,
+      "step": 8152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105738622432838e-06,
+      "loss": 0.3445,
+      "step": 8153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103461533882341e-06,
+      "loss": 0.3372,
+      "step": 8154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.101184913945205e-06,
+      "loss": 0.3471,
+      "step": 8155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.098908762802367e-06,
+      "loss": 0.3428,
+      "step": 8156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096633080634717e-06,
+      "loss": 0.3258,
+      "step": 8157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.094357867623105e-06,
+      "loss": 0.3387,
+      "step": 8158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0920831239483585e-06,
+      "loss": 0.3337,
+      "step": 8159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089808849791241e-06,
+      "loss": 0.3844,
+      "step": 8160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087535045332508e-06,
+      "loss": 0.3866,
+      "step": 8161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.085261710752861e-06,
+      "loss": 0.3664,
+      "step": 8162
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082988846232968e-06,
+      "loss": 0.3464,
+      "step": 8163
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080716451953463e-06,
+      "loss": 0.3371,
+      "step": 8164
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078444528094928e-06,
+      "loss": 0.3517,
+      "step": 8165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.076173074837934e-06,
+      "loss": 0.3803,
+      "step": 8166
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0739020923629854e-06,
+      "loss": 0.2958,
+      "step": 8167
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.071631580850568e-06,
+      "loss": 0.3313,
+      "step": 8168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0693615404811335e-06,
+      "loss": 0.3692,
+      "step": 8169
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067091971435074e-06,
+      "loss": 0.3561,
+      "step": 8170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064822873892771e-06,
+      "loss": 0.3965,
+      "step": 8171
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.062554248034546e-06,
+      "loss": 0.3588,
+      "step": 8172
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060286094040694e-06,
+      "loss": 0.317,
+      "step": 8173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05801841209148e-06,
+      "loss": 0.3463,
+      "step": 8174
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.055751202367113e-06,
+      "loss": 0.3402,
+      "step": 8175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.053484465047777e-06,
+      "loss": 0.3686,
+      "step": 8176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051218200313615e-06,
+      "loss": 0.32,
+      "step": 8177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048952408344734e-06,
+      "loss": 0.3303,
+      "step": 8178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046687089321203e-06,
+      "loss": 0.3583,
+      "step": 8179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0444222434230515e-06,
+      "loss": 0.3567,
+      "step": 8180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0421578708302734e-06,
+      "loss": 0.3677,
+      "step": 8181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039893971722824e-06,
+      "loss": 0.3396,
+      "step": 8182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.037630546280622e-06,
+      "loss": 0.3291,
+      "step": 8183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.035367594683547e-06,
+      "loss": 0.3347,
+      "step": 8184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033105117111441e-06,
+      "loss": 0.3499,
+      "step": 8185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030843113744112e-06,
+      "loss": 0.4063,
+      "step": 8186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.028581584761323e-06,
+      "loss": 0.3424,
+      "step": 8187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026320530342807e-06,
+      "loss": 0.3279,
+      "step": 8188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0240599506682556e-06,
+      "loss": 0.3796,
+      "step": 8189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.021799845917321e-06,
+      "loss": 0.3743,
+      "step": 8190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019540216269622e-06,
+      "loss": 0.3339,
+      "step": 8191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017281061904736e-06,
+      "loss": 0.3268,
+      "step": 8192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015022383002203e-06,
+      "loss": 0.3512,
+      "step": 8193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.012764179741527e-06,
+      "loss": 0.3288,
+      "step": 8194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010506452302178e-06,
+      "loss": 0.3406,
+      "step": 8195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.008249200863572e-06,
+      "loss": 0.3338,
+      "step": 8196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0059924256051095e-06,
+      "loss": 0.3245,
+      "step": 8197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003736126706138e-06,
+      "loss": 0.3414,
+      "step": 8198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0014803043459726e-06,
+      "loss": 0.339,
+      "step": 8199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999224958703892e-06,
+      "loss": 0.3325,
+      "step": 8200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.996970089959125e-06,
+      "loss": 0.3183,
+      "step": 8201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994715698290882e-06,
+      "loss": 0.3144,
+      "step": 8202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.992461783878324e-06,
+      "loss": 0.3544,
+      "step": 8203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990208346900567e-06,
+      "loss": 0.3275,
+      "step": 8204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.98795538753671e-06,
+      "loss": 0.3216,
+      "step": 8205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985702905965789e-06,
+      "loss": 0.3377,
+      "step": 8206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983450902366828e-06,
+      "loss": 0.3432,
+      "step": 8207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.981199376918788e-06,
+      "loss": 0.3487,
+      "step": 8208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978948329800606e-06,
+      "loss": 0.3547,
+      "step": 8209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.976697761191187e-06,
+      "loss": 0.3633,
+      "step": 8210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974447671269377e-06,
+      "loss": 0.3737,
+      "step": 8211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972198060214009e-06,
+      "loss": 0.342,
+      "step": 8212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.969948928203856e-06,
+      "loss": 0.3946,
+      "step": 8213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.967700275417663e-06,
+      "loss": 0.3308,
+      "step": 8214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.965452102034146e-06,
+      "loss": 0.3458,
+      "step": 8215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.963204408231962e-06,
+      "loss": 0.3227,
+      "step": 8216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960957194189747e-06,
+      "loss": 0.3841,
+      "step": 8217
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.95871046008609e-06,
+      "loss": 0.3441,
+      "step": 8218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.956464206099548e-06,
+      "loss": 0.3478,
+      "step": 8219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954218432408633e-06,
+      "loss": 0.3455,
+      "step": 8220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9519731391918255e-06,
+      "loss": 0.3568,
+      "step": 8221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949728326627564e-06,
+      "loss": 0.3357,
+      "step": 8222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.94748399489425e-06,
+      "loss": 0.3664,
+      "step": 8223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945240144170245e-06,
+      "loss": 0.3464,
+      "step": 8224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942996774633875e-06,
+      "loss": 0.3271,
+      "step": 8225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940753886463426e-06,
+      "loss": 0.3187,
+      "step": 8226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.938511479837147e-06,
+      "loss": 0.3278,
+      "step": 8227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.936269554933247e-06,
+      "loss": 0.3659,
+      "step": 8228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.934028111929897e-06,
+      "loss": 0.3701,
+      "step": 8229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931787151005232e-06,
+      "loss": 0.391,
+      "step": 8230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.929546672337345e-06,
+      "loss": 0.3149,
+      "step": 8231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.927306676104296e-06,
+      "loss": 0.3459,
+      "step": 8232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9250671624841e-06,
+      "loss": 0.3323,
+      "step": 8233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922828131654739e-06,
+      "loss": 0.3492,
+      "step": 8234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920589583794153e-06,
+      "loss": 0.3427,
+      "step": 8235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.918351519080251e-06,
+      "loss": 0.3381,
+      "step": 8236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916113937690886e-06,
+      "loss": 0.3315,
+      "step": 8237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913876839803895e-06,
+      "loss": 0.3505,
+      "step": 8238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911640225597062e-06,
+      "loss": 0.3466,
+      "step": 8239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909404095248139e-06,
+      "loss": 0.3353,
+      "step": 8240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907168448934836e-06,
+      "loss": 0.3305,
+      "step": 8241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90493328683482e-06,
+      "loss": 0.315,
+      "step": 8242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.902698609125734e-06,
+      "loss": 0.3367,
+      "step": 8243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9004644159851725e-06,
+      "loss": 0.3638,
+      "step": 8244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8982307075906826e-06,
+      "loss": 0.3356,
+      "step": 8245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895997484119798e-06,
+      "loss": 0.3077,
+      "step": 8246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893764745749985e-06,
+      "loss": 0.3555,
+      "step": 8247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.891532492658697e-06,
+      "loss": 0.3238,
+      "step": 8248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.889300725023329e-06,
+      "loss": 0.3397,
+      "step": 8249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.887069443021245e-06,
+      "loss": 0.3361,
+      "step": 8250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.884838646829781e-06,
+      "loss": 0.3276,
+      "step": 8251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.882608336626211e-06,
+      "loss": 0.3418,
+      "step": 8252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.880378512587791e-06,
+      "loss": 0.3079,
+      "step": 8253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.87814917489173e-06,
+      "loss": 0.3621,
+      "step": 8254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875920323715195e-06,
+      "loss": 0.3591,
+      "step": 8255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8736919592353295e-06,
+      "loss": 0.3407,
+      "step": 8256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.871464081629217e-06,
+      "loss": 0.3483,
+      "step": 8257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.869236691073915e-06,
+      "loss": 0.3485,
+      "step": 8258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.867009787746443e-06,
+      "loss": 0.3339,
+      "step": 8259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.864783371823776e-06,
+      "loss": 0.3441,
+      "step": 8260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862557443482854e-06,
+      "loss": 0.3493,
+      "step": 8261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.860332002900579e-06,
+      "loss": 0.3677,
+      "step": 8262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8581070502538095e-06,
+      "loss": 0.3556,
+      "step": 8263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.85588258571937e-06,
+      "loss": 0.3319,
+      "step": 8264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8536586094740455e-06,
+      "loss": 0.3609,
+      "step": 8265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.85143512169458e-06,
+      "loss": 0.3489,
+      "step": 8266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.849212122557681e-06,
+      "loss": 0.3546,
+      "step": 8267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846989612240014e-06,
+      "loss": 0.3449,
+      "step": 8268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8447675909182095e-06,
+      "loss": 0.3512,
+      "step": 8269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842546058768855e-06,
+      "loss": 0.3528,
+      "step": 8270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840325015968506e-06,
+      "loss": 0.3415,
+      "step": 8271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83810446269367e-06,
+      "loss": 0.3238,
+      "step": 8272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.835884399120823e-06,
+      "loss": 0.3366,
+      "step": 8273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.833664825426397e-06,
+      "loss": 0.3338,
+      "step": 8274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83144574178679e-06,
+      "loss": 0.3426,
+      "step": 8275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.829227148378356e-06,
+      "loss": 0.3372,
+      "step": 8276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827009045377418e-06,
+      "loss": 0.3585,
+      "step": 8277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824791432960242e-06,
+      "loss": 0.3586,
+      "step": 8278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.82257431130308e-06,
+      "loss": 0.3492,
+      "step": 8279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820357680582131e-06,
+      "loss": 0.3403,
+      "step": 8280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818141540973545e-06,
+      "loss": 0.3607,
+      "step": 8281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815925892653462e-06,
+      "loss": 0.389,
+      "step": 8282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8137107357979475e-06,
+      "loss": 0.3469,
+      "step": 8283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8114960705830583e-06,
+      "loss": 0.3904,
+      "step": 8284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8092818971848e-06,
+      "loss": 0.3671,
+      "step": 8285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8070682157791263e-06,
+      "loss": 0.3512,
+      "step": 8286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8048550265419803e-06,
+      "loss": 0.3459,
+      "step": 8287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8026423296492354e-06,
+      "loss": 0.3387,
+      "step": 8288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800430125276754e-06,
+      "loss": 0.3193,
+      "step": 8289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.798218413600335e-06,
+      "loss": 0.3423,
+      "step": 8290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79600719479575e-06,
+      "loss": 0.3616,
+      "step": 8291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7937964690387397e-06,
+      "loss": 0.3486,
+      "step": 8292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.791586236504987e-06,
+      "loss": 0.3487,
+      "step": 8293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7893764973701473e-06,
+      "loss": 0.3283,
+      "step": 8294
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7871672518098345e-06,
+      "loss": 0.3176,
+      "step": 8295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.78495849999962e-06,
+      "loss": 0.323,
+      "step": 8296
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7827502421150497e-06,
+      "loss": 0.3179,
+      "step": 8297
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.780542478331609e-06,
+      "loss": 0.3243,
+      "step": 8298
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7783352088247582e-06,
+      "loss": 0.3517,
+      "step": 8299
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7761284337699143e-06,
+      "loss": 0.3373,
+      "step": 8300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7739221533424565e-06,
+      "loss": 0.3721,
+      "step": 8301
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7717163677177215e-06,
+      "loss": 0.3566,
+      "step": 8302
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7695110770710118e-06,
+      "loss": 0.3451,
+      "step": 8303
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7673062815775864e-06,
+      "loss": 0.3266,
+      "step": 8304
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.3457,
+      "step": 8305
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7628981767514315e-06,
+      "loss": 0.3278,
+      "step": 8306
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.760694867769026e-06,
+      "loss": 0.3214,
+      "step": 8307
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7584920546405524e-06,
+      "loss": 0.3846,
+      "step": 8308
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7562897375410722e-06,
+      "loss": 0.3673,
+      "step": 8309
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7540879166456123e-06,
+      "loss": 0.3614,
+      "step": 8310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.751886592129155e-06,
+      "loss": 0.3284,
+      "step": 8311
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.749685764166645e-06,
+      "loss": 0.3148,
+      "step": 8312
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.74748543293299e-06,
+      "loss": 0.3492,
+      "step": 8313
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7452855986030544e-06,
+      "loss": 0.3543,
+      "step": 8314
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.743086261351665e-06,
+      "loss": 0.3411,
+      "step": 8315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.740887421353613e-06,
+      "loss": 0.3232,
+      "step": 8316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.738689078783635e-06,
+      "loss": 0.3797,
+      "step": 8317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.736491233816453e-06,
+      "loss": 0.3294,
+      "step": 8318
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.734293886626722e-06,
+      "loss": 0.3298,
+      "step": 8319
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7320970373890807e-06,
+      "loss": 0.3479,
+      "step": 8320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7299006862781184e-06,
+      "loss": 0.3252,
+      "step": 8321
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7277048334683753e-06,
+      "loss": 0.3237,
+      "step": 8322
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.725509479134376e-06,
+      "loss": 0.3264,
+      "step": 8323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7233146234505747e-06,
+      "loss": 0.3365,
+      "step": 8324
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.721120266591416e-06,
+      "loss": 0.3433,
+      "step": 8325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.718926408731288e-06,
+      "loss": 0.3287,
+      "step": 8326
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.716733050044534e-06,
+      "loss": 0.3543,
+      "step": 8327
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.714540190705479e-06,
+      "loss": 0.309,
+      "step": 8328
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.712347830888385e-06,
+      "loss": 0.33,
+      "step": 8329
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.710155970767485e-06,
+      "loss": 0.3779,
+      "step": 8330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.707964610516982e-06,
+      "loss": 0.3038,
+      "step": 8331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7057737503110157e-06,
+      "loss": 0.3647,
+      "step": 8332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7035833903237115e-06,
+      "loss": 0.3607,
+      "step": 8333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.701393530729136e-06,
+      "loss": 0.3657,
+      "step": 8334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6992041717013226e-06,
+      "loss": 0.3011,
+      "step": 8335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6970153134142685e-06,
+      "loss": 0.3307,
+      "step": 8336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6948269560419236e-06,
+      "loss": 0.3355,
+      "step": 8337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6926390997582127e-06,
+      "loss": 0.3417,
+      "step": 8338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.690451744736999e-06,
+      "loss": 0.33,
+      "step": 8339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6882648911521226e-06,
+      "loss": 0.3184,
+      "step": 8340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.686078539177378e-06,
+      "loss": 0.3488,
+      "step": 8341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.683892688986519e-06,
+      "loss": 0.3285,
+      "step": 8342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6817073407532632e-06,
+      "loss": 0.3215,
+      "step": 8343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.679522494651283e-06,
+      "loss": 0.3368,
+      "step": 8344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.677338150854217e-06,
+      "loss": 0.3534,
+      "step": 8345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.675154309535659e-06,
+      "loss": 0.3573,
+      "step": 8346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.672970970869164e-06,
+      "loss": 0.3591,
+      "step": 8347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.670788135028248e-06,
+      "loss": 0.3354,
+      "step": 8348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.668605802186389e-06,
+      "loss": 0.3289,
+      "step": 8349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.666423972517019e-06,
+      "loss": 0.3158,
+      "step": 8350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6642426461935363e-06,
+      "loss": 0.3235,
+      "step": 8351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6620618233892957e-06,
+      "loss": 0.333,
+      "step": 8352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6598815042776135e-06,
+      "loss": 0.3432,
+      "step": 8353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6577016890317652e-06,
+      "loss": 0.36,
+      "step": 8354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.655522377824985e-06,
+      "loss": 0.3787,
+      "step": 8355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.653343570830471e-06,
+      "loss": 0.3273,
+      "step": 8356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6511652682213794e-06,
+      "loss": 0.3348,
+      "step": 8357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.648987470170817e-06,
+      "loss": 0.3143,
+      "step": 8358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6468101768518738e-06,
+      "loss": 0.3673,
+      "step": 8359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6446333884375694e-06,
+      "loss": 0.34,
+      "step": 8360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6424571051009096e-06,
+      "loss": 0.3371,
+      "step": 8361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6402813270148496e-06,
+      "loss": 0.3431,
+      "step": 8362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.638106054352294e-06,
+      "loss": 0.3582,
+      "step": 8363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6359312872861297e-06,
+      "loss": 0.3353,
+      "step": 8364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6337570259891842e-06,
+      "loss": 0.3092,
+      "step": 8365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.631583270634248e-06,
+      "loss": 0.3437,
+      "step": 8366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6294100213940873e-06,
+      "loss": 0.365,
+      "step": 8367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6272372784414034e-06,
+      "loss": 0.3537,
+      "step": 8368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.625065041948881e-06,
+      "loss": 0.3285,
+      "step": 8369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.622893312089145e-06,
+      "loss": 0.3626,
+      "step": 8370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.620722089034787e-06,
+      "loss": 0.3633,
+      "step": 8371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.618551372958371e-06,
+      "loss": 0.3295,
+      "step": 8372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6163811640323965e-06,
+      "loss": 0.3086,
+      "step": 8373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.614211462429348e-06,
+      "loss": 0.3175,
+      "step": 8374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.612042268321647e-06,
+      "loss": 0.3545,
+      "step": 8375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.60987358188169e-06,
+      "loss": 0.37,
+      "step": 8376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.607705403281827e-06,
+      "loss": 0.3528,
+      "step": 8377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6055377326943698e-06,
+      "loss": 0.3217,
+      "step": 8378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6033705702915867e-06,
+      "loss": 0.3455,
+      "step": 8379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.60120391624571e-06,
+      "loss": 0.3798,
+      "step": 8380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5990377707289292e-06,
+      "loss": 0.3517,
+      "step": 8381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5968721339133927e-06,
+      "loss": 0.3945,
+      "step": 8382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5947070059712096e-06,
+      "loss": 0.341,
+      "step": 8383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.592542387074448e-06,
+      "loss": 0.3148,
+      "step": 8384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.590378277395137e-06,
+      "loss": 0.3324,
+      "step": 8385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.588214677105263e-06,
+      "loss": 0.3492,
+      "step": 8386
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.586051586376773e-06,
+      "loss": 0.3232,
+      "step": 8387
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5838890053815734e-06,
+      "loss": 0.3642,
+      "step": 8388
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5817269342915306e-06,
+      "loss": 0.3761,
+      "step": 8389
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5795653732784706e-06,
+      "loss": 0.3554,
+      "step": 8390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5774043225141774e-06,
+      "loss": 0.374,
+      "step": 8391
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5752437821703954e-06,
+      "loss": 0.3508,
+      "step": 8392
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.573083752418832e-06,
+      "loss": 0.3658,
+      "step": 8393
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.570924233431141e-06,
+      "loss": 0.3081,
+      "step": 8394
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.568765225378954e-06,
+      "loss": 0.3165,
+      "step": 8395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.56660672843385e-06,
+      "loss": 0.3774,
+      "step": 8396
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.56444874276737e-06,
+      "loss": 0.3843,
+      "step": 8397
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5622912685510193e-06,
+      "loss": 0.3393,
+      "step": 8398
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5601343059562475e-06,
+      "loss": 0.3336,
+      "step": 8399
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5579778551544843e-06,
+      "loss": 0.3322,
+      "step": 8400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5558219163171035e-06,
+      "loss": 0.3218,
+      "step": 8401
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.553666489615446e-06,
+      "loss": 0.3579,
+      "step": 8402
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.551511575220811e-06,
+      "loss": 0.3488,
+      "step": 8403
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.549357173304444e-06,
+      "loss": 0.384,
+      "step": 8404
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.547203284037577e-06,
+      "loss": 0.3273,
+      "step": 8405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5450499075913724e-06,
+      "loss": 0.3729,
+      "step": 8406
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.542897044136967e-06,
+      "loss": 0.3505,
+      "step": 8407
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540744693845464e-06,
+      "loss": 0.356,
+      "step": 8408
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5385928568879012e-06,
+      "loss": 0.3561,
+      "step": 8409
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5364415334353062e-06,
+      "loss": 0.3193,
+      "step": 8410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.53429072365864e-06,
+      "loss": 0.3302,
+      "step": 8411
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.532140427728832e-06,
+      "loss": 0.3365,
+      "step": 8412
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.529990645816782e-06,
+      "loss": 0.3365,
+      "step": 8413
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5278413780933298e-06,
+      "loss": 0.3386,
+      "step": 8414
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.525692624729287e-06,
+      "loss": 0.3509,
+      "step": 8415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5235443858954186e-06,
+      "loss": 0.3207,
+      "step": 8416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.521396661762452e-06,
+      "loss": 0.3428,
+      "step": 8417
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5192494525010724e-06,
+      "loss": 0.3336,
+      "step": 8418
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.517102758281925e-06,
+      "loss": 0.3367,
+      "step": 8419
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5149565792756113e-06,
+      "loss": 0.3589,
+      "step": 8420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5128109156526956e-06,
+      "loss": 0.3382,
+      "step": 8421
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5106657675836986e-06,
+      "loss": 0.324,
+      "step": 8422
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5085211352391014e-06,
+      "loss": 0.3452,
+      "step": 8423
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.506377018789342e-06,
+      "loss": 0.3659,
+      "step": 8424
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.504233418404821e-06,
+      "loss": 0.3277,
+      "step": 8425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5020903342558944e-06,
+      "loss": 0.3482,
+      "step": 8426
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.49994776651288e-06,
+      "loss": 0.3406,
+      "step": 8427
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4978057153460522e-06,
+      "loss": 0.3954,
+      "step": 8428
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4956641809256485e-06,
+      "loss": 0.3047,
+      "step": 8429
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4935231634218537e-06,
+      "loss": 0.3489,
+      "step": 8430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4913826630048286e-06,
+      "loss": 0.3196,
+      "step": 8431
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4892426798446823e-06,
+      "loss": 0.2803,
+      "step": 8432
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4871032141114836e-06,
+      "loss": 0.3253,
+      "step": 8433
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4849642659752656e-06,
+      "loss": 0.3581,
+      "step": 8434
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4828258356060065e-06,
+      "loss": 0.3532,
+      "step": 8435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.480687923173661e-06,
+      "loss": 0.3261,
+      "step": 8436
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.478550528848134e-06,
+      "loss": 0.3725,
+      "step": 8437
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4764136527992877e-06,
+      "loss": 0.3134,
+      "step": 8438
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4742772951969507e-06,
+      "loss": 0.3541,
+      "step": 8439
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4721414562108924e-06,
+      "loss": 0.3894,
+      "step": 8440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.470006136010865e-06,
+      "loss": 0.3624,
+      "step": 8441
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467871334766568e-06,
+      "loss": 0.3525,
+      "step": 8442
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4657370526476497e-06,
+      "loss": 0.3353,
+      "step": 8443
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4636032898237393e-06,
+      "loss": 0.3365,
+      "step": 8444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4614700464644004e-06,
+      "loss": 0.3454,
+      "step": 8445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4593373227391804e-06,
+      "loss": 0.3503,
+      "step": 8446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4572051188175624e-06,
+      "loss": 0.3654,
+      "step": 8447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4550734348689984e-06,
+      "loss": 0.3334,
+      "step": 8448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.452942271062908e-06,
+      "loss": 0.3173,
+      "step": 8449
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4508116275686488e-06,
+      "loss": 0.3319,
+      "step": 8450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4486815045555612e-06,
+      "loss": 0.366,
+      "step": 8451
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4465519021929206e-06,
+      "loss": 0.3517,
+      "step": 8452
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.444422820649973e-06,
+      "loss": 0.3214,
+      "step": 8453
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.442294260095932e-06,
+      "loss": 0.345,
+      "step": 8454
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.44016622069995e-06,
+      "loss": 0.3225,
+      "step": 8455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4380387026311513e-06,
+      "loss": 0.3544,
+      "step": 8456
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4359117060586146e-06,
+      "loss": 0.3553,
+      "step": 8457
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4337852311513787e-06,
+      "loss": 0.3418,
+      "step": 8458
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4316592780784395e-06,
+      "loss": 0.3182,
+      "step": 8459
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4295338470087537e-06,
+      "loss": 0.3558,
+      "step": 8460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.427408938111233e-06,
+      "loss": 0.3801,
+      "step": 8461
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4252845515547494e-06,
+      "loss": 0.3314,
+      "step": 8462
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4231606875081345e-06,
+      "loss": 0.3556,
+      "step": 8463
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.421037346140177e-06,
+      "loss": 0.3671,
+      "step": 8464
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4189145276196244e-06,
+      "loss": 0.3437,
+      "step": 8465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4167922321151823e-06,
+      "loss": 0.3875,
+      "step": 8466
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.414670459795516e-06,
+      "loss": 0.3416,
+      "step": 8467
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.412549210829247e-06,
+      "loss": 0.3362,
+      "step": 8468
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4104284853849577e-06,
+      "loss": 0.3511,
+      "step": 8469
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4083082836311875e-06,
+      "loss": 0.2737,
+      "step": 8470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4061886057364335e-06,
+      "loss": 0.3668,
+      "step": 8471
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.404069451869153e-06,
+      "loss": 0.3343,
+      "step": 8472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4019508221977613e-06,
+      "loss": 0.3618,
+      "step": 8473
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.39983271689063e-06,
+      "loss": 0.3737,
+      "step": 8474
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.397715136116093e-06,
+      "loss": 0.389,
+      "step": 8475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3955980800424325e-06,
+      "loss": 0.3344,
+      "step": 8476
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.393481548837906e-06,
+      "loss": 0.3227,
+      "step": 8477
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3913655426707184e-06,
+      "loss": 0.3272,
+      "step": 8478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3892500617090253e-06,
+      "loss": 0.3322,
+      "step": 8479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3871351061209625e-06,
+      "loss": 0.3578,
+      "step": 8480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3850206760745976e-06,
+      "loss": 0.3201,
+      "step": 8481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3829067717379804e-06,
+      "loss": 0.3384,
+      "step": 8482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.380793393279107e-06,
+      "loss": 0.3301,
+      "step": 8483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3786805408659252e-06,
+      "loss": 0.3504,
+      "step": 8484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3765682146663613e-06,
+      "loss": 0.3148,
+      "step": 8485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.374456414848274e-06,
+      "loss": 0.3692,
+      "step": 8486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3723451415795074e-06,
+      "loss": 0.3308,
+      "step": 8487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.370234395027838e-06,
+      "loss": 0.3234,
+      "step": 8488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.368124175361015e-06,
+      "loss": 0.3326,
+      "step": 8489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3660144827467513e-06,
+      "loss": 0.34,
+      "step": 8490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3639053173527014e-06,
+      "loss": 0.3385,
+      "step": 8491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3617966793464874e-06,
+      "loss": 0.3231,
+      "step": 8492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3596885688956893e-06,
+      "loss": 0.331,
+      "step": 8493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.357580986167841e-06,
+      "loss": 0.3546,
+      "step": 8494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3554739313304486e-06,
+      "loss": 0.3739,
+      "step": 8495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3533674045509524e-06,
+      "loss": 0.3248,
+      "step": 8496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.35126140599677e-06,
+      "loss": 0.365,
+      "step": 8497
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.349155935835269e-06,
+      "loss": 0.3827,
+      "step": 8498
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.347050994233776e-06,
+      "loss": 0.3328,
+      "step": 8499
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3449465813595784e-06,
+      "loss": 0.3412,
+      "step": 8500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3428426973799178e-06,
+      "loss": 0.3625,
+      "step": 8501
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.340739342461996e-06,
+      "loss": 0.3365,
+      "step": 8502
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3386365167729705e-06,
+      "loss": 0.3337,
+      "step": 8503
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3365342204799613e-06,
+      "loss": 0.3569,
+      "step": 8504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.334432453750042e-06,
+      "loss": 0.3678,
+      "step": 8505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.332331216750244e-06,
+      "loss": 0.3567,
+      "step": 8506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.330230509647561e-06,
+      "loss": 0.3659,
+      "step": 8507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3281303326089386e-06,
+      "loss": 0.3468,
+      "step": 8508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.326030685801286e-06,
+      "loss": 0.353,
+      "step": 8509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3239315693914666e-06,
+      "loss": 0.3139,
+      "step": 8510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3218329835463024e-06,
+      "loss": 0.3204,
+      "step": 8511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.319734928432574e-06,
+      "loss": 0.3383,
+      "step": 8512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.317637404217019e-06,
+      "loss": 0.3273,
+      "step": 8513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3155404110663324e-06,
+      "loss": 0.3526,
+      "step": 8514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3134439491471694e-06,
+      "loss": 0.364,
+      "step": 8515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3113480186261436e-06,
+      "loss": 0.3454,
+      "step": 8516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.309252619669815e-06,
+      "loss": 0.2938,
+      "step": 8517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3071577524447197e-06,
+      "loss": 0.3325,
+      "step": 8518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3050634171173425e-06,
+      "loss": 0.3062,
+      "step": 8519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3029696138541143e-06,
+      "loss": 0.32,
+      "step": 8520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.300876342821451e-06,
+      "loss": 0.3229,
+      "step": 8521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2987836041856958e-06,
+      "loss": 0.3319,
+      "step": 8522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.296691398113173e-06,
+      "loss": 0.3494,
+      "step": 8523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.294599724770158e-06,
+      "loss": 0.3767,
+      "step": 8524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.29250858432287e-06,
+      "loss": 0.3667,
+      "step": 8525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.290417976937511e-06,
+      "loss": 0.3602,
+      "step": 8526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2883279027802174e-06,
+      "loss": 0.3166,
+      "step": 8527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2862383620170957e-06,
+      "loss": 0.3641,
+      "step": 8528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.284149354814208e-06,
+      "loss": 0.3555,
+      "step": 8529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.282060881337568e-06,
+      "loss": 0.3366,
+      "step": 8530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2799729417531647e-06,
+      "loss": 0.3482,
+      "step": 8531
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.27788553622692e-06,
+      "loss": 0.3493,
+      "step": 8532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.27579866492473e-06,
+      "loss": 0.3386,
+      "step": 8533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2737123280124437e-06,
+      "loss": 0.3396,
+      "step": 8534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2716265256558645e-06,
+      "loss": 0.3593,
+      "step": 8535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2695412580207664e-06,
+      "loss": 0.3199,
+      "step": 8536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.267456525272861e-06,
+      "loss": 0.3418,
+      "step": 8537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.265372327577829e-06,
+      "loss": 0.3222,
+      "step": 8538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2632886651013107e-06,
+      "loss": 0.3123,
+      "step": 8539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.261205538008898e-06,
+      "loss": 0.3879,
+      "step": 8540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2591229464661433e-06,
+      "loss": 0.3638,
+      "step": 8541
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2570408906385555e-06,
+      "loss": 0.3654,
+      "step": 8542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2549593706915994e-06,
+      "loss": 0.3946,
+      "step": 8543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.252878386790701e-06,
+      "loss": 0.3357,
+      "step": 8544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2507979391012413e-06,
+      "loss": 0.3924,
+      "step": 8545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2487180277885565e-06,
+      "loss": 0.3371,
+      "step": 8546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.246638653017946e-06,
+      "loss": 0.337,
+      "step": 8547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2445598149546607e-06,
+      "loss": 0.3206,
+      "step": 8548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2424815137639132e-06,
+      "loss": 0.3699,
+      "step": 8549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.24040374961087e-06,
+      "loss": 0.3858,
+      "step": 8550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2383265226606576e-06,
+      "loss": 0.3285,
+      "step": 8551
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2362498330783563e-06,
+      "loss": 0.3443,
+      "step": 8552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2341736810290105e-06,
+      "loss": 0.4056,
+      "step": 8553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2320980666776123e-06,
+      "loss": 0.3172,
+      "step": 8554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.230022990189123e-06,
+      "loss": 0.3435,
+      "step": 8555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2279484517284442e-06,
+      "loss": 0.3624,
+      "step": 8556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.225874451460457e-06,
+      "loss": 0.3382,
+      "step": 8557
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2238009895499743e-06,
+      "loss": 0.3376,
+      "step": 8558
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.22172806616179e-06,
+      "loss": 0.3563,
+      "step": 8559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.219655681460645e-06,
+      "loss": 0.3456,
+      "step": 8560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2175838356112264e-06,
+      "loss": 0.3572,
+      "step": 8561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.215512528778204e-06,
+      "loss": 0.3232,
+      "step": 8562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.213441761126176e-06,
+      "loss": 0.3126,
+      "step": 8563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2113715328197203e-06,
+      "loss": 0.3479,
+      "step": 8564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.209301844023365e-06,
+      "loss": 0.3731,
+      "step": 8565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.207232694901583e-06,
+      "loss": 0.3656,
+      "step": 8566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2051640856188294e-06,
+      "loss": 0.3424,
+      "step": 8567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2030960163394898e-06,
+      "loss": 0.3321,
+      "step": 8568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2010284872279205e-06,
+      "loss": 0.3517,
+      "step": 8569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.198961498448444e-06,
+      "loss": 0.3707,
+      "step": 8570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.196895050165316e-06,
+      "loss": 0.3531,
+      "step": 8571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.194829142542775e-06,
+      "loss": 0.3422,
+      "step": 8572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1927637757449936e-06,
+      "loss": 0.3212,
+      "step": 8573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1906989499361164e-06,
+      "loss": 0.3344,
+      "step": 8574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.18863466528024e-06,
+      "loss": 0.3168,
+      "step": 8575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1865709219414153e-06,
+      "loss": 0.3325,
+      "step": 8576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1845077200836638e-06,
+      "loss": 0.3378,
+      "step": 8577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.182445059870942e-06,
+      "loss": 0.3614,
+      "step": 8578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1803829414671805e-06,
+      "loss": 0.3012,
+      "step": 8579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.17832136503626e-06,
+      "loss": 0.3553,
+      "step": 8580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1762603307420194e-06,
+      "loss": 0.3797,
+      "step": 8581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.174199838748254e-06,
+      "loss": 0.3313,
+      "step": 8582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.172139889218718e-06,
+      "loss": 0.3637,
+      "step": 8583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1700804823171203e-06,
+      "loss": 0.319,
+      "step": 8584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1680216182071265e-06,
+      "loss": 0.3637,
+      "step": 8585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1659632970523615e-06,
+      "loss": 0.3813,
+      "step": 8586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1639055190164038e-06,
+      "loss": 0.3939,
+      "step": 8587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1618482842627905e-06,
+      "loss": 0.3354,
+      "step": 8588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.159791592955017e-06,
+      "loss": 0.3528,
+      "step": 8589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1577354452565332e-06,
+      "loss": 0.3272,
+      "step": 8590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1556798413307466e-06,
+      "loss": 0.3582,
+      "step": 8591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1536247813410216e-06,
+      "loss": 0.3529,
+      "step": 8592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1515702654506775e-06,
+      "loss": 0.3445,
+      "step": 8593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.149516293822994e-06,
+      "loss": 0.3156,
+      "step": 8594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.147462866621206e-06,
+      "loss": 0.351,
+      "step": 8595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.145409984008506e-06,
+      "loss": 0.3368,
+      "step": 8596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.143357646148033e-06,
+      "loss": 0.3299,
+      "step": 8597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141305853202905e-06,
+      "loss": 0.3456,
+      "step": 8598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.139254605336172e-06,
+      "loss": 0.3439,
+      "step": 8599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.137203902710858e-06,
+      "loss": 0.3375,
+      "step": 8600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1351537454899407e-06,
+      "loss": 0.3453,
+      "step": 8601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1331041338363398e-06,
+      "loss": 0.3227,
+      "step": 8602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.131055067912958e-06,
+      "loss": 0.294,
+      "step": 8603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1290065478826305e-06,
+      "loss": 0.3038,
+      "step": 8604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1269585739081564e-06,
+      "loss": 0.3348,
+      "step": 8605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.124911146152304e-06,
+      "loss": 0.3427,
+      "step": 8606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1228642647777763e-06,
+      "loss": 0.3473,
+      "step": 8607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1208179299472564e-06,
+      "loss": 0.3725,
+      "step": 8608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1187721418233607e-06,
+      "loss": 0.3514,
+      "step": 8609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1167269005686762e-06,
+      "loss": 0.3321,
+      "step": 8610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1146822063457504e-06,
+      "loss": 0.3435,
+      "step": 8611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.112638059317069e-06,
+      "loss": 0.3571,
+      "step": 8612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1105944596450997e-06,
+      "loss": 0.3131,
+      "step": 8613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1085514074922407e-06,
+      "loss": 0.3047,
+      "step": 8614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.106508903020864e-06,
+      "loss": 0.3513,
+      "step": 8615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1044669463932908e-06,
+      "loss": 0.3387,
+      "step": 8616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.102425537771803e-06,
+      "loss": 0.347,
+      "step": 8617
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1003846773186354e-06,
+      "loss": 0.3714,
+      "step": 8618
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.09834436519598e-06,
+      "loss": 0.3098,
+      "step": 8619
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.096304601565987e-06,
+      "loss": 0.3454,
+      "step": 8620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0942653865907603e-06,
+      "loss": 0.3477,
+      "step": 8621
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.092226720432363e-06,
+      "loss": 0.3436,
+      "step": 8622
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0901886032528126e-06,
+      "loss": 0.3655,
+      "step": 8623
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0881510352140853e-06,
+      "loss": 0.3403,
+      "step": 8624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.086114016478109e-06,
+      "loss": 0.3261,
+      "step": 8625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.084077547206773e-06,
+      "loss": 0.3394,
+      "step": 8626
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0820416275619193e-06,
+      "loss": 0.3359,
+      "step": 8627
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.080006257705349e-06,
+      "loss": 0.3569,
+      "step": 8628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.077971437798819e-06,
+      "loss": 0.336,
+      "step": 8629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0759371680040397e-06,
+      "loss": 0.3343,
+      "step": 8630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0739034484826814e-06,
+      "loss": 0.3437,
+      "step": 8631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0718702793963717e-06,
+      "loss": 0.3322,
+      "step": 8632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0698376609066828e-06,
+      "loss": 0.417,
+      "step": 8633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0678055931751604e-06,
+      "loss": 0.3235,
+      "step": 8634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.065774076363297e-06,
+      "loss": 0.3399,
+      "step": 8635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0637431106325412e-06,
+      "loss": 0.3487,
+      "step": 8636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0617126961443032e-06,
+      "loss": 0.3433,
+      "step": 8637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0596828330599348e-06,
+      "loss": 0.3487,
+      "step": 8638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0576535215407656e-06,
+      "loss": 0.3506,
+      "step": 8639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055624761748066e-06,
+      "loss": 0.3259,
+      "step": 8640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0535965538430667e-06,
+      "loss": 0.3327,
+      "step": 8641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0515688979869584e-06,
+      "loss": 0.3102,
+      "step": 8642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0495417943408744e-06,
+      "loss": 0.3335,
+      "step": 8643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0475152430659262e-06,
+      "loss": 0.3427,
+      "step": 8644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.045489244323161e-06,
+      "loss": 0.3599,
+      "step": 8645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0434637982735894e-06,
+      "loss": 0.3664,
+      "step": 8646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0414389050781877e-06,
+      "loss": 0.3269,
+      "step": 8647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039414564897869e-06,
+      "loss": 0.3755,
+      "step": 8648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.037390777893523e-06,
+      "loss": 0.3409,
+      "step": 8649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.035367544225978e-06,
+      "loss": 0.3627,
+      "step": 8650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0333448640560235e-06,
+      "loss": 0.3711,
+      "step": 8651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0313227375444176e-06,
+      "loss": 0.3575,
+      "step": 8652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0293011648518555e-06,
+      "loss": 0.3378,
+      "step": 8653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.027280146139e-06,
+      "loss": 0.3934,
+      "step": 8654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.025259681566465e-06,
+      "loss": 0.351,
+      "step": 8655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0232397712948235e-06,
+      "loss": 0.3175,
+      "step": 8656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0212204154846024e-06,
+      "loss": 0.3224,
+      "step": 8657
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0192016142962866e-06,
+      "loss": 0.3442,
+      "step": 8658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017183367890314e-06,
+      "loss": 0.3728,
+      "step": 8659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.01516567642708e-06,
+      "loss": 0.3723,
+      "step": 8660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.013148540066936e-06,
+      "loss": 0.3578,
+      "step": 8661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0111319589701902e-06,
+      "loss": 0.3332,
+      "step": 8662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0091159332971042e-06,
+      "loss": 0.3269,
+      "step": 8663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.007100463207897e-06,
+      "loss": 0.3394,
+      "step": 8664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.005085548862744e-06,
+      "loss": 0.3781,
+      "step": 8665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.003071190421776e-06,
+      "loss": 0.3279,
+      "step": 8666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0010573880450778e-06,
+      "loss": 0.391,
+      "step": 8667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9990441418926964e-06,
+      "loss": 0.3703,
+      "step": 8668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997031452124619e-06,
+      "loss": 0.3139,
+      "step": 8669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.995019318900809e-06,
+      "loss": 0.3043,
+      "step": 8670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.993007742381173e-06,
+      "loss": 0.3913,
+      "step": 8671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9909967227255777e-06,
+      "loss": 0.3181,
+      "step": 8672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9889862600938447e-06,
+      "loss": 0.3408,
+      "step": 8673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.986976354645743e-06,
+      "loss": 0.3466,
+      "step": 8674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.984967006541013e-06,
+      "loss": 0.3753,
+      "step": 8675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.982958215939341e-06,
+      "loss": 0.3921,
+      "step": 8676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9809499830003717e-06,
+      "loss": 0.3659,
+      "step": 8677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9789423078837055e-06,
+      "loss": 0.3415,
+      "step": 8678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9769351907488896e-06,
+      "loss": 0.3273,
+      "step": 8679
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9749286317554436e-06,
+      "loss": 0.334,
+      "step": 8680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9729226310628346e-06,
+      "loss": 0.3462,
+      "step": 8681
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9709171888304755e-06,
+      "loss": 0.3429,
+      "step": 8682
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9689123052177558e-06,
+      "loss": 0.3755,
+      "step": 8683
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.966907980383996e-06,
+      "loss": 0.3487,
+      "step": 8684
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9649042144884987e-06,
+      "loss": 0.3321,
+      "step": 8685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9629010076904996e-06,
+      "loss": 0.3399,
+      "step": 8686
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.960898360149197e-06,
+      "loss": 0.3273,
+      "step": 8687
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.958896272023757e-06,
+      "loss": 0.3639,
+      "step": 8688
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9568947434732777e-06,
+      "loss": 0.3088,
+      "step": 8689
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9548937746568384e-06,
+      "loss": 0.3237,
+      "step": 8690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9528933657334523e-06,
+      "loss": 0.3339,
+      "step": 8691
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9508935168620976e-06,
+      "loss": 0.3427,
+      "step": 8692
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.948894228201715e-06,
+      "loss": 0.3784,
+      "step": 8693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9468954999111864e-06,
+      "loss": 0.3374,
+      "step": 8694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9448973321493578e-06,
+      "loss": 0.3509,
+      "step": 8695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.942899725075029e-06,
+      "loss": 0.3277,
+      "step": 8696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9409026788469553e-06,
+      "loss": 0.3838,
+      "step": 8697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.938906193623846e-06,
+      "loss": 0.3322,
+      "step": 8698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.936910269564368e-06,
+      "loss": 0.3445,
+      "step": 8699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.934914906827142e-06,
+      "loss": 0.3585,
+      "step": 8700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9329201055707468e-06,
+      "loss": 0.3868,
+      "step": 8701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.930925865953712e-06,
+      "loss": 0.3439,
+      "step": 8702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.3674,
+      "step": 8703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9269390722716305e-06,
+      "loss": 0.3425,
+      "step": 8704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.924946518523426e-06,
+      "loss": 0.3364,
+      "step": 8705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.922954527048265e-06,
+      "loss": 0.3285,
+      "step": 8706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9209630980044544e-06,
+      "loss": 0.3752,
+      "step": 8707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.918972231550261e-06,
+      "loss": 0.3485,
+      "step": 8708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9169819278439027e-06,
+      "loss": 0.3583,
+      "step": 8709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.914992187043555e-06,
+      "loss": 0.3452,
+      "step": 8710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.913003009307346e-06,
+      "loss": 0.3046,
+      "step": 8711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9110143947933633e-06,
+      "loss": 0.3295,
+      "step": 8712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9090263436596455e-06,
+      "loss": 0.3416,
+      "step": 8713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9070388560641918e-06,
+      "loss": 0.3634,
+      "step": 8714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.905051932164945e-06,
+      "loss": 0.34,
+      "step": 8715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9030655721198186e-06,
+      "loss": 0.3221,
+      "step": 8716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.901079776086674e-06,
+      "loss": 0.3666,
+      "step": 8717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89909454422332e-06,
+      "loss": 0.3581,
+      "step": 8718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.897109876687539e-06,
+      "loss": 0.3002,
+      "step": 8719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8951257736370465e-06,
+      "loss": 0.3554,
+      "step": 8720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8931422352295314e-06,
+      "loss": 0.335,
+      "step": 8721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8911592616226335e-06,
+      "loss": 0.3348,
+      "step": 8722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.889176852973935e-06,
+      "loss": 0.3628,
+      "step": 8723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.887195009440994e-06,
+      "loss": 0.3381,
+      "step": 8724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.885213731181301e-06,
+      "loss": 0.3341,
+      "step": 8725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.883233018352327e-06,
+      "loss": 0.3366,
+      "step": 8726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.881252871111474e-06,
+      "loss": 0.3713,
+      "step": 8727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.879273289616109e-06,
+      "loss": 0.3123,
+      "step": 8728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8772942740235655e-06,
+      "loss": 0.3352,
+      "step": 8729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8753158244911106e-06,
+      "loss": 0.3275,
+      "step": 8730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.87333794117598e-06,
+      "loss": 0.3555,
+      "step": 8731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.871360624235361e-06,
+      "loss": 0.3416,
+      "step": 8732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8693838738263944e-06,
+      "loss": 0.3211,
+      "step": 8733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8674076901061853e-06,
+      "loss": 0.3735,
+      "step": 8734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8654320732317786e-06,
+      "loss": 0.3202,
+      "step": 8735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8634570233601832e-06,
+      "loss": 0.3615,
+      "step": 8736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8614825406483625e-06,
+      "loss": 0.3364,
+      "step": 8737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8595086252532312e-06,
+      "loss": 0.3332,
+      "step": 8738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.85753527733167e-06,
+      "loss": 0.3088,
+      "step": 8739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.855562497040496e-06,
+      "loss": 0.3397,
+      "step": 8740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.853590284536496e-06,
+      "loss": 0.406,
+      "step": 8741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8516186399764057e-06,
+      "loss": 0.3371,
+      "step": 8742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8496475635169176e-06,
+      "loss": 0.3497,
+      "step": 8743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.847677055314678e-06,
+      "loss": 0.3205,
+      "step": 8744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8457071155262885e-06,
+      "loss": 0.3122,
+      "step": 8745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8437377443083036e-06,
+      "loss": 0.3879,
+      "step": 8746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8417689418172376e-06,
+      "loss": 0.3269,
+      "step": 8747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8398007082095546e-06,
+      "loss": 0.3708,
+      "step": 8748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8378330436416748e-06,
+      "loss": 0.3573,
+      "step": 8749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8358659482699748e-06,
+      "loss": 0.3929,
+      "step": 8750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8338994222507843e-06,
+      "loss": 0.366,
+      "step": 8751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8319334657403874e-06,
+      "loss": 0.3294,
+      "step": 8752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.829968078895026e-06,
+      "loss": 0.3207,
+      "step": 8753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.828003261870893e-06,
+      "loss": 0.3654,
+      "step": 8754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.826039014824141e-06,
+      "loss": 0.3282,
+      "step": 8755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8240753379108645e-06,
+      "loss": 0.3791,
+      "step": 8756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8221122312871307e-06,
+      "loss": 0.3875,
+      "step": 8757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8201496951089537e-06,
+      "loss": 0.3267,
+      "step": 8758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8181877295322923e-06,
+      "loss": 0.3425,
+      "step": 8759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8162263347130805e-06,
+      "loss": 0.3283,
+      "step": 8760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8142655108071838e-06,
+      "loss": 0.3587,
+      "step": 8761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8123052579704436e-06,
+      "loss": 0.3479,
+      "step": 8762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.810345576358645e-06,
+      "loss": 0.3424,
+      "step": 8763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8083864661275216e-06,
+      "loss": 0.3253,
+      "step": 8764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8064279274327797e-06,
+      "loss": 0.3337,
+      "step": 8765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8044699604300605e-06,
+      "loss": 0.3362,
+      "step": 8766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.802512565274974e-06,
+      "loss": 0.3457,
+      "step": 8767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8005557421230765e-06,
+      "loss": 0.3157,
+      "step": 8768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7985994911298795e-06,
+      "loss": 0.3638,
+      "step": 8769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7966438124508622e-06,
+      "loss": 0.3681,
+      "step": 8770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7946887062414374e-06,
+      "loss": 0.3456,
+      "step": 8771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7927341726569857e-06,
+      "loss": 0.347,
+      "step": 8772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.790780211852838e-06,
+      "loss": 0.3293,
+      "step": 8773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7888268239842784e-06,
+      "loss": 0.3513,
+      "step": 8774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786874009206557e-06,
+      "loss": 0.336,
+      "step": 8775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.784921767674861e-06,
+      "loss": 0.3644,
+      "step": 8776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.782970099544341e-06,
+      "loss": 0.3227,
+      "step": 8777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7810190049701026e-06,
+      "loss": 0.3598,
+      "step": 8778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7790684841072037e-06,
+      "loss": 0.336,
+      "step": 8779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.777118537110658e-06,
+      "loss": 0.3757,
+      "step": 8780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7751691641354326e-06,
+      "loss": 0.3424,
+      "step": 8781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7732203653364487e-06,
+      "loss": 0.3367,
+      "step": 8782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7712721408685838e-06,
+      "loss": 0.3544,
+      "step": 8783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7693244908866677e-06,
+      "loss": 0.3088,
+      "step": 8784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.767377415545485e-06,
+      "loss": 0.3304,
+      "step": 8785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7654309149997737e-06,
+      "loss": 0.3305,
+      "step": 8786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7634849894042303e-06,
+      "loss": 0.3741,
+      "step": 8787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7615396389135006e-06,
+      "loss": 0.3585,
+      "step": 8788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7595948636821867e-06,
+      "loss": 0.3353,
+      "step": 8789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7576506638648458e-06,
+      "loss": 0.3388,
+      "step": 8790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.755707039615988e-06,
+      "loss": 0.3216,
+      "step": 8791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7537639910900783e-06,
+      "loss": 0.3428,
+      "step": 8792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7518215184415354e-06,
+      "loss": 0.3133,
+      "step": 8793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.749879621824736e-06,
+      "loss": 0.3379,
+      "step": 8794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.747938301394e-06,
+      "loss": 0.357,
+      "step": 8795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.745997557303619e-06,
+      "loss": 0.3575,
+      "step": 8796
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.744057389707818e-06,
+      "loss": 0.3405,
+      "step": 8797
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7421177987607973e-06,
+      "loss": 0.3329,
+      "step": 8798
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7401787846166994e-06,
+      "loss": 0.3274,
+      "step": 8799
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.738240347429615e-06,
+      "loss": 0.3395,
+      "step": 8800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7363024873536093e-06,
+      "loss": 0.3142,
+      "step": 8801
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.734365204542675e-06,
+      "loss": 0.3379,
+      "step": 8802
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.732428499150784e-06,
+      "loss": 0.3387,
+      "step": 8803
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.730492371331851e-06,
+      "loss": 0.3116,
+      "step": 8804
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7285568212397353e-06,
+      "loss": 0.3252,
+      "step": 8805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7266218490282738e-06,
+      "loss": 0.3201,
+      "step": 8806
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7246874548512325e-06,
+      "loss": 0.2779,
+      "step": 8807
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7227536388623445e-06,
+      "loss": 0.3588,
+      "step": 8808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.720820401215304e-06,
+      "loss": 0.3548,
+      "step": 8809
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7188877420637372e-06,
+      "loss": 0.3123,
+      "step": 8810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7169556615612512e-06,
+      "loss": 0.352,
+      "step": 8811
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7150241598613846e-06,
+      "loss": 0.3649,
+      "step": 8812
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.71309323711764e-06,
+      "loss": 0.3716,
+      "step": 8813
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7111628934834745e-06,
+      "loss": 0.3597,
+      "step": 8814
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7092331291122976e-06,
+      "loss": 0.3872,
+      "step": 8815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.707303944157472e-06,
+      "loss": 0.3416,
+      "step": 8816
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7053753387723146e-06,
+      "loss": 0.3387,
+      "step": 8817
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7034473131100993e-06,
+      "loss": 0.3211,
+      "step": 8818
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7015198673240473e-06,
+      "loss": 0.3292,
+      "step": 8819
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.699593001567341e-06,
+      "loss": 0.3317,
+      "step": 8820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6976667159931136e-06,
+      "loss": 0.3726,
+      "step": 8821
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.695741010754449e-06,
+      "loss": 0.3436,
+      "step": 8822
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6938158860043916e-06,
+      "loss": 0.3035,
+      "step": 8823
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6918913418959337e-06,
+      "loss": 0.4133,
+      "step": 8824
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6899673785820244e-06,
+      "loss": 0.3458,
+      "step": 8825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6880439962155668e-06,
+      "loss": 0.3637,
+      "step": 8826
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6861211949494172e-06,
+      "loss": 0.3506,
+      "step": 8827
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.684198974936384e-06,
+      "loss": 0.3033,
+      "step": 8828
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6822773363292333e-06,
+      "loss": 0.3342,
+      "step": 8829
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6803562792806835e-06,
+      "loss": 0.3453,
+      "step": 8830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.678435803943399e-06,
+      "loss": 0.3401,
+      "step": 8831
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6765159104700135e-06,
+      "loss": 0.3266,
+      "step": 8832
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6745965990131017e-06,
+      "loss": 0.3275,
+      "step": 8833
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6726778697251978e-06,
+      "loss": 0.3783,
+      "step": 8834
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6707597227587913e-06,
+      "loss": 0.3427,
+      "step": 8835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6688421582663114e-06,
+      "loss": 0.3412,
+      "step": 8836
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.666925176400167e-06,
+      "loss": 0.3799,
+      "step": 8837
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.665008777312691e-06,
+      "loss": 0.3293,
+      "step": 8838
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.663092961156195e-06,
+      "loss": 0.3455,
+      "step": 8839
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6611777280829344e-06,
+      "loss": 0.3601,
+      "step": 8840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6592630782451066e-06,
+      "loss": 0.3412,
+      "step": 8841
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6573490117948887e-06,
+      "loss": 0.3503,
+      "step": 8842
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6554355288843847e-06,
+      "loss": 0.3543,
+      "step": 8843
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.653522629665667e-06,
+      "loss": 0.319,
+      "step": 8844
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6516103142907658e-06,
+      "loss": 0.3449,
+      "step": 8845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6496985829116463e-06,
+      "loss": 0.3697,
+      "step": 8846
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.64778743568025e-06,
+      "loss": 0.2894,
+      "step": 8847
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.645876872748454e-06,
+      "loss": 0.3409,
+      "step": 8848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.643966894268093e-06,
+      "loss": 0.3339,
+      "step": 8849
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.642057500390969e-06,
+      "loss": 0.3405,
+      "step": 8850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6401486912688136e-06,
+      "loss": 0.3319,
+      "step": 8851
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6382404670533368e-06,
+      "loss": 0.376,
+      "step": 8852
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.636332827896182e-06,
+      "loss": 0.3388,
+      "step": 8853
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6344257739489563e-06,
+      "loss": 0.3553,
+      "step": 8854
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6325193053632203e-06,
+      "loss": 0.3295,
+      "step": 8855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6306134222904832e-06,
+      "loss": 0.381,
+      "step": 8856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.628708124882212e-06,
+      "loss": 0.3635,
+      "step": 8857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6268034132898246e-06,
+      "loss": 0.3135,
+      "step": 8858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.624899287664696e-06,
+      "loss": 0.3674,
+      "step": 8859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622995748158149e-06,
+      "loss": 0.3262,
+      "step": 8860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.621092794921465e-06,
+      "loss": 0.3359,
+      "step": 8861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6191904281058756e-06,
+      "loss": 0.3504,
+      "step": 8862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6172886478625683e-06,
+      "loss": 0.3294,
+      "step": 8863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6153874543426803e-06,
+      "loss": 0.3255,
+      "step": 8864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6134868476973054e-06,
+      "loss": 0.357,
+      "step": 8865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61158682807749e-06,
+      "loss": 0.3384,
+      "step": 8866
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.609687395634235e-06,
+      "loss": 0.3231,
+      "step": 8867
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.607788550518491e-06,
+      "loss": 0.3475,
+      "step": 8868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.605890292881165e-06,
+      "loss": 0.3337,
+      "step": 8869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.603992622873117e-06,
+      "loss": 0.3468,
+      "step": 8870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.602095540645162e-06,
+      "loss": 0.3717,
+      "step": 8871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6001990463480575e-06,
+      "loss": 0.3098,
+      "step": 8872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598303140132532e-06,
+      "loss": 0.3264,
+      "step": 8873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.596407822149255e-06,
+      "loss": 0.3286,
+      "step": 8874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5945130925488527e-06,
+      "loss": 0.3587,
+      "step": 8875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5926189514819065e-06,
+      "loss": 0.3097,
+      "step": 8876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.590725399098941e-06,
+      "loss": 0.3444,
+      "step": 8877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.588832435550449e-06,
+      "loss": 0.3207,
+      "step": 8878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5869400609868677e-06,
+      "loss": 0.3167,
+      "step": 8879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.585048275558589e-06,
+      "loss": 0.3381,
+      "step": 8880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.583157079415961e-06,
+      "loss": 0.3652,
+      "step": 8881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5812664727092716e-06,
+      "loss": 0.3457,
+      "step": 8882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.579376455588786e-06,
+      "loss": 0.332,
+      "step": 8883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.577487028204698e-06,
+      "loss": 0.3254,
+      "step": 8884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5755981907071683e-06,
+      "loss": 0.3607,
+      "step": 8885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.573709943246314e-06,
+      "loss": 0.356,
+      "step": 8886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5718222859721875e-06,
+      "loss": 0.3316,
+      "step": 8887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.569935219034819e-06,
+      "loss": 0.3404,
+      "step": 8888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.568048742584169e-06,
+      "loss": 0.3482,
+      "step": 8889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5661628567701614e-06,
+      "loss": 0.364,
+      "step": 8890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5642775617426806e-06,
+      "loss": 0.3452,
+      "step": 8891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5623928576515467e-06,
+      "loss": 0.3482,
+      "step": 8892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.560508744646546e-06,
+      "loss": 0.3479,
+      "step": 8893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5586252228774132e-06,
+      "loss": 0.3811,
+      "step": 8894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5567422924938377e-06,
+      "loss": 0.3356,
+      "step": 8895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55485995364546e-06,
+      "loss": 0.3687,
+      "step": 8896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5529782064818744e-06,
+      "loss": 0.3445,
+      "step": 8897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.551097051152629e-06,
+      "loss": 0.3615,
+      "step": 8898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5492164878072235e-06,
+      "loss": 0.3526,
+      "step": 8899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.54733651659511e-06,
+      "loss": 0.3387,
+      "step": 8900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.545457137665698e-06,
+      "loss": 0.3454,
+      "step": 8901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5435783511683444e-06,
+      "loss": 0.3454,
+      "step": 8902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5417001572523627e-06,
+      "loss": 0.3853,
+      "step": 8903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5398225560670154e-06,
+      "loss": 0.3259,
+      "step": 8904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.537945547761522e-06,
+      "loss": 0.3685,
+      "step": 8905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.536069132485054e-06,
+      "loss": 0.3162,
+      "step": 8906
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5341933103867344e-06,
+      "loss": 0.3564,
+      "step": 8907
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532318081615639e-06,
+      "loss": 0.3207,
+      "step": 8908
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.530443446320797e-06,
+      "loss": 0.3502,
+      "step": 8909
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5285694046511924e-06,
+      "loss": 0.3071,
+      "step": 8910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5266959567557593e-06,
+      "loss": 0.323,
+      "step": 8911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.524823102783387e-06,
+      "loss": 0.3496,
+      "step": 8912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.52295084288291e-06,
+      "loss": 0.3316,
+      "step": 8913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5210791772031274e-06,
+      "loss": 0.3717,
+      "step": 8914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.519208105892784e-06,
+      "loss": 0.3684,
+      "step": 8915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.51733762910058e-06,
+      "loss": 0.3462,
+      "step": 8916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5154677469751687e-06,
+      "loss": 0.3106,
+      "step": 8917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5135984596651453e-06,
+      "loss": 0.3533,
+      "step": 8918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5117297673190766e-06,
+      "loss": 0.3441,
+      "step": 8919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.509861670085473e-06,
+      "loss": 0.3231,
+      "step": 8920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5079941681127862e-06,
+      "loss": 0.3565,
+      "step": 8921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5061272615494446e-06,
+      "loss": 0.3574,
+      "step": 8922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5042609505438052e-06,
+      "loss": 0.3442,
+      "step": 8923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.502395235244198e-06,
+      "loss": 0.3673,
+      "step": 8924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5005301157988902e-06,
+      "loss": 0.32,
+      "step": 8925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4986655923561066e-06,
+      "loss": 0.3371,
+      "step": 8926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496801665064035e-06,
+      "loss": 0.3265,
+      "step": 8927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.494938334070792e-06,
+      "loss": 0.3304,
+      "step": 8928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.493075599524477e-06,
+      "loss": 0.3377,
+      "step": 8929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4912134615731165e-06,
+      "loss": 0.3344,
+      "step": 8930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.489351920364699e-06,
+      "loss": 0.3171,
+      "step": 8931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.487490976047173e-06,
+      "loss": 0.3699,
+      "step": 8932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.485630628768427e-06,
+      "loss": 0.3398,
+      "step": 8933
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4837708786763104e-06,
+      "loss": 0.3664,
+      "step": 8934
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.48191172591862e-06,
+      "loss": 0.3913,
+      "step": 8935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4800531706431097e-06,
+      "loss": 0.3185,
+      "step": 8936
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.478195212997482e-06,
+      "loss": 0.3188,
+      "step": 8937
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4763378531293948e-06,
+      "loss": 0.3316,
+      "step": 8938
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.474481091186457e-06,
+      "loss": 0.3022,
+      "step": 8939
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.472624927316232e-06,
+      "loss": 0.3274,
+      "step": 8940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.470769361666231e-06,
+      "loss": 0.3327,
+      "step": 8941
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.468914394383922e-06,
+      "loss": 0.3088,
+      "step": 8942
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4670600256167245e-06,
+      "loss": 0.3482,
+      "step": 8943
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4652062555120103e-06,
+      "loss": 0.3358,
+      "step": 8944
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4633530842171027e-06,
+      "loss": 0.3708,
+      "step": 8945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4615005118792777e-06,
+      "loss": 0.3765,
+      "step": 8946
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.459648538645766e-06,
+      "loss": 0.3307,
+      "step": 8947
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4577971646637468e-06,
+      "loss": 0.3226,
+      "step": 8948
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4559463900803538e-06,
+      "loss": 0.3669,
+      "step": 8949
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.454096215042673e-06,
+      "loss": 0.3291,
+      "step": 8950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.452246639697744e-06,
+      "loss": 0.3529,
+      "step": 8951
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4503976641925563e-06,
+      "loss": 0.3149,
+      "step": 8952
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.448549288674055e-06,
+      "loss": 0.3321,
+      "step": 8953
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.446701513289128e-06,
+      "loss": 0.3247,
+      "step": 8954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4448543381846314e-06,
+      "loss": 0.3446,
+      "step": 8955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4430077635073647e-06,
+      "loss": 0.3919,
+      "step": 8956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4411617894040705e-06,
+      "loss": 0.3348,
+      "step": 8957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4393164160214667e-06,
+      "loss": 0.3056,
+      "step": 8958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4374716435061963e-06,
+      "loss": 0.3433,
+      "step": 8959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.435627472004879e-06,
+      "loss": 0.3732,
+      "step": 8960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.433783901664074e-06,
+      "loss": 0.3323,
+      "step": 8961
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4319409326302867e-06,
+      "loss": 0.3458,
+      "step": 8962
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4300985650499953e-06,
+      "loss": 0.3206,
+      "step": 8963
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.428256799069604e-06,
+      "loss": 0.3612,
+      "step": 8964
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4264156348354963e-06,
+      "loss": 0.3304,
+      "step": 8965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4245750724939844e-06,
+      "loss": 0.337,
+      "step": 8966
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422735112191342e-06,
+      "loss": 0.3523,
+      "step": 8967
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4208957540738067e-06,
+      "loss": 0.3493,
+      "step": 8968
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.419056998287547e-06,
+      "loss": 0.3463,
+      "step": 8969
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.417218844978696e-06,
+      "loss": 0.3439,
+      "step": 8970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4153812942933363e-06,
+      "loss": 0.3736,
+      "step": 8971
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4135443463775022e-06,
+      "loss": 0.3347,
+      "step": 8972
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.411708001377188e-06,
+      "loss": 0.3514,
+      "step": 8973
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409872259438323e-06,
+      "loss": 0.3347,
+      "step": 8974
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4080371207068033e-06,
+      "loss": 0.3208,
+      "step": 8975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.406202585328471e-06,
+      "loss": 0.3422,
+      "step": 8976
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4043686534491184e-06,
+      "loss": 0.3705,
+      "step": 8977
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.402535325214502e-06,
+      "loss": 0.3479,
+      "step": 8978
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4007026007703136e-06,
+      "loss": 0.3207,
+      "step": 8979
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3988704802622063e-06,
+      "loss": 0.3629,
+      "step": 8980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.397038963835784e-06,
+      "loss": 0.3719,
+      "step": 8981
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3952080516366004e-06,
+      "loss": 0.3502,
+      "step": 8982
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.393377743810166e-06,
+      "loss": 0.3939,
+      "step": 8983
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3915480405019374e-06,
+      "loss": 0.3263,
+      "step": 8984
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3897189418573275e-06,
+      "loss": 0.3378,
+      "step": 8985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.387890448021698e-06,
+      "loss": 0.3551,
+      "step": 8986
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.386062559140366e-06,
+      "loss": 0.367,
+      "step": 8987
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3842352753585983e-06,
+      "loss": 0.3537,
+      "step": 8988
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3824085968216127e-06,
+      "loss": 0.3732,
+      "step": 8989
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3805825236745804e-06,
+      "loss": 0.3118,
+      "step": 8990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.378757056062626e-06,
+      "loss": 0.3415,
+      "step": 8991
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3769321941308223e-06,
+      "loss": 0.3084,
+      "step": 8992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.375107938024196e-06,
+      "loss": 0.3084,
+      "step": 8993
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3732842878877283e-06,
+      "loss": 0.3533,
+      "step": 8994
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.371461243866342e-06,
+      "loss": 0.334,
+      "step": 8995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3696388061049267e-06,
+      "loss": 0.356,
+      "step": 8996
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.367816974748317e-06,
+      "loss": 0.3593,
+      "step": 8997
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.36599574994129e-06,
+      "loss": 0.3239,
+      "step": 8998
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3641751318285946e-06,
+      "loss": 0.3368,
+      "step": 8999
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3623551205549077e-06,
+      "loss": 0.3476,
+      "step": 9000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.36053571626488e-06,
+      "loss": 0.3014,
+      "step": 9001
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3587169191031047e-06,
+      "loss": 0.3792,
+      "step": 9002
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3568987292141166e-06,
+      "loss": 0.3361,
+      "step": 9003
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.355081146742424e-06,
+      "loss": 0.3252,
+      "step": 9004
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3532641718324667e-06,
+      "loss": 0.3417,
+      "step": 9005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.351447804628646e-06,
+      "loss": 0.3532,
+      "step": 9006
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3496320452753154e-06,
+      "loss": 0.335,
+      "step": 9007
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3478168939167734e-06,
+      "loss": 0.3409,
+      "step": 9008
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3460023506972838e-06,
+      "loss": 0.3371,
+      "step": 9009
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3441884157610463e-06,
+      "loss": 0.3678,
+      "step": 9010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.342375089252219e-06,
+      "loss": 0.3233,
+      "step": 9011
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3405623713149138e-06,
+      "loss": 0.3512,
+      "step": 9012
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.338750262093189e-06,
+      "loss": 0.337,
+      "step": 9013
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3369387617310656e-06,
+      "loss": 0.3674,
+      "step": 9014
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3351278703725e-06,
+      "loss": 0.3252,
+      "step": 9015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3333175881614123e-06,
+      "loss": 0.3094,
+      "step": 9016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3315079152416687e-06,
+      "loss": 0.3099,
+      "step": 9017
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3296988517570883e-06,
+      "loss": 0.4111,
+      "step": 9018
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3278903978514443e-06,
+      "loss": 0.3797,
+      "step": 9019
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3260825536684583e-06,
+      "loss": 0.3386,
+      "step": 9020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.324275319351804e-06,
+      "loss": 0.3081,
+      "step": 9021
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.322468695045108e-06,
+      "loss": 0.331,
+      "step": 9022
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3206626808919462e-06,
+      "loss": 0.3426,
+      "step": 9023
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3188572770358476e-06,
+      "loss": 0.3892,
+      "step": 9024
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3170524836202936e-06,
+      "loss": 0.3277,
+      "step": 9025
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.315248300788714e-06,
+      "loss": 0.3452,
+      "step": 9026
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.313444728684493e-06,
+      "loss": 0.3604,
+      "step": 9027
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3116417674509663e-06,
+      "loss": 0.3075,
+      "step": 9028
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3098394172314175e-06,
+      "loss": 0.3649,
+      "step": 9029
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.308037678169085e-06,
+      "loss": 0.3313,
+      "step": 9030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3062365504071593e-06,
+      "loss": 0.3496,
+      "step": 9031
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3044360340887783e-06,
+      "loss": 0.3148,
+      "step": 9032
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3026361293570386e-06,
+      "loss": 0.3505,
+      "step": 9033
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.300836836354974e-06,
+      "loss": 0.3477,
+      "step": 9034
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2990381552255902e-06,
+      "loss": 0.3496,
+      "step": 9035
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2972400861118216e-06,
+      "loss": 0.3418,
+      "step": 9036
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2954426291565745e-06,
+      "loss": 0.359,
+      "step": 9037
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2936457845026984e-06,
+      "loss": 0.3823,
+      "step": 9038
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.291849552292982e-06,
+      "loss": 0.3348,
+      "step": 9039
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2900539326701908e-06,
+      "loss": 0.3323,
+      "step": 9040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.288258925777014e-06,
+      "loss": 0.3368,
+      "step": 9041
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.286464531756115e-06,
+      "loss": 0.3474,
+      "step": 9042
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2846707507500997e-06,
+      "loss": 0.3153,
+      "step": 9043
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.282877582901515e-06,
+      "loss": 0.3333,
+      "step": 9044
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2810850283528794e-06,
+      "loss": 0.3647,
+      "step": 9045
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2792930872466458e-06,
+      "loss": 0.354,
+      "step": 9046
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2775017597252215e-06,
+      "loss": 0.3754,
+      "step": 9047
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2757110459309784e-06,
+      "loss": 0.3468,
+      "step": 9048
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.273920946006217e-06,
+      "loss": 0.3241,
+      "step": 9049
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.272131460093213e-06,
+      "loss": 0.3167,
+      "step": 9050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2703425883341736e-06,
+      "loss": 0.3507,
+      "step": 9051
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2685543308712665e-06,
+      "loss": 0.3393,
+      "step": 9052
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.266766687846611e-06,
+      "loss": 0.3599,
+      "step": 9053
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2649796594022745e-06,
+      "loss": 0.361,
+      "step": 9054
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2631932456802765e-06,
+      "loss": 0.3422,
+      "step": 9055
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.261407446822589e-06,
+      "loss": 0.3083,
+      "step": 9056
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.259622262971134e-06,
+      "loss": 0.3013,
+      "step": 9057
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.257837694267784e-06,
+      "loss": 0.3185,
+      "step": 9058
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2560537408543646e-06,
+      "loss": 0.3319,
+      "step": 9059
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2542704028726504e-06,
+      "loss": 0.3681,
+      "step": 9060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.252487680464367e-06,
+      "loss": 0.3448,
+      "step": 9061
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2507055737711935e-06,
+      "loss": 0.3503,
+      "step": 9062
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.248924082934759e-06,
+      "loss": 0.3331,
+      "step": 9063
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2471432080966415e-06,
+      "loss": 0.3726,
+      "step": 9064
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.245362949398373e-06,
+      "loss": 0.3383,
+      "step": 9065
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.243583306981435e-06,
+      "loss": 0.3449,
+      "step": 9066
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.241804280987261e-06,
+      "loss": 0.3529,
+      "step": 9067
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.240025871557233e-06,
+      "loss": 0.3572,
+      "step": 9068
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.23824807883269e-06,
+      "loss": 0.3579,
+      "step": 9069
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2364709029549104e-06,
+      "loss": 0.346,
+      "step": 9070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.234694344065137e-06,
+      "loss": 0.3722,
+      "step": 9071
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.232918402304557e-06,
+      "loss": 0.3777,
+      "step": 9072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2311430778143083e-06,
+      "loss": 0.3471,
+      "step": 9073
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229368370735483e-06,
+      "loss": 0.3557,
+      "step": 9074
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2275942812091146e-06,
+      "loss": 0.3606,
+      "step": 9075
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2258208093762037e-06,
+      "loss": 0.3464,
+      "step": 9076
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2240479553776837e-06,
+      "loss": 0.3593,
+      "step": 9077
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2222757193544553e-06,
+      "loss": 0.3306,
+      "step": 9078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2205041014473627e-06,
+      "loss": 0.3428,
+      "step": 9079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2187331017971924e-06,
+      "loss": 0.3312,
+      "step": 9080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.216962720544703e-06,
+      "loss": 0.3355,
+      "step": 9081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2151929578305808e-06,
+      "loss": 0.3287,
+      "step": 9082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2134238137954755e-06,
+      "loss": 0.3402,
+      "step": 9083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2116552885799926e-06,
+      "loss": 0.33,
+      "step": 9084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2098873823246715e-06,
+      "loss": 0.3505,
+      "step": 9085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.208120095170022e-06,
+      "loss": 0.3389,
+      "step": 9086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.206353427256487e-06,
+      "loss": 0.3635,
+      "step": 9087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.20458737872447e-06,
+      "loss": 0.3422,
+      "step": 9088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2028219497143298e-06,
+      "loss": 0.3168,
+      "step": 9089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2010571403663595e-06,
+      "loss": 0.3419,
+      "step": 9090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.199292950820824e-06,
+      "loss": 0.3863,
+      "step": 9091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.19752938121792e-06,
+      "loss": 0.3667,
+      "step": 9092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1957664316978045e-06,
+      "loss": 0.3559,
+      "step": 9093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1940041024005855e-06,
+      "loss": 0.3541,
+      "step": 9094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1922423934663195e-06,
+      "loss": 0.3241,
+      "step": 9095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1904813050350137e-06,
+      "loss": 0.3257,
+      "step": 9096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1887208372466263e-06,
+      "loss": 0.3199,
+      "step": 9097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.186960990241066e-06,
+      "loss": 0.3026,
+      "step": 9098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1852017641581935e-06,
+      "loss": 0.3319,
+      "step": 9099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183443159137819e-06,
+      "loss": 0.3409,
+      "step": 9100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1816851753197023e-06,
+      "loss": 0.3781,
+      "step": 9101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1799278128435565e-06,
+      "loss": 0.3503,
+      "step": 9102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1781710718490425e-06,
+      "loss": 0.3512,
+      "step": 9103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.176414952475774e-06,
+      "loss": 0.3301,
+      "step": 9104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1746594548633136e-06,
+      "loss": 0.3339,
+      "step": 9105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.172904579151176e-06,
+      "loss": 0.3479,
+      "step": 9106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1711503254788245e-06,
+      "loss": 0.3578,
+      "step": 9107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.169396693985677e-06,
+      "loss": 0.3368,
+      "step": 9108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.167643684811096e-06,
+      "loss": 0.3702,
+      "step": 9109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1658912980944025e-06,
+      "loss": 0.3193,
+      "step": 9110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.164139533974855e-06,
+      "loss": 0.3334,
+      "step": 9111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1623883925916776e-06,
+      "loss": 0.3475,
+      "step": 9112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1606378740840363e-06,
+      "loss": 0.3354,
+      "step": 9113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.158887978591049e-06,
+      "loss": 0.3204,
+      "step": 9114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.157138706251788e-06,
+      "loss": 0.332,
+      "step": 9115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1553900572052644e-06,
+      "loss": 0.3027,
+      "step": 9116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.153642031590455e-06,
+      "loss": 0.3188,
+      "step": 9117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1518946295462808e-06,
+      "loss": 0.3418,
+      "step": 9118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1501478512116037e-06,
+      "loss": 0.3921,
+      "step": 9119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1484016967252553e-06,
+      "loss": 0.3162,
+      "step": 9120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.146656166225998e-06,
+      "loss": 0.3309,
+      "step": 9121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1449112598525635e-06,
+      "loss": 0.334,
+      "step": 9122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.143166977743615e-06,
+      "loss": 0.3422,
+      "step": 9123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1414233200377767e-06,
+      "loss": 0.3362,
+      "step": 9124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1396802868736275e-06,
+      "loss": 0.3351,
+      "step": 9125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1379378783896833e-06,
+      "loss": 0.3662,
+      "step": 9126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1361960947244266e-06,
+      "loss": 0.3562,
+      "step": 9127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1344549360162737e-06,
+      "loss": 0.3319,
+      "step": 9128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1327144024035996e-06,
+      "loss": 0.3712,
+      "step": 9129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1309744940247355e-06,
+      "loss": 0.3622,
+      "step": 9130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1292352110179506e-06,
+      "loss": 0.3282,
+      "step": 9131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.127496553521471e-06,
+      "loss": 0.3484,
+      "step": 9132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1257585216734744e-06,
+      "loss": 0.3581,
+      "step": 9133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.124021115612085e-06,
+      "loss": 0.3961,
+      "step": 9134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.12228433547538e-06,
+      "loss": 0.3537,
+      "step": 9135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1205481814013852e-06,
+      "loss": 0.3391,
+      "step": 9136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.118812653528077e-06,
+      "loss": 0.3878,
+      "step": 9137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.117077751993384e-06,
+      "loss": 0.3375,
+      "step": 9138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.115343476935182e-06,
+      "loss": 0.3194,
+      "step": 9139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1136098284912986e-06,
+      "loss": 0.3522,
+      "step": 9140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1118768067995123e-06,
+      "loss": 0.3733,
+      "step": 9141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.11014441199755e-06,
+      "loss": 0.3229,
+      "step": 9142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.108412644223089e-06,
+      "loss": 0.2978,
+      "step": 9143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.106681503613759e-06,
+      "loss": 0.3709,
+      "step": 9144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1049509903071376e-06,
+      "loss": 0.386,
+      "step": 9145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1032211044407514e-06,
+      "loss": 0.3612,
+      "step": 9146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1014918461520826e-06,
+      "loss": 0.319,
+      "step": 9147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0997632155785572e-06,
+      "loss": 0.3342,
+      "step": 9148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0980352128575543e-06,
+      "loss": 0.3205,
+      "step": 9149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0963078381264036e-06,
+      "loss": 0.3294,
+      "step": 9150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0945810915223876e-06,
+      "loss": 0.3268,
+      "step": 9151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.092854973182724e-06,
+      "loss": 0.3193,
+      "step": 9152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0911294832446037e-06,
+      "loss": 0.3593,
+      "step": 9153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0894046218451513e-06,
+      "loss": 0.3344,
+      "step": 9154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.087680389121447e-06,
+      "loss": 0.3573,
+      "step": 9155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.085956785210521e-06,
+      "loss": 0.3297,
+      "step": 9156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0842338102493465e-06,
+      "loss": 0.3513,
+      "step": 9157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0825114643748613e-06,
+      "loss": 0.3571,
+      "step": 9158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0807897477239415e-06,
+      "loss": 0.3313,
+      "step": 9159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.079068660433412e-06,
+      "loss": 0.3557,
+      "step": 9160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0773482026400606e-06,
+      "loss": 0.3458,
+      "step": 9161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0756283744806072e-06,
+      "loss": 0.2962,
+      "step": 9162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.073909176091741e-06,
+      "loss": 0.3689,
+      "step": 9163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.072190607610083e-06,
+      "loss": 0.3665,
+      "step": 9164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0704726691722134e-06,
+      "loss": 0.3406,
+      "step": 9165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0687553609146684e-06,
+      "loss": 0.3849,
+      "step": 9166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0670386829739187e-06,
+      "loss": 0.3372,
+      "step": 9167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0653226354863964e-06,
+      "loss": 0.3683,
+      "step": 9168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0636072185884805e-06,
+      "loss": 0.3023,
+      "step": 9169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.061892432416497e-06,
+      "loss": 0.3745,
+      "step": 9170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0601782771067313e-06,
+      "loss": 0.3659,
+      "step": 9171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.058464752795405e-06,
+      "loss": 0.3434,
+      "step": 9172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0567518596187e-06,
+      "loss": 0.3859,
+      "step": 9173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.055039597712741e-06,
+      "loss": 0.3449,
+      "step": 9174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0533279672136097e-06,
+      "loss": 0.3808,
+      "step": 9175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0516169682573317e-06,
+      "loss": 0.3346,
+      "step": 9176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0499066009798853e-06,
+      "loss": 0.3586,
+      "step": 9177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0481968655171967e-06,
+      "loss": 0.3446,
+      "step": 9178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.046487762005146e-06,
+      "loss": 0.3584,
+      "step": 9179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.044779290579556e-06,
+      "loss": 0.3435,
+      "step": 9180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0430714513762074e-06,
+      "loss": 0.3285,
+      "step": 9181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0413642445308236e-06,
+      "loss": 0.3014,
+      "step": 9182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0396576701790828e-06,
+      "loss": 0.3186,
+      "step": 9183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0379517284566097e-06,
+      "loss": 0.3708,
+      "step": 9184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0362464194989808e-06,
+      "loss": 0.3111,
+      "step": 9185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0345417434417214e-06,
+      "loss": 0.3615,
+      "step": 9186
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.032837700420306e-06,
+      "loss": 0.3425,
+      "step": 9187
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.031134290570159e-06,
+      "loss": 0.3282,
+      "step": 9188
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0294315140266563e-06,
+      "loss": 0.3367,
+      "step": 9189
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.027729370925121e-06,
+      "loss": 0.3659,
+      "step": 9190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0260278614008265e-06,
+      "loss": 0.3674,
+      "step": 9191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0243269855890002e-06,
+      "loss": 0.2912,
+      "step": 9192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.022626743624807e-06,
+      "loss": 0.3667,
+      "step": 9193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0209271356433756e-06,
+      "loss": 0.3175,
+      "step": 9194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0192281617797803e-06,
+      "loss": 0.3285,
+      "step": 9195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0175298221690353e-06,
+      "loss": 0.324,
+      "step": 9196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0158321169461204e-06,
+      "loss": 0.3208,
+      "step": 9197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0141350462459485e-06,
+      "loss": 0.3294,
+      "step": 9198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0124386102033964e-06,
+      "loss": 0.3142,
+      "step": 9199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0107428089532864e-06,
+      "loss": 0.3352,
+      "step": 9200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0090476426303784e-06,
+      "loss": 0.3351,
+      "step": 9201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0073531113694032e-06,
+      "loss": 0.3719,
+      "step": 9202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.005659215305019e-06,
+      "loss": 0.3394,
+      "step": 9203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0039659545718537e-06,
+      "loss": 0.34,
+      "step": 9204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0022733293044694e-06,
+      "loss": 0.3255,
+      "step": 9205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0005813396373806e-06,
+      "loss": 0.3649,
+      "step": 9206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9988899857050647e-06,
+      "loss": 0.3452,
+      "step": 9207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9971992676419295e-06,
+      "loss": 0.3892,
+      "step": 9208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9955091855823427e-06,
+      "loss": 0.3219,
+      "step": 9209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9938197396606196e-06,
+      "loss": 0.3203,
+      "step": 9210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9921309300110226e-06,
+      "loss": 0.3592,
+      "step": 9211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9904427567677732e-06,
+      "loss": 0.3386,
+      "step": 9212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9887552200650272e-06,
+      "loss": 0.3383,
+      "step": 9213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.987068320036902e-06,
+      "loss": 0.3501,
+      "step": 9214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9853820568174577e-06,
+      "loss": 0.3677,
+      "step": 9215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.983696430540707e-06,
+      "loss": 0.3747,
+      "step": 9216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98201144134061e-06,
+      "loss": 0.3336,
+      "step": 9217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9803270893510797e-06,
+      "loss": 0.3528,
+      "step": 9218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978643374705973e-06,
+      "loss": 0.3303,
+      "step": 9219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9769602975391002e-06,
+      "loss": 0.3688,
+      "step": 9220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9752778579842214e-06,
+      "loss": 0.3595,
+      "step": 9221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973596056175042e-06,
+      "loss": 0.3495,
+      "step": 9222
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9719148922452215e-06,
+      "loss": 0.3411,
+      "step": 9223
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.970234366328364e-06,
+      "loss": 0.308,
+      "step": 9224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9685544785580278e-06,
+      "loss": 0.3478,
+      "step": 9225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.966875229067716e-06,
+      "loss": 0.3561,
+      "step": 9226
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9651966179908843e-06,
+      "loss": 0.3012,
+      "step": 9227
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9635186454609355e-06,
+      "loss": 0.3346,
+      "step": 9228
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9618413116112224e-06,
+      "loss": 0.3506,
+      "step": 9229
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9601646165750476e-06,
+      "loss": 0.3251,
+      "step": 9230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.958488560485663e-06,
+      "loss": 0.3786,
+      "step": 9231
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956813143476267e-06,
+      "loss": 0.3477,
+      "step": 9232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.955138365680015e-06,
+      "loss": 0.3209,
+      "step": 9233
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9534642272299965e-06,
+      "loss": 0.2799,
+      "step": 9234
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951790728259266e-06,
+      "loss": 0.3326,
+      "step": 9235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.950117868900824e-06,
+      "loss": 0.3724,
+      "step": 9236
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.948445649287607e-06,
+      "loss": 0.3225,
+      "step": 9237
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.946774069552523e-06,
+      "loss": 0.3478,
+      "step": 9238
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945103129828403e-06,
+      "loss": 0.3031,
+      "step": 9239
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.943432830248052e-06,
+      "loss": 0.3497,
+      "step": 9240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.941763170944212e-06,
+      "loss": 0.3319,
+      "step": 9241
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9400941520495673e-06,
+      "loss": 0.3588,
+      "step": 9242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9384257736967703e-06,
+      "loss": 0.3526,
+      "step": 9243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9367580360184023e-06,
+      "loss": 0.3408,
+      "step": 9244
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.935090939147006e-06,
+      "loss": 0.3167,
+      "step": 9245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.933424483215072e-06,
+      "loss": 0.3564,
+      "step": 9246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9317586683550303e-06,
+      "loss": 0.3739,
+      "step": 9247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.930093494699281e-06,
+      "loss": 0.3295,
+      "step": 9248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.928428962380148e-06,
+      "loss": 0.327,
+      "step": 9249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9267650715299203e-06,
+      "loss": 0.3794,
+      "step": 9250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9251018222808317e-06,
+      "loss": 0.3618,
+      "step": 9251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.923439214765063e-06,
+      "loss": 0.3235,
+      "step": 9252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9217772491147513e-06,
+      "loss": 0.2909,
+      "step": 9253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9201159254619727e-06,
+      "loss": 0.3128,
+      "step": 9254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.918455243938757e-06,
+      "loss": 0.324,
+      "step": 9255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9167952046770842e-06,
+      "loss": 0.3474,
+      "step": 9256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9151358078088822e-06,
+      "loss": 0.3377,
+      "step": 9257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.913477053466026e-06,
+      "loss": 0.3732,
+      "step": 9258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9118189417803436e-06,
+      "loss": 0.3636,
+      "step": 9259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910161472883607e-06,
+      "loss": 0.3864,
+      "step": 9260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9085046469075417e-06,
+      "loss": 0.3689,
+      "step": 9261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9068484639838193e-06,
+      "loss": 0.3151,
+      "step": 9262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9051929242440604e-06,
+      "loss": 0.3838,
+      "step": 9263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9035380278198346e-06,
+      "loss": 0.3221,
+      "step": 9264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9018837748426633e-06,
+      "loss": 0.3515,
+      "step": 9265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9002301654440114e-06,
+      "loss": 0.3514,
+      "step": 9266
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.898577199755297e-06,
+      "loss": 0.3614,
+      "step": 9267
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8969248779078865e-06,
+      "loss": 0.3431,
+      "step": 9268
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.895273200033092e-06,
+      "loss": 0.3846,
+      "step": 9269
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.893622166262179e-06,
+      "loss": 0.3438,
+      "step": 9270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8919717767263578e-06,
+      "loss": 0.3239,
+      "step": 9271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8903220315567928e-06,
+      "loss": 0.3855,
+      "step": 9272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8886729308845852e-06,
+      "loss": 0.3516,
+      "step": 9273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.887024474840804e-06,
+      "loss": 0.3326,
+      "step": 9274
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8853766635564464e-06,
+      "loss": 0.3431,
+      "step": 9275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8837294971624743e-06,
+      "loss": 0.3874,
+      "step": 9276
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8820829757897952e-06,
+      "loss": 0.3782,
+      "step": 9277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.880437099569251e-06,
+      "loss": 0.3555,
+      "step": 9278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8787918686316575e-06,
+      "loss": 0.3466,
+      "step": 9279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8771472831077541e-06,
+      "loss": 0.3246,
+      "step": 9280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8755033431282476e-06,
+      "loss": 0.3641,
+      "step": 9281
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8738600488237858e-06,
+      "loss": 0.3069,
+      "step": 9282
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8722174003249594e-06,
+      "loss": 0.3621,
+      "step": 9283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.870575397762323e-06,
+      "loss": 0.3621,
+      "step": 9284
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8689340412663636e-06,
+      "loss": 0.3323,
+      "step": 9285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8672933309675234e-06,
+      "loss": 0.3406,
+      "step": 9286
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8656532669962036e-06,
+      "loss": 0.3037,
+      "step": 9287
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.864013849482732e-06,
+      "loss": 0.3495,
+      "step": 9288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.862375078557408e-06,
+      "loss": 0.3117,
+      "step": 9289
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8607369543504627e-06,
+      "loss": 0.3292,
+      "step": 9290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8590994769920834e-06,
+      "loss": 0.3464,
+      "step": 9291
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8574626466124057e-06,
+      "loss": 0.3014,
+      "step": 9292
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8558264633415112e-06,
+      "loss": 0.3668,
+      "step": 9293
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8541909273094338e-06,
+      "loss": 0.3295,
+      "step": 9294
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8525560386461528e-06,
+      "loss": 0.3654,
+      "step": 9295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8509217974815964e-06,
+      "loss": 0.3473,
+      "step": 9296
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8492882039456417e-06,
+      "loss": 0.3466,
+      "step": 9297
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8476552581681172e-06,
+      "loss": 0.3807,
+      "step": 9298
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8460229602787948e-06,
+      "loss": 0.3602,
+      "step": 9299
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8443913104073984e-06,
+      "loss": 0.3525,
+      "step": 9300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.842760308683601e-06,
+      "loss": 0.346,
+      "step": 9301
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.84112995523702e-06,
+      "loss": 0.3123,
+      "step": 9302
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8395002501972248e-06,
+      "loss": 0.32,
+      "step": 9303
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8378711936937331e-06,
+      "loss": 0.3352,
+      "step": 9304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8362427858560094e-06,
+      "loss": 0.3161,
+      "step": 9305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8346150268134676e-06,
+      "loss": 0.3464,
+      "step": 9306
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.832987916695471e-06,
+      "loss": 0.3205,
+      "step": 9307
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.831361455631332e-06,
+      "loss": 0.3344,
+      "step": 9308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8297356437503023e-06,
+      "loss": 0.3889,
+      "step": 9309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8281104811815964e-06,
+      "loss": 0.3432,
+      "step": 9310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8264859680543678e-06,
+      "loss": 0.3084,
+      "step": 9311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8248621044977221e-06,
+      "loss": 0.3121,
+      "step": 9312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.823238890640714e-06,
+      "loss": 0.3214,
+      "step": 9313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8216163266123354e-06,
+      "loss": 0.3467,
+      "step": 9314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8199944125415459e-06,
+      "loss": 0.3454,
+      "step": 9315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8183731485572387e-06,
+      "loss": 0.3545,
+      "step": 9316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8167525347882598e-06,
+      "loss": 0.3179,
+      "step": 9317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8151325713634084e-06,
+      "loss": 0.34,
+      "step": 9318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8135132584114167e-06,
+      "loss": 0.3832,
+      "step": 9319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.811894596060989e-06,
+      "loss": 0.3541,
+      "step": 9320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8102765844407545e-06,
+      "loss": 0.3609,
+      "step": 9321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.808659223679301e-06,
+      "loss": 0.3364,
+      "step": 9322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.807042513905174e-06,
+      "loss": 0.3408,
+      "step": 9323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.805426455246846e-06,
+      "loss": 0.3246,
+      "step": 9324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8038110478327597e-06,
+      "loss": 0.3832,
+      "step": 9325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8021962917912883e-06,
+      "loss": 0.3348,
+      "step": 9326
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8005821872507601e-06,
+      "loss": 0.3274,
+      "step": 9327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7989687343394602e-06,
+      "loss": 0.381,
+      "step": 9328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7973559331856049e-06,
+      "loss": 0.3637,
+      "step": 9329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7957437839173764e-06,
+      "loss": 0.3229,
+      "step": 9330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7941322866628896e-06,
+      "loss": 0.338,
+      "step": 9331
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7925214415502169e-06,
+      "loss": 0.3368,
+      "step": 9332
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7909112487073754e-06,
+      "loss": 0.3479,
+      "step": 9333
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7893017082623331e-06,
+      "loss": 0.3553,
+      "step": 9334
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7876928203430022e-06,
+      "loss": 0.3648,
+      "step": 9335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.786084585077248e-06,
+      "loss": 0.3212,
+      "step": 9336
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7844770025928782e-06,
+      "loss": 0.3218,
+      "step": 9337
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.782870073017654e-06,
+      "loss": 0.3218,
+      "step": 9338
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.781263796479281e-06,
+      "loss": 0.3407,
+      "step": 9339
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7796581731054152e-06,
+      "loss": 0.3402,
+      "step": 9340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7780532030236576e-06,
+      "loss": 0.3366,
+      "step": 9341
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7764488863615625e-06,
+      "loss": 0.3267,
+      "step": 9342
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7748452232466263e-06,
+      "loss": 0.3616,
+      "step": 9343
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7732422138062976e-06,
+      "loss": 0.3406,
+      "step": 9344
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7716398581679718e-06,
+      "loss": 0.3496,
+      "step": 9345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7700381564589918e-06,
+      "loss": 0.3257,
+      "step": 9346
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7684371088066509e-06,
+      "loss": 0.316,
+      "step": 9347
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.766836715338186e-06,
+      "loss": 0.3793,
+      "step": 9348
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7652369761807875e-06,
+      "loss": 0.3465,
+      "step": 9349
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7636378914615847e-06,
+      "loss": 0.3647,
+      "step": 9350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7620394613076675e-06,
+      "loss": 0.3187,
+      "step": 9351
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7604416858460648e-06,
+      "loss": 0.3135,
+      "step": 9352
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7588445652037578e-06,
+      "loss": 0.3479,
+      "step": 9353
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7572480995076734e-06,
+      "loss": 0.3456,
+      "step": 9354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7556522888846816e-06,
+      "loss": 0.3734,
+      "step": 9355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.754057133461613e-06,
+      "loss": 0.3682,
+      "step": 9356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7524626333652373e-06,
+      "loss": 0.35,
+      "step": 9357
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7508687887222674e-06,
+      "loss": 0.3292,
+      "step": 9358
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7492755996593802e-06,
+      "loss": 0.3737,
+      "step": 9359
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7476830663031796e-06,
+      "loss": 0.3693,
+      "step": 9360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.74609118878024e-06,
+      "loss": 0.33,
+      "step": 9361
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7444999672170638e-06,
+      "loss": 0.3436,
+      "step": 9362
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7429094017401094e-06,
+      "loss": 0.3472,
+      "step": 9363
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7413194924757904e-06,
+      "loss": 0.3879,
+      "step": 9364
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.739730239550451e-06,
+      "loss": 0.3375,
+      "step": 9365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7381416430904052e-06,
+      "loss": 0.3415,
+      "step": 9366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7365537032218927e-06,
+      "loss": 0.3561,
+      "step": 9367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7349664200711136e-06,
+      "loss": 0.3776,
+      "step": 9368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7333797937642195e-06,
+      "loss": 0.3215,
+      "step": 9369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.731793824427297e-06,
+      "loss": 0.3397,
+      "step": 9370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7302085121863887e-06,
+      "loss": 0.353,
+      "step": 9371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7286238571674841e-06,
+      "loss": 0.3317,
+      "step": 9372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.727039859496521e-06,
+      "loss": 0.3642,
+      "step": 9373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7254565192993823e-06,
+      "loss": 0.3465,
+      "step": 9374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7238738367019003e-06,
+      "loss": 0.3733,
+      "step": 9375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.722291811829856e-06,
+      "loss": 0.3804,
+      "step": 9376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7207104448089763e-06,
+      "loss": 0.3498,
+      "step": 9377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7191297357649361e-06,
+      "loss": 0.3162,
+      "step": 9378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7175496848233597e-06,
+      "loss": 0.3493,
+      "step": 9379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7159702921098165e-06,
+      "loss": 0.3295,
+      "step": 9380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.714391557749827e-06,
+      "loss": 0.3503,
+      "step": 9381
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7128134818688557e-06,
+      "loss": 0.338,
+      "step": 9382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7112360645923166e-06,
+      "loss": 0.3287,
+      "step": 9383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7096593060455713e-06,
+      "loss": 0.3135,
+      "step": 9384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7080832063539288e-06,
+      "loss": 0.3198,
+      "step": 9385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7065077656426477e-06,
+      "loss": 0.3588,
+      "step": 9386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704932984036929e-06,
+      "loss": 0.3448,
+      "step": 9387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703358861661928e-06,
+      "loss": 0.3611,
+      "step": 9388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7017853986427423e-06,
+      "loss": 0.345,
+      "step": 9389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7002125951044213e-06,
+      "loss": 0.3635,
+      "step": 9390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6986404511719545e-06,
+      "loss": 0.3536,
+      "step": 9391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.69706896697029e-06,
+      "loss": 0.3337,
+      "step": 9392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6954981426243166e-06,
+      "loss": 0.3624,
+      "step": 9393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6939279782588703e-06,
+      "loss": 0.3139,
+      "step": 9394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6923584739987398e-06,
+      "loss": 0.318,
+      "step": 9395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6907896299686488e-06,
+      "loss": 0.3342,
+      "step": 9396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6892214462932876e-06,
+      "loss": 0.3242,
+      "step": 9397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.687653923097281e-06,
+      "loss": 0.3615,
+      "step": 9398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6860870605051983e-06,
+      "loss": 0.3686,
+      "step": 9399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6845208586415718e-06,
+      "loss": 0.3267,
+      "step": 9400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6829553176308621e-06,
+      "loss": 0.3315,
+      "step": 9401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6813904375974965e-06,
+      "loss": 0.3762,
+      "step": 9402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.679826218665832e-06,
+      "loss": 0.3306,
+      "step": 9403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6782626609601815e-06,
+      "loss": 0.3769,
+      "step": 9404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6766997646048133e-06,
+      "loss": 0.3313,
+      "step": 9405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6751375297239257e-06,
+      "loss": 0.3469,
+      "step": 9406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6735759564416787e-06,
+      "loss": 0.3192,
+      "step": 9407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6720150448821725e-06,
+      "loss": 0.3594,
+      "step": 9408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.670454795169455e-06,
+      "loss": 0.3845,
+      "step": 9409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.668895207427531e-06,
+      "loss": 0.3225,
+      "step": 9410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6673362817803373e-06,
+      "loss": 0.4137,
+      "step": 9411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.665778018351768e-06,
+      "loss": 0.3515,
+      "step": 9412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.664220417265663e-06,
+      "loss": 0.3537,
+      "step": 9413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6626634786458085e-06,
+      "loss": 0.3546,
+      "step": 9414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6611072026159392e-06,
+      "loss": 0.3226,
+      "step": 9415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6595515892997349e-06,
+      "loss": 0.3545,
+      "step": 9416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6579966388208257e-06,
+      "loss": 0.354,
+      "step": 9417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.656442351302787e-06,
+      "loss": 0.3662,
+      "step": 9418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6548887268691426e-06,
+      "loss": 0.3511,
+      "step": 9419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6533357656433623e-06,
+      "loss": 0.3549,
+      "step": 9420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6517834677488653e-06,
+      "loss": 0.3654,
+      "step": 9421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.650231833309015e-06,
+      "loss": 0.3372,
+      "step": 9422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6486808624471252e-06,
+      "loss": 0.3528,
+      "step": 9423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6471305552864547e-06,
+      "loss": 0.3393,
+      "step": 9424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.645580911950212e-06,
+      "loss": 0.3523,
+      "step": 9425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6440319325615494e-06,
+      "loss": 0.3502,
+      "step": 9426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.64248361724357e-06,
+      "loss": 0.3379,
+      "step": 9427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6409359661193203e-06,
+      "loss": 0.3085,
+      "step": 9428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.639388979311799e-06,
+      "loss": 0.3509,
+      "step": 9429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6378426569439477e-06,
+      "loss": 0.3933,
+      "step": 9430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.636296999138659e-06,
+      "loss": 0.3284,
+      "step": 9431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6347520060187628e-06,
+      "loss": 0.3613,
+      "step": 9432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6332076777070515e-06,
+      "loss": 0.3456,
+      "step": 9433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6316640143262575e-06,
+      "loss": 0.3553,
+      "step": 9434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6301210159990522e-06,
+      "loss": 0.3287,
+      "step": 9435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.628578682848071e-06,
+      "loss": 0.3421,
+      "step": 9436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.627037014995878e-06,
+      "loss": 0.3632,
+      "step": 9437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6254960125649999e-06,
+      "loss": 0.3358,
+      "step": 9438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6239556756779051e-06,
+      "loss": 0.3633,
+      "step": 9439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6224160044570003e-06,
+      "loss": 0.352,
+      "step": 9440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6208769990246575e-06,
+      "loss": 0.3471,
+      "step": 9441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6193386595031757e-06,
+      "loss": 0.3433,
+      "step": 9442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6178009860148203e-06,
+      "loss": 0.2877,
+      "step": 9443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6162639786817869e-06,
+      "loss": 0.3255,
+      "step": 9444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6147276376262256e-06,
+      "loss": 0.3514,
+      "step": 9445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6131919629702408e-06,
+      "loss": 0.317,
+      "step": 9446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6116569548358685e-06,
+      "loss": 0.3834,
+      "step": 9447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6101226133451043e-06,
+      "loss": 0.3355,
+      "step": 9448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.608588938619884e-06,
+      "loss": 0.3303,
+      "step": 9449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.607055930782091e-06,
+      "loss": 0.3162,
+      "step": 9450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.605523589953566e-06,
+      "loss": 0.3289,
+      "step": 9451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6039919162560802e-06,
+      "loss": 0.3767,
+      "step": 9452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6024609098113609e-06,
+      "loss": 0.3755,
+      "step": 9453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6009305707410828e-06,
+      "loss": 0.3067,
+      "step": 9454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5994008991668654e-06,
+      "loss": 0.368,
+      "step": 9455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5978718952102744e-06,
+      "loss": 0.3901,
+      "step": 9456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5963435589928267e-06,
+      "loss": 0.3278,
+      "step": 9457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.59481589063598e-06,
+      "loss": 0.3513,
+      "step": 9458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5932888902611453e-06,
+      "loss": 0.3131,
+      "step": 9459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5917625579896745e-06,
+      "loss": 0.3381,
+      "step": 9460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5902368939428702e-06,
+      "loss": 0.3869,
+      "step": 9461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5887118982419813e-06,
+      "loss": 0.367,
+      "step": 9462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587187571008203e-06,
+      "loss": 0.3368,
+      "step": 9463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5856639123626771e-06,
+      "loss": 0.3756,
+      "step": 9464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5841409224264936e-06,
+      "loss": 0.3398,
+      "step": 9465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.582618601320689e-06,
+      "loss": 0.3473,
+      "step": 9466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5810969491662442e-06,
+      "loss": 0.3595,
+      "step": 9467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5795759660840904e-06,
+      "loss": 0.3431,
+      "step": 9468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5780556521951029e-06,
+      "loss": 0.3632,
+      "step": 9469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5765360076201085e-06,
+      "loss": 0.3357,
+      "step": 9470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5750170324798697e-06,
+      "loss": 0.3757,
+      "step": 9471
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.573498726895114e-06,
+      "loss": 0.3043,
+      "step": 9472
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5719810909864941e-06,
+      "loss": 0.3664,
+      "step": 9473
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5704641248746277e-06,
+      "loss": 0.3181,
+      "step": 9474
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.568947828680073e-06,
+      "loss": 0.3373,
+      "step": 9475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5674322025233268e-06,
+      "loss": 0.3455,
+      "step": 9476
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5659172465248485e-06,
+      "loss": 0.3094,
+      "step": 9477
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.564402960805027e-06,
+      "loss": 0.3137,
+      "step": 9478
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5628893454842143e-06,
+      "loss": 0.3099,
+      "step": 9479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5613764006826992e-06,
+      "loss": 0.3516,
+      "step": 9480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5598641265207137e-06,
+      "loss": 0.3216,
+      "step": 9481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5583525231184514e-06,
+      "loss": 0.3375,
+      "step": 9482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5568415905960378e-06,
+      "loss": 0.3172,
+      "step": 9483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5553313290735506e-06,
+      "loss": 0.3466,
+      "step": 9484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5538217386710152e-06,
+      "loss": 0.3641,
+      "step": 9485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5523128195084004e-06,
+      "loss": 0.37,
+      "step": 9486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.550804571705632e-06,
+      "loss": 0.3081,
+      "step": 9487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5492969953825653e-06,
+      "loss": 0.3567,
+      "step": 9488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5477900906590148e-06,
+      "loss": 0.3488,
+      "step": 9489
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.546283857654739e-06,
+      "loss": 0.3721,
+      "step": 9490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5447782964894386e-06,
+      "loss": 0.328,
+      "step": 9491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5432734072827715e-06,
+      "loss": 0.3305,
+      "step": 9492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5417691901543286e-06,
+      "loss": 0.3484,
+      "step": 9493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5402656452236575e-06,
+      "loss": 0.3562,
+      "step": 9494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.538762772610246e-06,
+      "loss": 0.3203,
+      "step": 9495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5372605724335343e-06,
+      "loss": 0.3847,
+      "step": 9496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5357590448129046e-06,
+      "loss": 0.3333,
+      "step": 9497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5342581898676867e-06,
+      "loss": 0.3023,
+      "step": 9498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5327580077171589e-06,
+      "loss": 0.384,
+      "step": 9499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5312584984805445e-06,
+      "loss": 0.3329,
+      "step": 9500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5297596622770117e-06,
+      "loss": 0.3315,
+      "step": 9501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5282614992256795e-06,
+      "loss": 0.3671,
+      "step": 9502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5267640094456083e-06,
+      "loss": 0.3458,
+      "step": 9503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5252671930558094e-06,
+      "loss": 0.3593,
+      "step": 9504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.523771050175238e-06,
+      "loss": 0.356,
+      "step": 9505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5222755809227962e-06,
+      "loss": 0.3324,
+      "step": 9506
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.520780785417334e-06,
+      "loss": 0.3174,
+      "step": 9507
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5192866637776449e-06,
+      "loss": 0.3642,
+      "step": 9508
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.517793216122473e-06,
+      "loss": 0.3395,
+      "step": 9509
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5163004425705041e-06,
+      "loss": 0.3147,
+      "step": 9510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.514808343240376e-06,
+      "loss": 0.361,
+      "step": 9511
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5133169182506635e-06,
+      "loss": 0.3158,
+      "step": 9512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.511826167719903e-06,
+      "loss": 0.2979,
+      "step": 9513
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5103360917665599e-06,
+      "loss": 0.3397,
+      "step": 9514
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5088466905090592e-06,
+      "loss": 0.3341,
+      "step": 9515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5073579640657687e-06,
+      "loss": 0.3495,
+      "step": 9516
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5058699125549948e-06,
+      "loss": 0.3396,
+      "step": 9517
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5043825360950059e-06,
+      "loss": 0.3263,
+      "step": 9518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.502895834804e-06,
+      "loss": 0.3196,
+      "step": 9519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5014098088001305e-06,
+      "loss": 0.3509,
+      "step": 9520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4999244582015026e-06,
+      "loss": 0.3613,
+      "step": 9521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.498439783126151e-06,
+      "loss": 0.3301,
+      "step": 9522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4969557836920757e-06,
+      "loss": 0.3072,
+      "step": 9523
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4954724600172076e-06,
+      "loss": 0.3396,
+      "step": 9524
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4939898122194297e-06,
+      "loss": 0.308,
+      "step": 9525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4925078404165805e-06,
+      "loss": 0.3183,
+      "step": 9526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.491026544726425e-06,
+      "loss": 0.3289,
+      "step": 9527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4895459252666955e-06,
+      "loss": 0.3184,
+      "step": 9528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4880659821550547e-06,
+      "loss": 0.3568,
+      "step": 9529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4865867155091196e-06,
+      "loss": 0.4064,
+      "step": 9530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4851081254464506e-06,
+      "loss": 0.344,
+      "step": 9531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4836302120845559e-06,
+      "loss": 0.339,
+      "step": 9532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.482152975540888e-06,
+      "loss": 0.3335,
+      "step": 9533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.480676415932848e-06,
+      "loss": 0.358,
+      "step": 9534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4792005333777804e-06,
+      "loss": 0.3025,
+      "step": 9535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4777253279929805e-06,
+      "loss": 0.3422,
+      "step": 9536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4762507998956832e-06,
+      "loss": 0.3098,
+      "step": 9537
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.474776949203075e-06,
+      "loss": 0.3705,
+      "step": 9538
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4733037760322866e-06,
+      "loss": 0.3633,
+      "step": 9539
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4718312805003953e-06,
+      "loss": 0.4168,
+      "step": 9540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4703594627244222e-06,
+      "loss": 0.388,
+      "step": 9541
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.468888322821339e-06,
+      "loss": 0.3684,
+      "step": 9542
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4674178609080603e-06,
+      "loss": 0.3312,
+      "step": 9543
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4659480771014456e-06,
+      "loss": 0.3362,
+      "step": 9544
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.464478971518305e-06,
+      "loss": 0.3745,
+      "step": 9545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4630105442753917e-06,
+      "loss": 0.3515,
+      "step": 9546
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4615427954894068e-06,
+      "loss": 0.3253,
+      "step": 9547
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4600757252769903e-06,
+      "loss": 0.3578,
+      "step": 9548
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4586093337547392e-06,
+      "loss": 0.3068,
+      "step": 9549
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4571436210391922e-06,
+      "loss": 0.3522,
+      "step": 9550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4556785872468305e-06,
+      "loss": 0.4007,
+      "step": 9551
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4542142324940878e-06,
+      "loss": 0.3358,
+      "step": 9552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4527505568973322e-06,
+      "loss": 0.2935,
+      "step": 9553
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.451287560572895e-06,
+      "loss": 0.3367,
+      "step": 9554
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4498252436370398e-06,
+      "loss": 0.3208,
+      "step": 9555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4483636062059814e-06,
+      "loss": 0.3451,
+      "step": 9556
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4469026483958837e-06,
+      "loss": 0.359,
+      "step": 9557
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4454423703228438e-06,
+      "loss": 0.3621,
+      "step": 9558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4439827721029232e-06,
+      "loss": 0.3776,
+      "step": 9559
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4425238538521147e-06,
+      "loss": 0.3421,
+      "step": 9560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4410656156863622e-06,
+      "loss": 0.3444,
+      "step": 9561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.439608057721561e-06,
+      "loss": 0.3786,
+      "step": 9562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4381511800735381e-06,
+      "loss": 0.329,
+      "step": 9563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4366949828580867e-06,
+      "loss": 0.3129,
+      "step": 9564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4352394661909253e-06,
+      "loss": 0.3371,
+      "step": 9565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.433784630187729e-06,
+      "loss": 0.3473,
+      "step": 9566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4323304749641243e-06,
+      "loss": 0.3699,
+      "step": 9567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4308770006356687e-06,
+      "loss": 0.365,
+      "step": 9568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4294242073178765e-06,
+      "loss": 0.3676,
+      "step": 9569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4279720951262054e-06,
+      "loss": 0.3435,
+      "step": 9570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4265206641760588e-06,
+      "loss": 0.3296,
+      "step": 9571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4250699145827852e-06,
+      "loss": 0.3387,
+      "step": 9572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4236198464616779e-06,
+      "loss": 0.3172,
+      "step": 9573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4221704599279807e-06,
+      "loss": 0.384,
+      "step": 9574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4207217550968788e-06,
+      "loss": 0.3405,
+      "step": 9575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4192737320835038e-06,
+      "loss": 0.3429,
+      "step": 9576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4178263910029345e-06,
+      "loss": 0.3578,
+      "step": 9577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4163797319701955e-06,
+      "loss": 0.303,
+      "step": 9578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4149337551002552e-06,
+      "loss": 0.3528,
+      "step": 9579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4134884605080302e-06,
+      "loss": 0.3562,
+      "step": 9580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.412043848308382e-06,
+      "loss": 0.3385,
+      "step": 9581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4105999186161167e-06,
+      "loss": 0.3363,
+      "step": 9582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4091566715459882e-06,
+      "loss": 0.3378,
+      "step": 9583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4077141072126943e-06,
+      "loss": 0.3031,
+      "step": 9584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4062722257308803e-06,
+      "loss": 0.3343,
+      "step": 9585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4048310272151354e-06,
+      "loss": 0.3773,
+      "step": 9586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4033905117799962e-06,
+      "loss": 0.3389,
+      "step": 9587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4019506795399462e-06,
+      "loss": 0.3467,
+      "step": 9588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4005115306094053e-06,
+      "loss": 0.3589,
+      "step": 9589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3990730651027539e-06,
+      "loss": 0.313,
+      "step": 9590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3976352831343087e-06,
+      "loss": 0.3074,
+      "step": 9591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3961981848183326e-06,
+      "loss": 0.3306,
+      "step": 9592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3947617702690386e-06,
+      "loss": 0.3363,
+      "step": 9593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3933260396005766e-06,
+      "loss": 0.3466,
+      "step": 9594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.391890992927053e-06,
+      "loss": 0.3066,
+      "step": 9595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3904566303625156e-06,
+      "loss": 0.3311,
+      "step": 9596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3890229520209497e-06,
+      "loss": 0.3645,
+      "step": 9597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3875899580163033e-06,
+      "loss": 0.3315,
+      "step": 9598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3861576484624506e-06,
+      "loss": 0.2961,
+      "step": 9599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3847260234732296e-06,
+      "loss": 0.3586,
+      "step": 9600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3832950831624093e-06,
+      "loss": 0.3438,
+      "step": 9601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3818648276437097e-06,
+      "loss": 0.3409,
+      "step": 9602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3804352570308043e-06,
+      "loss": 0.3446,
+      "step": 9603
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.379006371437296e-06,
+      "loss": 0.3401,
+      "step": 9604
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3775781709767488e-06,
+      "loss": 0.3367,
+      "step": 9605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3761506557626624e-06,
+      "loss": 0.3562,
+      "step": 9606
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3747238259084827e-06,
+      "loss": 0.3416,
+      "step": 9607
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3732976815276111e-06,
+      "loss": 0.3487,
+      "step": 9608
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3718722227333804e-06,
+      "loss": 0.3548,
+      "step": 9609
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3704474496390774e-06,
+      "loss": 0.355,
+      "step": 9610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.369023362357932e-06,
+      "loss": 0.3358,
+      "step": 9611
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3675999610031222e-06,
+      "loss": 0.3464,
+      "step": 9612
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3661772456877675e-06,
+      "loss": 0.3792,
+      "step": 9613
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3647552165249355e-06,
+      "loss": 0.3311,
+      "step": 9614
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.363333873627638e-06,
+      "loss": 0.3534,
+      "step": 9615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3619132171088344e-06,
+      "loss": 0.3242,
+      "step": 9616
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3604932470814258e-06,
+      "loss": 0.3868,
+      "step": 9617
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.359073963658263e-06,
+      "loss": 0.3166,
+      "step": 9618
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3576553669521385e-06,
+      "loss": 0.3026,
+      "step": 9619
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.356237457075793e-06,
+      "loss": 0.3691,
+      "step": 9620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3548202341419115e-06,
+      "loss": 0.3533,
+      "step": 9621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.353403698263125e-06,
+      "loss": 0.2991,
+      "step": 9622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3519878495520067e-06,
+      "loss": 0.3366,
+      "step": 9623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.35057268812108e-06,
+      "loss": 0.3276,
+      "step": 9624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3491582140828118e-06,
+      "loss": 0.3633,
+      "step": 9625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.347744427549612e-06,
+      "loss": 0.3843,
+      "step": 9626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.346331328633841e-06,
+      "loss": 0.3856,
+      "step": 9627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3449189174477983e-06,
+      "loss": 0.382,
+      "step": 9628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3435071941037359e-06,
+      "loss": 0.3113,
+      "step": 9629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3420961587138403e-06,
+      "loss": 0.3311,
+      "step": 9630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3406858113902554e-06,
+      "loss": 0.331,
+      "step": 9631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3392761522450658e-06,
+      "loss": 0.3354,
+      "step": 9632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3378671813902988e-06,
+      "loss": 0.3378,
+      "step": 9633
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3364588989379312e-06,
+      "loss": 0.3408,
+      "step": 9634
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.335051304999877e-06,
+      "loss": 0.3427,
+      "step": 9635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3336443996880076e-06,
+      "loss": 0.3061,
+      "step": 9636
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.332238183114134e-06,
+      "loss": 0.3463,
+      "step": 9637
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3308326553900053e-06,
+      "loss": 0.3405,
+      "step": 9638
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.32942781662733e-06,
+      "loss": 0.3357,
+      "step": 9639
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3280236669377467e-06,
+      "loss": 0.3568,
+      "step": 9640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3266202064328548e-06,
+      "loss": 0.3727,
+      "step": 9641
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3252174352241864e-06,
+      "loss": 0.3057,
+      "step": 9642
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.32381535342322e-06,
+      "loss": 0.3236,
+      "step": 9643
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3224139611413922e-06,
+      "loss": 0.3497,
+      "step": 9644
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.321013258490068e-06,
+      "loss": 0.3215,
+      "step": 9645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3196132455805667e-06,
+      "loss": 0.356,
+      "step": 9646
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3182139225241508e-06,
+      "loss": 0.3392,
+      "step": 9647
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3168152894320274e-06,
+      "loss": 0.3576,
+      "step": 9648
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3154173464153542e-06,
+      "loss": 0.3971,
+      "step": 9649
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3140200935852233e-06,
+      "loss": 0.38,
+      "step": 9650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3126235310526813e-06,
+      "loss": 0.3465,
+      "step": 9651
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3112276589287153e-06,
+      "loss": 0.3289,
+      "step": 9652
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.309832477324261e-06,
+      "loss": 0.3244,
+      "step": 9653
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.308437986350194e-06,
+      "loss": 0.3384,
+      "step": 9654
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.307044186117341e-06,
+      "loss": 0.316,
+      "step": 9655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3056510767364693e-06,
+      "loss": 0.3603,
+      "step": 9656
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3042586583182948e-06,
+      "loss": 0.3512,
+      "step": 9657
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3028669309734744e-06,
+      "loss": 0.328,
+      "step": 9658
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3014758948126127e-06,
+      "loss": 0.369,
+      "step": 9659
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3000855499462606e-06,
+      "loss": 0.3076,
+      "step": 9660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2986958964849106e-06,
+      "loss": 0.3563,
+      "step": 9661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2973069345390033e-06,
+      "loss": 0.3058,
+      "step": 9662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2959186642189225e-06,
+      "loss": 0.3585,
+      "step": 9663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.294531085634999e-06,
+      "loss": 0.3418,
+      "step": 9664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2931441988975047e-06,
+      "loss": 0.3643,
+      "step": 9665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2917580041166623e-06,
+      "loss": 0.3384,
+      "step": 9666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2903725014026336e-06,
+      "loss": 0.3463,
+      "step": 9667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2889876908655297e-06,
+      "loss": 0.3506,
+      "step": 9668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2876035726154046e-06,
+      "loss": 0.3345,
+      "step": 9669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2862201467622592e-06,
+      "loss": 0.3277,
+      "step": 9670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2848374134160336e-06,
+      "loss": 0.3378,
+      "step": 9671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2834553726866216e-06,
+      "loss": 0.3876,
+      "step": 9672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2820740246838592e-06,
+      "loss": 0.3097,
+      "step": 9673
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2806933695175173e-06,
+      "loss": 0.3554,
+      "step": 9674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2793134072973312e-06,
+      "loss": 0.3136,
+      "step": 9675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2779341381329603e-06,
+      "loss": 0.3404,
+      "step": 9676
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2765555621340243e-06,
+      "loss": 0.3412,
+      "step": 9677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2751776794100835e-06,
+      "loss": 0.3473,
+      "step": 9678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2738004900706358e-06,
+      "loss": 0.2966,
+      "step": 9679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.272423994225137e-06,
+      "loss": 0.3509,
+      "step": 9680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2710481919829743e-06,
+      "loss": 0.3011,
+      "step": 9681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.269673083453492e-06,
+      "loss": 0.3225,
+      "step": 9682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.268298668745971e-06,
+      "loss": 0.3297,
+      "step": 9683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2669249479696355e-06,
+      "loss": 0.3469,
+      "step": 9684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2655519212336686e-06,
+      "loss": 0.3451,
+      "step": 9685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2641795886471797e-06,
+      "loss": 0.3233,
+      "step": 9686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2628079503192347e-06,
+      "loss": 0.3453,
+      "step": 9687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2614370063588422e-06,
+      "loss": 0.3398,
+      "step": 9688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2600667568749502e-06,
+      "loss": 0.2865,
+      "step": 9689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2586972019764643e-06,
+      "loss": 0.3232,
+      "step": 9690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2573283417722192e-06,
+      "loss": 0.3638,
+      "step": 9691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.255960176371005e-06,
+      "loss": 0.3582,
+      "step": 9692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2545927058815533e-06,
+      "loss": 0.3698,
+      "step": 9693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2532259304125383e-06,
+      "loss": 0.2994,
+      "step": 9694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2518598500725842e-06,
+      "loss": 0.3273,
+      "step": 9695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2504944649702555e-06,
+      "loss": 0.3313,
+      "step": 9696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.249129775214064e-06,
+      "loss": 0.3436,
+      "step": 9697
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2477657809124632e-06,
+      "loss": 0.3639,
+      "step": 9698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2464024821738551e-06,
+      "loss": 0.3589,
+      "step": 9699
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2450398791065843e-06,
+      "loss": 0.372,
+      "step": 9700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2436779718189395e-06,
+      "loss": 0.3093,
+      "step": 9701
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2423167604191566e-06,
+      "loss": 0.3963,
+      "step": 9702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2409562450154133e-06,
+      "loss": 0.3572,
+      "step": 9703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2395964257158333e-06,
+      "loss": 0.3493,
+      "step": 9704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2382373026284856e-06,
+      "loss": 0.3762,
+      "step": 9705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2368788758613837e-06,
+      "loss": 0.3603,
+      "step": 9706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2355211455224836e-06,
+      "loss": 0.3778,
+      "step": 9707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2341641117196901e-06,
+      "loss": 0.2914,
+      "step": 9708
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2328077745608503e-06,
+      "loss": 0.3495,
+      "step": 9709
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.231452134153751e-06,
+      "loss": 0.3202,
+      "step": 9710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2300971906061354e-06,
+      "loss": 0.3308,
+      "step": 9711
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2287429440256781e-06,
+      "loss": 0.369,
+      "step": 9712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2273893945200087e-06,
+      "loss": 0.3273,
+      "step": 9713
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.226036542196698e-06,
+      "loss": 0.3698,
+      "step": 9714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2246843871632564e-06,
+      "loss": 0.3427,
+      "step": 9715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.223332929527148e-06,
+      "loss": 0.3776,
+      "step": 9716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2219821693957712e-06,
+      "loss": 0.3306,
+      "step": 9717
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2206321068764803e-06,
+      "loss": 0.358,
+      "step": 9718
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2192827420765663e-06,
+      "loss": 0.3274,
+      "step": 9719
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2179340751032631e-06,
+      "loss": 0.3489,
+      "step": 9720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2165861060637597e-06,
+      "loss": 0.3429,
+      "step": 9721
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2152388350651766e-06,
+      "loss": 0.3281,
+      "step": 9722
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2138922622145854e-06,
+      "loss": 0.3472,
+      "step": 9723
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.212546387619008e-06,
+      "loss": 0.3051,
+      "step": 9724
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2112012113853955e-06,
+      "loss": 0.3631,
+      "step": 9725
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2098567336206625e-06,
+      "loss": 0.3045,
+      "step": 9726
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2085129544316498e-06,
+      "loss": 0.374,
+      "step": 9727
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2071698739251537e-06,
+      "loss": 0.3138,
+      "step": 9728
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.205827492207914e-06,
+      "loss": 0.332,
+      "step": 9729
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2044858093866097e-06,
+      "loss": 0.3631,
+      "step": 9730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2031448255678736e-06,
+      "loss": 0.355,
+      "step": 9731
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2018045408582723e-06,
+      "loss": 0.3458,
+      "step": 9732
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2004649553643233e-06,
+      "loss": 0.3484,
+      "step": 9733
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1991260691924877e-06,
+      "loss": 0.3307,
+      "step": 9734
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1977878824491685e-06,
+      "loss": 0.3729,
+      "step": 9735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1964503952407169e-06,
+      "loss": 0.3475,
+      "step": 9736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1951136076734248e-06,
+      "loss": 0.3006,
+      "step": 9737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1937775198535318e-06,
+      "loss": 0.3275,
+      "step": 9738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1924421318872182e-06,
+      "loss": 0.3376,
+      "step": 9739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1911074438806137e-06,
+      "loss": 0.3212,
+      "step": 9740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1897734559397866e-06,
+      "loss": 0.3256,
+      "step": 9741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1884401681707535e-06,
+      "loss": 0.3275,
+      "step": 9742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1871075806794752e-06,
+      "loss": 0.3296,
+      "step": 9743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1857756935718556e-06,
+      "loss": 0.3745,
+      "step": 9744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1844445069537414e-06,
+      "loss": 0.3267,
+      "step": 9745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1831140209309277e-06,
+      "loss": 0.343,
+      "step": 9746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1817842356091513e-06,
+      "loss": 0.3281,
+      "step": 9747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.180455151094092e-06,
+      "loss": 0.3504,
+      "step": 9748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1791267674913787e-06,
+      "loss": 0.3558,
+      "step": 9749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.17779908490658e-06,
+      "loss": 0.3317,
+      "step": 9750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1764721034452075e-06,
+      "loss": 0.3232,
+      "step": 9751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1751458232127256e-06,
+      "loss": 0.3886,
+      "step": 9752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1738202443145307e-06,
+      "loss": 0.3581,
+      "step": 9753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1724953668559746e-06,
+      "loss": 0.3287,
+      "step": 9754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1711711909423506e-06,
+      "loss": 0.3485,
+      "step": 9755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1698477166788879e-06,
+      "loss": 0.3388,
+      "step": 9756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1685249441707735e-06,
+      "loss": 0.366,
+      "step": 9757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1672028735231256e-06,
+      "loss": 0.328,
+      "step": 9758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1658815048410133e-06,
+      "loss": 0.3386,
+      "step": 9759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.164560838229456e-06,
+      "loss": 0.3253,
+      "step": 9760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1632408737934008e-06,
+      "loss": 0.3351,
+      "step": 9761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1619216116377586e-06,
+      "loss": 0.3457,
+      "step": 9762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1606030518673671e-06,
+      "loss": 0.3628,
+      "step": 9763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1592851945870154e-06,
+      "loss": 0.3266,
+      "step": 9764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1579680399014448e-06,
+      "loss": 0.3334,
+      "step": 9765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.156651587915324e-06,
+      "loss": 0.3313,
+      "step": 9766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1553358387332825e-06,
+      "loss": 0.3205,
+      "step": 9767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1540207924598812e-06,
+      "loss": 0.4002,
+      "step": 9768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.152706449199632e-06,
+      "loss": 0.3142,
+      "step": 9769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1513928090569892e-06,
+      "loss": 0.3277,
+      "step": 9770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1500798721363516e-06,
+      "loss": 0.2962,
+      "step": 9771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1487676385420598e-06,
+      "loss": 0.3444,
+      "step": 9772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1474561083784031e-06,
+      "loss": 0.3533,
+      "step": 9773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.146145281749611e-06,
+      "loss": 0.3353,
+      "step": 9774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1448351587598583e-06,
+      "loss": 0.3268,
+      "step": 9775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1435257395132637e-06,
+      "loss": 0.3787,
+      "step": 9776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1422170241138908e-06,
+      "loss": 0.4107,
+      "step": 9777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1409090126657463e-06,
+      "loss": 0.3203,
+      "step": 9778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1396017052727804e-06,
+      "loss": 0.3411,
+      "step": 9779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.13829510203889e-06,
+      "loss": 0.3322,
+      "step": 9780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1369892030679142e-06,
+      "loss": 0.3544,
+      "step": 9781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1356840084636345e-06,
+      "loss": 0.3518,
+      "step": 9782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.134379518329779e-06,
+      "loss": 0.3916,
+      "step": 9783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1330757327700193e-06,
+      "loss": 0.3609,
+      "step": 9784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1317726518879712e-06,
+      "loss": 0.3646,
+      "step": 9785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1304702757871944e-06,
+      "loss": 0.3842,
+      "step": 9786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1291686045711869e-06,
+      "loss": 0.3191,
+      "step": 9787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1278676383434017e-06,
+      "loss": 0.2936,
+      "step": 9788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1265673772072284e-06,
+      "loss": 0.3485,
+      "step": 9789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1252678212660017e-06,
+      "loss": 0.3456,
+      "step": 9790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.123968970623004e-06,
+      "loss": 0.3077,
+      "step": 9791
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.12267082538145e-06,
+      "loss": 0.3426,
+      "step": 9792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1213733856445153e-06,
+      "loss": 0.3296,
+      "step": 9793
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1200766515153073e-06,
+      "loss": 0.3043,
+      "step": 9794
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1187806230968801e-06,
+      "loss": 0.3682,
+      "step": 9795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.117485300492237e-06,
+      "loss": 0.3738,
+      "step": 9796
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1161906838043124e-06,
+      "loss": 0.3382,
+      "step": 9797
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1148967731360017e-06,
+      "loss": 0.3499,
+      "step": 9798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1136035685901292e-06,
+      "loss": 0.3493,
+      "step": 9799
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1123110702694707e-06,
+      "loss": 0.3058,
+      "step": 9800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1110192782767482e-06,
+      "loss": 0.3401,
+      "step": 9801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1097281927146164e-06,
+      "loss": 0.3768,
+      "step": 9802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10843781368569e-06,
+      "loss": 0.347,
+      "step": 9803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1071481412925122e-06,
+      "loss": 0.32,
+      "step": 9804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1058591756375768e-06,
+      "loss": 0.3362,
+      "step": 9805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1045709168233276e-06,
+      "loss": 0.3374,
+      "step": 9806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10328336495214e-06,
+      "loss": 0.337,
+      "step": 9807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1019965201263405e-06,
+      "loss": 0.3313,
+      "step": 9808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.100710382448198e-06,
+      "loss": 0.3665,
+      "step": 9809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0994249520199263e-06,
+      "loss": 0.3087,
+      "step": 9810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0981402289436804e-06,
+      "loss": 0.3402,
+      "step": 9811
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.096856213321561e-06,
+      "loss": 0.3356,
+      "step": 9812
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.095572905255613e-06,
+      "loss": 0.3438,
+      "step": 9813
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0942903048478247e-06,
+      "loss": 0.3141,
+      "step": 9814
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0930084122001261e-06,
+      "loss": 0.348,
+      "step": 9815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0917272274143931e-06,
+      "loss": 0.3612,
+      "step": 9816
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0904467505924442e-06,
+      "loss": 0.3438,
+      "step": 9817
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.089166981836044e-06,
+      "loss": 0.3245,
+      "step": 9818
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0878879212468984e-06,
+      "loss": 0.3268,
+      "step": 9819
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0866095689266565e-06,
+      "loss": 0.2991,
+      "step": 9820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0853319249769146e-06,
+      "loss": 0.3251,
+      "step": 9821
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0840549894992103e-06,
+      "loss": 0.3668,
+      "step": 9822
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0827787625950193e-06,
+      "loss": 0.3476,
+      "step": 9823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0815032443657737e-06,
+      "loss": 0.3708,
+      "step": 9824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.080228434912839e-06,
+      "loss": 0.3479,
+      "step": 9825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0789543343375296e-06,
+      "loss": 0.3174,
+      "step": 9826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0776809427411017e-06,
+      "loss": 0.338,
+      "step": 9827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0764082602247495e-06,
+      "loss": 0.3502,
+      "step": 9828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.075136286889623e-06,
+      "loss": 0.3328,
+      "step": 9829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0738650228368065e-06,
+      "loss": 0.3057,
+      "step": 9830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0725944681673317e-06,
+      "loss": 0.3758,
+      "step": 9831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0713246229821738e-06,
+      "loss": 0.3525,
+      "step": 9832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0700554873822456e-06,
+      "loss": 0.3455,
+      "step": 9833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0687870614684147e-06,
+      "loss": 0.3375,
+      "step": 9834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0675193453414857e-06,
+      "loss": 0.3598,
+      "step": 9835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0662523391022005e-06,
+      "loss": 0.3227,
+      "step": 9836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0649860428512604e-06,
+      "loss": 0.3562,
+      "step": 9837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0637204566892944e-06,
+      "loss": 0.3096,
+      "step": 9838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0624555807168878e-06,
+      "loss": 0.3691,
+      "step": 9839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0611914150345591e-06,
+      "loss": 0.3247,
+      "step": 9840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0599279597427737e-06,
+      "loss": 0.3243,
+      "step": 9841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0586652149419484e-06,
+      "loss": 0.352,
+      "step": 9842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.057403180732429e-06,
+      "loss": 0.34,
+      "step": 9843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.056141857214521e-06,
+      "loss": 0.3328,
+      "step": 9844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0548812444884583e-06,
+      "loss": 0.3307,
+      "step": 9845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0536213426544252e-06,
+      "loss": 0.316,
+      "step": 9846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.052362151812556e-06,
+      "loss": 0.3428,
+      "step": 9847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0511036720629142e-06,
+      "loss": 0.3359,
+      "step": 9848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0498459035055186e-06,
+      "loss": 0.3847,
+      "step": 9849
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.048588846240326e-06,
+      "loss": 0.4155,
+      "step": 9850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0473325003672386e-06,
+      "loss": 0.3705,
+      "step": 9851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.046076865986102e-06,
+      "loss": 0.3493,
+      "step": 9852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0448219431967033e-06,
+      "loss": 0.3496,
+      "step": 9853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0435677320987758e-06,
+      "loss": 0.349,
+      "step": 9854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0423142327919954e-06,
+      "loss": 0.3343,
+      "step": 9855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0410614453759794e-06,
+      "loss": 0.3322,
+      "step": 9856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0398093699502898e-06,
+      "loss": 0.3319,
+      "step": 9857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0385580066144341e-06,
+      "loss": 0.33,
+      "step": 9858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0373073554678602e-06,
+      "loss": 0.3434,
+      "step": 9859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0360574166099624e-06,
+      "loss": 0.3436,
+      "step": 9860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0348081901400741e-06,
+      "loss": 0.3059,
+      "step": 9861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0335596761574761e-06,
+      "loss": 0.3439,
+      "step": 9862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0323118747613902e-06,
+      "loss": 0.362,
+      "step": 9863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0310647860509837e-06,
+      "loss": 0.3297,
+      "step": 9864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.029818410125365e-06,
+      "loss": 0.3425,
+      "step": 9865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.028572747083587e-06,
+      "loss": 0.3268,
+      "step": 9866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0273277970246466e-06,
+      "loss": 0.3359,
+      "step": 9867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.026083560047484e-06,
+      "loss": 0.3569,
+      "step": 9868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.024840036250977e-06,
+      "loss": 0.3354,
+      "step": 9869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0235972257339566e-06,
+      "loss": 0.33,
+      "step": 9870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0223551285951938e-06,
+      "loss": 0.3213,
+      "step": 9871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.021113744933394e-06,
+      "loss": 0.3475,
+      "step": 9872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.019873074847222e-06,
+      "loss": 0.3265,
+      "step": 9873
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0186331184352681e-06,
+      "loss": 0.3394,
+      "step": 9874
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.017393875796081e-06,
+      "loss": 0.3214,
+      "step": 9875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0161553470281482e-06,
+      "loss": 0.3075,
+      "step": 9876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.014917532229891e-06,
+      "loss": 0.3454,
+      "step": 9877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0136804314996918e-06,
+      "loss": 0.3328,
+      "step": 9878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0124440449358553e-06,
+      "loss": 0.3486,
+      "step": 9879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0112083726366505e-06,
+      "loss": 0.3417,
+      "step": 9880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0099734147002737e-06,
+      "loss": 0.3264,
+      "step": 9881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0087391712248685e-06,
+      "loss": 0.3733,
+      "step": 9882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0075056423085305e-06,
+      "loss": 0.3667,
+      "step": 9883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0062728280492862e-06,
+      "loss": 0.2953,
+      "step": 9884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0050407285451126e-06,
+      "loss": 0.3274,
+      "step": 9885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0038093438939257e-06,
+      "loss": 0.3355,
+      "step": 9886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0025786741935871e-06,
+      "loss": 0.3809,
+      "step": 9887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.001348719541906e-06,
+      "loss": 0.329,
+      "step": 9888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0001194800366254e-06,
+      "loss": 0.3212,
+      "step": 9889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.988909557754366e-07,
+      "loss": 0.2966,
+      "step": 9890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.97663146855975e-07,
+      "loss": 0.3199,
+      "step": 9891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.964360533758176e-07,
+      "loss": 0.3304,
+      "step": 9892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.952096754324847e-07,
+      "loss": 0.3767,
+      "step": 9893
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.93984013123439e-07,
+      "loss": 0.2967,
+      "step": 9894
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.927590665460874e-07,
+      "loss": 0.3185,
+      "step": 9895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.915348357977795e-07,
+      "loss": 0.2995,
+      "step": 9896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.3152,
+      "step": 9897
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.890885221774115e-07,
+      "loss": 0.3329,
+      "step": 9898
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.878664394997662e-07,
+      "loss": 0.3422,
+      "step": 9899
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.866450730399934e-07,
+      "loss": 0.3303,
+      "step": 9900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.854244228951603e-07,
+      "loss": 0.3516,
+      "step": 9901
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.84204489162275e-07,
+      "loss": 0.3176,
+      "step": 9902
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.829852719382882e-07,
+      "loss": 0.3748,
+      "step": 9903
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.817667713200951e-07,
+      "loss": 0.3284,
+      "step": 9904
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.805489874045316e-07,
+      "loss": 0.3292,
+      "step": 9905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.7933192028838e-07,
+      "loss": 0.3477,
+      "step": 9906
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.78115570068362e-07,
+      "loss": 0.3115,
+      "step": 9907
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.768999368411457e-07,
+      "loss": 0.3589,
+      "step": 9908
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.75685020703342e-07,
+      "loss": 0.3386,
+      "step": 9909
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.744708217514964e-07,
+      "loss": 0.3218,
+      "step": 9910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.732573400821132e-07,
+      "loss": 0.3556,
+      "step": 9911
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.720445757916285e-07,
+      "loss": 0.353,
+      "step": 9912
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.708325289764186e-07,
+      "loss": 0.3389,
+      "step": 9913
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.696211997328164e-07,
+      "loss": 0.3253,
+      "step": 9914
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.684105881570816e-07,
+      "loss": 0.3328,
+      "step": 9915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.672006943454294e-07,
+      "loss": 0.3565,
+      "step": 9916
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.659915183940138e-07,
+      "loss": 0.3424,
+      "step": 9917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.64783060398926e-07,
+      "loss": 0.3213,
+      "step": 9918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.635753204562126e-07,
+      "loss": 0.3476,
+      "step": 9919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.62368298661851e-07,
+      "loss": 0.3687,
+      "step": 9920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.611619951117657e-07,
+      "loss": 0.3808,
+      "step": 9921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.599564099018278e-07,
+      "loss": 0.3673,
+      "step": 9922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.58751543127845e-07,
+      "loss": 0.3944,
+      "step": 9923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.575473948855762e-07,
+      "loss": 0.3655,
+      "step": 9924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.56343965270714e-07,
+      "loss": 0.3644,
+      "step": 9925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.551412543788996e-07,
+      "loss": 0.3274,
+      "step": 9926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.539392623057143e-07,
+      "loss": 0.3458,
+      "step": 9927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.527379891466837e-07,
+      "loss": 0.3495,
+      "step": 9928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.515374349972806e-07,
+      "loss": 0.3419,
+      "step": 9929
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.503375999529107e-07,
+      "loss": 0.3452,
+      "step": 9930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.491384841089301e-07,
+      "loss": 0.3564,
+      "step": 9931
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.479400875606348e-07,
+      "loss": 0.3466,
+      "step": 9932
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.467424104032663e-07,
+      "loss": 0.32,
+      "step": 9933
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.455454527320051e-07,
+      "loss": 0.3615,
+      "step": 9934
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.443492146419786e-07,
+      "loss": 0.3128,
+      "step": 9935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.431536962282539e-07,
+      "loss": 0.3304,
+      "step": 9936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.419588975858407e-07,
+      "loss": 0.3756,
+      "step": 9937
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.407648188096951e-07,
+      "loss": 0.367,
+      "step": 9938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.395714599947126e-07,
+      "loss": 0.3317,
+      "step": 9939
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.383788212357326e-07,
+      "loss": 0.3295,
+      "step": 9940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.371869026275371e-07,
+      "loss": 0.3533,
+      "step": 9941
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.359957042648515e-07,
+      "loss": 0.3296,
+      "step": 9942
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.348052262423424e-07,
+      "loss": 0.3021,
+      "step": 9943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.336154686546217e-07,
+      "loss": 0.3537,
+      "step": 9944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.324264315962416e-07,
+      "loss": 0.3614,
+      "step": 9945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.312381151616978e-07,
+      "loss": 0.345,
+      "step": 9946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.30050519445429e-07,
+      "loss": 0.3534,
+      "step": 9947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.288636445418197e-07,
+      "loss": 0.3645,
+      "step": 9948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.276774905451868e-07,
+      "loss": 0.3537,
+      "step": 9949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.264920575498049e-07,
+      "loss": 0.3387,
+      "step": 9950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.253073456498762e-07,
+      "loss": 0.3017,
+      "step": 9951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.241233549395578e-07,
+      "loss": 0.3124,
+      "step": 9952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.229400855129467e-07,
+      "loss": 0.3563,
+      "step": 9953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.217575374640719e-07,
+      "loss": 0.3813,
+      "step": 9954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.205757108869229e-07,
+      "loss": 0.325,
+      "step": 9955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.193946058754155e-07,
+      "loss": 0.3666,
+      "step": 9956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.182142225234192e-07,
+      "loss": 0.3626,
+      "step": 9957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.170345609247433e-07,
+      "loss": 0.3562,
+      "step": 9958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.158556211731318e-07,
+      "loss": 0.3227,
+      "step": 9959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.146774033622874e-07,
+      "loss": 0.368,
+      "step": 9960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.134999075858386e-07,
+      "loss": 0.353,
+      "step": 9961
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.123231339373661e-07,
+      "loss": 0.3597,
+      "step": 9962
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.111470825103952e-07,
+      "loss": 0.328,
+      "step": 9963
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.09971753398382e-07,
+      "loss": 0.3612,
+      "step": 9964
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.08797146694742e-07,
+      "loss": 0.3224,
+      "step": 9965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.076232624928183e-07,
+      "loss": 0.3357,
+      "step": 9966
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.064501008859039e-07,
+      "loss": 0.3139,
+      "step": 9967
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.052776619672343e-07,
+      "loss": 0.3484,
+      "step": 9968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.041059458299817e-07,
+      "loss": 0.3464,
+      "step": 9969
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.029349525672737e-07,
+      "loss": 0.3384,
+      "step": 9970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.017646822721648e-07,
+      "loss": 0.304,
+      "step": 9971
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.005951350376629e-07,
+      "loss": 0.3929,
+      "step": 9972
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.994263109567136e-07,
+      "loss": 0.3334,
+      "step": 9973
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.982582101222082e-07,
+      "loss": 0.3455,
+      "step": 9974
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.970908326269767e-07,
+      "loss": 0.3252,
+      "step": 9975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.95924178563794e-07,
+      "loss": 0.3599,
+      "step": 9976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.94758248025378e-07,
+      "loss": 0.348,
+      "step": 9977
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.935930411043892e-07,
+      "loss": 0.3192,
+      "step": 9978
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.924285578934266e-07,
+      "loss": 0.3461,
+      "step": 9979
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.912647984850376e-07,
+      "loss": 0.35,
+      "step": 9980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.901017629717079e-07,
+      "loss": 0.3104,
+      "step": 9981
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.889394514458672e-07,
+      "loss": 0.3386,
+      "step": 9982
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.877778639998869e-07,
+      "loss": 0.3423,
+      "step": 9983
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.866170007260821e-07,
+      "loss": 0.354,
+      "step": 9984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.854568617167103e-07,
+      "loss": 0.3576,
+      "step": 9985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.842974470639687e-07,
+      "loss": 0.3285,
+      "step": 9986
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.831387568600014e-07,
+      "loss": 0.3394,
+      "step": 9987
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.819807911968902e-07,
+      "loss": 0.3272,
+      "step": 9988
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.808235501666662e-07,
+      "loss": 0.337,
+      "step": 9989
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.796670338612911e-07,
+      "loss": 0.3603,
+      "step": 9990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.785112423726827e-07,
+      "loss": 0.3372,
+      "step": 9991
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.773561757926897e-07,
+      "loss": 0.3449,
+      "step": 9992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.762018342131129e-07,
+      "loss": 0.3452,
+      "step": 9993
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.750482177256903e-07,
+      "loss": 0.3245,
+      "step": 9994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.738953264220984e-07,
+      "loss": 0.3479,
+      "step": 9995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.727431603939662e-07,
+      "loss": 0.3328,
+      "step": 9996
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.71591719732855e-07,
+      "loss": 0.3557,
+      "step": 9997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.704410045302736e-07,
+      "loss": 0.3229,
+      "step": 9998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.692910148776756e-07,
+      "loss": 0.3269,
+      "step": 9999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.68141750866448e-07,
+      "loss": 0.3384,
+      "step": 10000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.669932125879333e-07,
+      "loss": 0.326,
+      "step": 10001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.658454001334015e-07,
+      "loss": 0.3468,
+      "step": 10002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.646983135940745e-07,
+      "loss": 0.3153,
+      "step": 10003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.635519530611192e-07,
+      "loss": 0.394,
+      "step": 10004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.624063186256327e-07,
+      "loss": 0.3387,
+      "step": 10005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.612614103786665e-07,
+      "loss": 0.3079,
+      "step": 10006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.601172284112069e-07,
+      "loss": 0.3367,
+      "step": 10007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.589737728141867e-07,
+      "loss": 0.2795,
+      "step": 10008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.578310436784775e-07,
+      "loss": 0.3256,
+      "step": 10009
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.566890410948958e-07,
+      "loss": 0.3278,
+      "step": 10010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.555477651541999e-07,
+      "loss": 0.3325,
+      "step": 10011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.544072159470895e-07,
+      "loss": 0.3724,
+      "step": 10012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.532673935642077e-07,
+      "loss": 0.3686,
+      "step": 10013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.521282980961387e-07,
+      "loss": 0.3733,
+      "step": 10014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.5098992963341e-07,
+      "loss": 0.3166,
+      "step": 10015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.498522882664894e-07,
+      "loss": 0.3111,
+      "step": 10016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.48715374085789e-07,
+      "loss": 0.3593,
+      "step": 10017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.47579187181663e-07,
+      "loss": 0.3432,
+      "step": 10018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.46443727644406e-07,
+      "loss": 0.3717,
+      "step": 10019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.453089955642568e-07,
+      "loss": 0.378,
+      "step": 10020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.441749910313956e-07,
+      "loss": 0.3239,
+      "step": 10021
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.430417141359437e-07,
+      "loss": 0.3502,
+      "step": 10022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.419091649679667e-07,
+      "loss": 0.3229,
+      "step": 10023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.407773436174704e-07,
+      "loss": 0.3252,
+      "step": 10024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.396462501744052e-07,
+      "loss": 0.3596,
+      "step": 10025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.385158847286579e-07,
+      "loss": 0.3121,
+      "step": 10026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.373862473700667e-07,
+      "loss": 0.3715,
+      "step": 10027
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.362573381884042e-07,
+      "loss": 0.3384,
+      "step": 10028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.351291572733888e-07,
+      "loss": 0.3231,
+      "step": 10029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.340017047146809e-07,
+      "loss": 0.3337,
+      "step": 10030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.328749806018765e-07,
+      "loss": 0.3228,
+      "step": 10031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.317489850245252e-07,
+      "loss": 0.3665,
+      "step": 10032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.306237180721121e-07,
+      "loss": 0.3474,
+      "step": 10033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.294991798340634e-07,
+      "loss": 0.3045,
+      "step": 10034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.283753703997521e-07,
+      "loss": 0.3694,
+      "step": 10035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.272522898584845e-07,
+      "loss": 0.3348,
+      "step": 10036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.261299382995214e-07,
+      "loss": 0.3633,
+      "step": 10037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.250083158120537e-07,
+      "loss": 0.2933,
+      "step": 10038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.238874224852211e-07,
+      "loss": 0.3707,
+      "step": 10039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.22767258408107e-07,
+      "loss": 0.3523,
+      "step": 10040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.21647823669729e-07,
+      "loss": 0.332,
+      "step": 10041
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.205291183590569e-07,
+      "loss": 0.3113,
+      "step": 10042
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.194111425649931e-07,
+      "loss": 0.3481,
+      "step": 10043
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.182938963763842e-07,
+      "loss": 0.3439,
+      "step": 10044
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.171773798820282e-07,
+      "loss": 0.3342,
+      "step": 10045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.160615931706506e-07,
+      "loss": 0.2961,
+      "step": 10046
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.149465363309295e-07,
+      "loss": 0.3226,
+      "step": 10047
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.138322094514795e-07,
+      "loss": 0.3356,
+      "step": 10048
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.127186126208597e-07,
+      "loss": 0.3343,
+      "step": 10049
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.116057459275706e-07,
+      "loss": 0.3263,
+      "step": 10050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.104936094600546e-07,
+      "loss": 0.3164,
+      "step": 10051
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.093822033066967e-07,
+      "loss": 0.3387,
+      "step": 10052
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.082715275558218e-07,
+      "loss": 0.3377,
+      "step": 10053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.071615822957002e-07,
+      "loss": 0.3132,
+      "step": 10054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.060523676145404e-07,
+      "loss": 0.3181,
+      "step": 10055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.049438836004963e-07,
+      "loss": 0.374,
+      "step": 10056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.038361303416598e-07,
+      "loss": 0.3348,
+      "step": 10057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.02729107926069e-07,
+      "loss": 0.3399,
+      "step": 10058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.016228164417017e-07,
+      "loss": 0.3622,
+      "step": 10059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.005172559764762e-07,
+      "loss": 0.3442,
+      "step": 10060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.994124266182568e-07,
+      "loss": 0.3106,
+      "step": 10061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.983083284548432e-07,
+      "loss": 0.3552,
+      "step": 10062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.972049615739841e-07,
+      "loss": 0.3292,
+      "step": 10063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.961023260633671e-07,
+      "loss": 0.3628,
+      "step": 10064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.9500042201062e-07,
+      "loss": 0.356,
+      "step": 10065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.938992495033171e-07,
+      "loss": 0.3697,
+      "step": 10066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.92798808628965e-07,
+      "loss": 0.3198,
+      "step": 10067
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.916990994750251e-07,
+      "loss": 0.3346,
+      "step": 10068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.906001221288906e-07,
+      "loss": 0.3321,
+      "step": 10069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.895018766779028e-07,
+      "loss": 0.347,
+      "step": 10070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.884043632093419e-07,
+      "loss": 0.3113,
+      "step": 10071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.873075818104248e-07,
+      "loss": 0.3517,
+      "step": 10072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.862115325683217e-07,
+      "loss": 0.3342,
+      "step": 10073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.851162155701386e-07,
+      "loss": 0.3535,
+      "step": 10074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.840216309029169e-07,
+      "loss": 0.3694,
+      "step": 10075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.829277786536538e-07,
+      "loss": 0.3145,
+      "step": 10076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.81834658909274e-07,
+      "loss": 0.3364,
+      "step": 10077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.807422717566582e-07,
+      "loss": 0.343,
+      "step": 10078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.796506172826135e-07,
+      "loss": 0.3465,
+      "step": 10079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.785596955738994e-07,
+      "loss": 0.3542,
+      "step": 10080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.774695067172167e-07,
+      "loss": 0.3702,
+      "step": 10081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.763800507992015e-07,
+      "loss": 0.3739,
+      "step": 10082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.7529132790644e-07,
+      "loss": 0.3364,
+      "step": 10083
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.742033381254533e-07,
+      "loss": 0.3099,
+      "step": 10084
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.731160815427041e-07,
+      "loss": 0.3869,
+      "step": 10085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.720295582446069e-07,
+      "loss": 0.3478,
+      "step": 10086
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.709437683175047e-07,
+      "loss": 0.3123,
+      "step": 10087
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.698587118476897e-07,
+      "loss": 0.3373,
+      "step": 10088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.687743889213939e-07,
+      "loss": 0.3273,
+      "step": 10089
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.676907996247906e-07,
+      "loss": 0.3268,
+      "step": 10090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.666079440439978e-07,
+      "loss": 0.3328,
+      "step": 10091
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.655258222650708e-07,
+      "loss": 0.3676,
+      "step": 10092
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.644444343740087e-07,
+      "loss": 0.3161,
+      "step": 10093
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.633637804567539e-07,
+      "loss": 0.3668,
+      "step": 10094
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.622838605991867e-07,
+      "loss": 0.3664,
+      "step": 10095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.335,
+      "step": 10096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.601262234063578e-07,
+      "loss": 0.3401,
+      "step": 10097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.590485062425678e-07,
+      "loss": 0.3678,
+      "step": 10098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.579715234814123e-07,
+      "loss": 0.3693,
+      "step": 10099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.568952752084835e-07,
+      "loss": 0.358,
+      "step": 10100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.55819761509311e-07,
+      "loss": 0.338,
+      "step": 10101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.547449824693709e-07,
+      "loss": 0.3438,
+      "step": 10102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.536709381740781e-07,
+      "loss": 0.3247,
+      "step": 10103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.525976287087888e-07,
+      "loss": 0.3451,
+      "step": 10104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.515250541588026e-07,
+      "loss": 0.3091,
+      "step": 10105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.504532146093601e-07,
+      "loss": 0.3378,
+      "step": 10106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.493821101456444e-07,
+      "loss": 0.3699,
+      "step": 10107
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.483117408527729e-07,
+      "loss": 0.3359,
+      "step": 10108
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.472421068158175e-07,
+      "loss": 0.3642,
+      "step": 10109
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.461732081197848e-07,
+      "loss": 0.35,
+      "step": 10110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.451050448496167e-07,
+      "loss": 0.3542,
+      "step": 10111
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.440376170902097e-07,
+      "loss": 0.3408,
+      "step": 10112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.429709249263895e-07,
+      "loss": 0.3285,
+      "step": 10113
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.419049684429324e-07,
+      "loss": 0.3458,
+      "step": 10114
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.408397477245544e-07,
+      "loss": 0.3338,
+      "step": 10115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.397752628559051e-07,
+      "loss": 0.3465,
+      "step": 10116
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.387115139215895e-07,
+      "loss": 0.3651,
+      "step": 10117
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.376485010061385e-07,
+      "loss": 0.332,
+      "step": 10118
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.365862241940413e-07,
+      "loss": 0.2976,
+      "step": 10119
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.355246835697139e-07,
+      "loss": 0.3204,
+      "step": 10120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.344638792175184e-07,
+      "loss": 0.3256,
+      "step": 10121
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.334038112217656e-07,
+      "loss": 0.3618,
+      "step": 10122
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.323444796666979e-07,
+      "loss": 0.3074,
+      "step": 10123
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.312858846365034e-07,
+      "loss": 0.3604,
+      "step": 10124
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.302280262153127e-07,
+      "loss": 0.3358,
+      "step": 10125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.291709044871931e-07,
+      "loss": 0.3461,
+      "step": 10126
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.281145195361638e-07,
+      "loss": 0.4078,
+      "step": 10127
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.270588714461724e-07,
+      "loss": 0.3922,
+      "step": 10128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.26003960301116e-07,
+      "loss": 0.3307,
+      "step": 10129
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.24949786184831e-07,
+      "loss": 0.3322,
+      "step": 10130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.238963491810935e-07,
+      "loss": 0.3151,
+      "step": 10131
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.22843649373629e-07,
+      "loss": 0.3343,
+      "step": 10132
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.217916868460917e-07,
+      "loss": 0.396,
+      "step": 10133
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.207404616820868e-07,
+      "loss": 0.3175,
+      "step": 10134
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.196899739651564e-07,
+      "loss": 0.3811,
+      "step": 10135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.186402237787871e-07,
+      "loss": 0.3158,
+      "step": 10136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.175912112064054e-07,
+      "loss": 0.3066,
+      "step": 10137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.165429363313769e-07,
+      "loss": 0.323,
+      "step": 10138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.154953992370139e-07,
+      "loss": 0.3179,
+      "step": 10139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.144486000065643e-07,
+      "loss": 0.3432,
+      "step": 10140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.134025387232213e-07,
+      "loss": 0.3399,
+      "step": 10141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.123572154701175e-07,
+      "loss": 0.3498,
+      "step": 10142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.113126303303275e-07,
+      "loss": 0.3422,
+      "step": 10143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.102687833868671e-07,
+      "loss": 0.3399,
+      "step": 10144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.092256747226944e-07,
+      "loss": 0.3471,
+      "step": 10145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.081833044207076e-07,
+      "loss": 0.3475,
+      "step": 10146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.07141672563747e-07,
+      "loss": 0.3279,
+      "step": 10147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.061007792345942e-07,
+      "loss": 0.3319,
+      "step": 10148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.050606245159675e-07,
+      "loss": 0.3227,
+      "step": 10149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.040212084905363e-07,
+      "loss": 0.2984,
+      "step": 10150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.029825312409045e-07,
+      "loss": 0.3354,
+      "step": 10151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.019445928496149e-07,
+      "loss": 0.3566,
+      "step": 10152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.009073933991595e-07,
+      "loss": 0.3335,
+      "step": 10153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.998709329719633e-07,
+      "loss": 0.3592,
+      "step": 10154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.988352116504005e-07,
+      "loss": 0.3526,
+      "step": 10155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.97800229516783e-07,
+      "loss": 0.3467,
+      "step": 10156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.967659866533572e-07,
+      "loss": 0.3415,
+      "step": 10157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.95732483142324e-07,
+      "loss": 0.346,
+      "step": 10158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.946997190658156e-07,
+      "loss": 0.3529,
+      "step": 10159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.936676945059073e-07,
+      "loss": 0.346,
+      "step": 10160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.926364095446192e-07,
+      "loss": 0.356,
+      "step": 10161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.916058642639079e-07,
+      "loss": 0.3503,
+      "step": 10162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.905760587456778e-07,
+      "loss": 0.3461,
+      "step": 10163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.895469930717657e-07,
+      "loss": 0.3429,
+      "step": 10164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.885186673239553e-07,
+      "loss": 0.316,
+      "step": 10165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.87491081583972e-07,
+      "loss": 0.364,
+      "step": 10166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.864642359334772e-07,
+      "loss": 0.3362,
+      "step": 10167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.854381304540825e-07,
+      "loss": 0.3447,
+      "step": 10168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.844127652273302e-07,
+      "loss": 0.3417,
+      "step": 10169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.83388140334711e-07,
+      "loss": 0.3461,
+      "step": 10170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.823642558576537e-07,
+      "loss": 0.3362,
+      "step": 10171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.813411118775304e-07,
+      "loss": 0.3171,
+      "step": 10172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.803187084756524e-07,
+      "loss": 0.326,
+      "step": 10173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.792970457332715e-07,
+      "loss": 0.3783,
+      "step": 10174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.782761237315838e-07,
+      "loss": 0.3583,
+      "step": 10175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.772559425517233e-07,
+      "loss": 0.3335,
+      "step": 10176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.762365022747675e-07,
+      "loss": 0.3096,
+      "step": 10177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.752178029817325e-07,
+      "loss": 0.3068,
+      "step": 10178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.741998447535792e-07,
+      "loss": 0.3544,
+      "step": 10179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.73182627671205e-07,
+      "loss": 0.3731,
+      "step": 10180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.721661518154532e-07,
+      "loss": 0.3195,
+      "step": 10181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.711504172671035e-07,
+      "loss": 0.3584,
+      "step": 10182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.701354241068814e-07,
+      "loss": 0.333,
+      "step": 10183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.691211724154489e-07,
+      "loss": 0.3273,
+      "step": 10184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.681076622734117e-07,
+      "loss": 0.2929,
+      "step": 10185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.670948937613175e-07,
+      "loss": 0.3148,
+      "step": 10186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.66082866959653e-07,
+      "loss": 0.3224,
+      "step": 10187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.65071581948844e-07,
+      "loss": 0.3416,
+      "step": 10188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.64061038809265e-07,
+      "loss": 0.4031,
+      "step": 10189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.630512376212217e-07,
+      "loss": 0.343,
+      "step": 10190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.620421784649678e-07,
+      "loss": 0.3491,
+      "step": 10191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.610338614206979e-07,
+      "loss": 0.3797,
+      "step": 10192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.600262865685414e-07,
+      "loss": 0.3435,
+      "step": 10193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.590194539885764e-07,
+      "loss": 0.3413,
+      "step": 10194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.580133637608154e-07,
+      "loss": 0.3444,
+      "step": 10195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.57008015965217e-07,
+      "loss": 0.3222,
+      "step": 10196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.560034106816815e-07,
+      "loss": 0.3802,
+      "step": 10197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.549995479900406e-07,
+      "loss": 0.3519,
+      "step": 10198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.539964279700817e-07,
+      "loss": 0.3472,
+      "step": 10199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.529940507015197e-07,
+      "loss": 0.3069,
+      "step": 10200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.519924162640168e-07,
+      "loss": 0.3382,
+      "step": 10201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.50991524737179e-07,
+      "loss": 0.3464,
+      "step": 10202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.49991376200545e-07,
+      "loss": 0.3459,
+      "step": 10203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.489919707336046e-07,
+      "loss": 0.3237,
+      "step": 10204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.479933084157786e-07,
+      "loss": 0.3129,
+      "step": 10205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.469953893264358e-07,
+      "loss": 0.3197,
+      "step": 10206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.459982135448817e-07,
+      "loss": 0.3058,
+      "step": 10207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.450017811503662e-07,
+      "loss": 0.3475,
+      "step": 10208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.44006092222077e-07,
+      "loss": 0.3543,
+      "step": 10209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.430111468391454e-07,
+      "loss": 0.3611,
+      "step": 10210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.420169450806413e-07,
+      "loss": 0.3451,
+      "step": 10211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.410234870255772e-07,
+      "loss": 0.3366,
+      "step": 10212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.400307727529042e-07,
+      "loss": 0.361,
+      "step": 10213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.390388023415184e-07,
+      "loss": 0.3335,
+      "step": 10214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.380475758702532e-07,
+      "loss": 0.3509,
+      "step": 10215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.370570934178833e-07,
+      "loss": 0.3424,
+      "step": 10216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.360673550631258e-07,
+      "loss": 0.3439,
+      "step": 10217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.350783608846378e-07,
+      "loss": 0.3269,
+      "step": 10218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.340901109610165e-07,
+      "loss": 0.3294,
+      "step": 10219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.331026053708011e-07,
+      "loss": 0.3531,
+      "step": 10220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.321158441924713e-07,
+      "loss": 0.3417,
+      "step": 10221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.311298275044475e-07,
+      "loss": 0.325,
+      "step": 10222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.301445553850927e-07,
+      "loss": 0.3504,
+      "step": 10223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.291600279127052e-07,
+      "loss": 0.3426,
+      "step": 10224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.281762451655315e-07,
+      "loss": 0.3322,
+      "step": 10225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.271932072217535e-07,
+      "loss": 0.326,
+      "step": 10226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.262109141594974e-07,
+      "loss": 0.3508,
+      "step": 10227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.252293660568287e-07,
+      "loss": 0.3479,
+      "step": 10228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.242485629917494e-07,
+      "loss": 0.3168,
+      "step": 10229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.232685050422127e-07,
+      "loss": 0.3104,
+      "step": 10230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.222891922861008e-07,
+      "loss": 0.3492,
+      "step": 10231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.213106248012457e-07,
+      "loss": 0.3444,
+      "step": 10232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.203328026654176e-07,
+      "loss": 0.3731,
+      "step": 10233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.193557259563221e-07,
+      "loss": 0.3515,
+      "step": 10234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.183793947516159e-07,
+      "loss": 0.3419,
+      "step": 10235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.174038091288848e-07,
+      "loss": 0.3202,
+      "step": 10236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.164289691656633e-07,
+      "loss": 0.3515,
+      "step": 10237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.154548749394274e-07,
+      "loss": 0.3579,
+      "step": 10238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.144815265275861e-07,
+      "loss": 0.3248,
+      "step": 10239
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.135089240074987e-07,
+      "loss": 0.3232,
+      "step": 10240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.12537067456458e-07,
+      "loss": 0.3908,
+      "step": 10241
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.115659569516974e-07,
+      "loss": 0.3357,
+      "step": 10242
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.105955925704e-07,
+      "loss": 0.3064,
+      "step": 10243
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.09625974389676e-07,
+      "loss": 0.349,
+      "step": 10244
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.086571024865895e-07,
+      "loss": 0.3728,
+      "step": 10245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.076889769381355e-07,
+      "loss": 0.3338,
+      "step": 10246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.06721597821256e-07,
+      "loss": 0.3214,
+      "step": 10247
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.057549652128292e-07,
+      "loss": 0.3295,
+      "step": 10248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.04789079189676e-07,
+      "loss": 0.3495,
+      "step": 10249
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.038239398285594e-07,
+      "loss": 0.3604,
+      "step": 10250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.028595472061804e-07,
+      "loss": 0.3809,
+      "step": 10251
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.018959013991821e-07,
+      "loss": 0.3226,
+      "step": 10252
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.009330024841475e-07,
+      "loss": 0.3117,
+      "step": 10253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.999708505376023e-07,
+      "loss": 0.3444,
+      "step": 10254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.990094456360096e-07,
+      "loss": 0.3667,
+      "step": 10255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.98048787855775e-07,
+      "loss": 0.3531,
+      "step": 10256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.970888772732453e-07,
+      "loss": 0.3617,
+      "step": 10257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.961297139647059e-07,
+      "loss": 0.3631,
+      "step": 10258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.95171298006385e-07,
+      "loss": 0.3416,
+      "step": 10259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.942136294744494e-07,
+      "loss": 0.3265,
+      "step": 10260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.932567084450092e-07,
+      "loss": 0.3312,
+      "step": 10261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.923005349941124e-07,
+      "loss": 0.3756,
+      "step": 10262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.913451091977484e-07,
+      "loss": 0.3648,
+      "step": 10263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.903904311318498e-07,
+      "loss": 0.3568,
+      "step": 10264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.894365008722813e-07,
+      "loss": 0.3348,
+      "step": 10265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.884833184948602e-07,
+      "loss": 0.3334,
+      "step": 10266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.875308840753369e-07,
+      "loss": 0.3139,
+      "step": 10267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.86579197689403e-07,
+      "loss": 0.3535,
+      "step": 10268
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.856282594126927e-07,
+      "loss": 0.3683,
+      "step": 10269
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.846780693207765e-07,
+      "loss": 0.32,
+      "step": 10270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.837286274891718e-07,
+      "loss": 0.3283,
+      "step": 10271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.827799339933327e-07,
+      "loss": 0.3196,
+      "step": 10272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.818319889086532e-07,
+      "loss": 0.3318,
+      "step": 10273
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.80884792310471e-07,
+      "loss": 0.3577,
+      "step": 10274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.799383442740581e-07,
+      "loss": 0.3345,
+      "step": 10275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.789926448746374e-07,
+      "loss": 0.328,
+      "step": 10276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.780476941873603e-07,
+      "loss": 0.341,
+      "step": 10277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.771034922873264e-07,
+      "loss": 0.3476,
+      "step": 10278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.76160039249577e-07,
+      "loss": 0.3101,
+      "step": 10279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.752173351490842e-07,
+      "loss": 0.335,
+      "step": 10280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.742753800607759e-07,
+      "loss": 0.357,
+      "step": 10281
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.733341740595044e-07,
+      "loss": 0.3329,
+      "step": 10282
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.723937172200711e-07,
+      "loss": 0.3267,
+      "step": 10283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.714540096172205e-07,
+      "loss": 0.3824,
+      "step": 10284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.705150513256297e-07,
+      "loss": 0.3113,
+      "step": 10285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.695768424199222e-07,
+      "loss": 0.3293,
+      "step": 10286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.686393829746583e-07,
+      "loss": 0.3683,
+      "step": 10287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.67702673064342e-07,
+      "loss": 0.3101,
+      "step": 10288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.667667127634147e-07,
+      "loss": 0.3392,
+      "step": 10289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.658315021462613e-07,
+      "loss": 0.3586,
+      "step": 10290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.648970412872035e-07,
+      "loss": 0.3213,
+      "step": 10291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.639633302605063e-07,
+      "loss": 0.311,
+      "step": 10292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.630303691403738e-07,
+      "loss": 0.3293,
+      "step": 10293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.620981580009521e-07,
+      "loss": 0.3304,
+      "step": 10294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.611666969163243e-07,
+      "loss": 0.3465,
+      "step": 10295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.602359859605189e-07,
+      "loss": 0.3723,
+      "step": 10296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.593060252074989e-07,
+      "loss": 0.3569,
+      "step": 10297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.58376814731173e-07,
+      "loss": 0.3657,
+      "step": 10298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.574483546053866e-07,
+      "loss": 0.3323,
+      "step": 10299
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.565206449039284e-07,
+      "loss": 0.3187,
+      "step": 10300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.55593685700524e-07,
+      "loss": 0.3528,
+      "step": 10301
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.54667477068842e-07,
+      "loss": 0.3583,
+      "step": 10302
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.537420190824916e-07,
+      "loss": 0.3695,
+      "step": 10303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.528173118150204e-07,
+      "loss": 0.322,
+      "step": 10304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.518933553399187e-07,
+      "loss": 0.3588,
+      "step": 10305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.509701497306119e-07,
+      "loss": 0.363,
+      "step": 10306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.500476950604738e-07,
+      "loss": 0.3554,
+      "step": 10307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.491259914028125e-07,
+      "loss": 0.309,
+      "step": 10308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.48205038830878e-07,
+      "loss": 0.3374,
+      "step": 10309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.472848374178629e-07,
+      "loss": 0.3592,
+      "step": 10310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.463653872368946e-07,
+      "loss": 0.3689,
+      "step": 10311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.454466883610465e-07,
+      "loss": 0.3079,
+      "step": 10312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.445287408633304e-07,
+      "loss": 0.3666,
+      "step": 10313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.436115448166946e-07,
+      "loss": 0.3601,
+      "step": 10314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.426951002940362e-07,
+      "loss": 0.3566,
+      "step": 10315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.417794073681826e-07,
+      "loss": 0.3674,
+      "step": 10316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.408644661119111e-07,
+      "loss": 0.3233,
+      "step": 10317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.399502765979304e-07,
+      "loss": 0.3359,
+      "step": 10318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.390368388988943e-07,
+      "loss": 0.359,
+      "step": 10319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.381241530873982e-07,
+      "loss": 0.3376,
+      "step": 10320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.372122192359719e-07,
+      "loss": 0.3414,
+      "step": 10321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.363010374170951e-07,
+      "loss": 0.3957,
+      "step": 10322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.353906077031767e-07,
+      "loss": 0.3927,
+      "step": 10323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.344809301665699e-07,
+      "loss": 0.3118,
+      "step": 10324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.335720048795756e-07,
+      "loss": 0.3563,
+      "step": 10325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.326638319144229e-07,
+      "loss": 0.335,
+      "step": 10326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.317564113432883e-07,
+      "loss": 0.3176,
+      "step": 10327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.308497432382875e-07,
+      "loss": 0.3044,
+      "step": 10328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.299438276714753e-07,
+      "loss": 0.3888,
+      "step": 10329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.29038664714846e-07,
+      "loss": 0.3232,
+      "step": 10330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.281342544403378e-07,
+      "loss": 0.3178,
+      "step": 10331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.272305969198243e-07,
+      "loss": 0.3951,
+      "step": 10332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.263276922251226e-07,
+      "loss": 0.3345,
+      "step": 10333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.254255404279885e-07,
+      "loss": 0.327,
+      "step": 10334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.24524141600119e-07,
+      "loss": 0.3188,
+      "step": 10335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.236234958131503e-07,
+      "loss": 0.352,
+      "step": 10336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.227236031386584e-07,
+      "loss": 0.3381,
+      "step": 10337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.218244636481607e-07,
+      "loss": 0.3349,
+      "step": 10338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.209260774131153e-07,
+      "loss": 0.3494,
+      "step": 10339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.200284445049176e-07,
+      "loss": 0.3225,
+      "step": 10340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.191315649949046e-07,
+      "loss": 0.3364,
+      "step": 10341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.182354389543554e-07,
+      "loss": 0.3482,
+      "step": 10342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.173400664544859e-07,
+      "loss": 0.3408,
+      "step": 10343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.164454475664549e-07,
+      "loss": 0.3055,
+      "step": 10344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.155515823613599e-07,
+      "loss": 0.331,
+      "step": 10345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.146584709102397e-07,
+      "loss": 0.3347,
+      "step": 10346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.137661132840677e-07,
+      "loss": 0.3583,
+      "step": 10347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.128745095537668e-07,
+      "loss": 0.3083,
+      "step": 10348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.119836597901951e-07,
+      "loss": 0.3533,
+      "step": 10349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.110935640641457e-07,
+      "loss": 0.3503,
+      "step": 10350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.102042224463622e-07,
+      "loss": 0.3294,
+      "step": 10351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.09315635007519e-07,
+      "loss": 0.3465,
+      "step": 10352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.084278018182387e-07,
+      "loss": 0.3021,
+      "step": 10353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.075407229490781e-07,
+      "loss": 0.329,
+      "step": 10354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.066543984705319e-07,
+      "loss": 0.3347,
+      "step": 10355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.057688284530449e-07,
+      "loss": 0.3623,
+      "step": 10356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.048840129669885e-07,
+      "loss": 0.3292,
+      "step": 10357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.039999520826899e-07,
+      "loss": 0.2903,
+      "step": 10358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.031166458704007e-07,
+      "loss": 0.3283,
+      "step": 10359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.022340944003213e-07,
+      "loss": 0.33,
+      "step": 10360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.013522977425933e-07,
+      "loss": 0.319,
+      "step": 10361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.004712559672908e-07,
+      "loss": 0.3454,
+      "step": 10362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.995909691444367e-07,
+      "loss": 0.3413,
+      "step": 10363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.987114373439871e-07,
+      "loss": 0.3308,
+      "step": 10364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.978326606358397e-07,
+      "loss": 0.3449,
+      "step": 10365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.969546390898383e-07,
+      "loss": 0.322,
+      "step": 10366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.960773727757562e-07,
+      "loss": 0.3045,
+      "step": 10367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.952008617633153e-07,
+      "loss": 0.342,
+      "step": 10368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.943251061221721e-07,
+      "loss": 0.3809,
+      "step": 10369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.934501059219243e-07,
+      "loss": 0.36,
+      "step": 10370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.925758612321163e-07,
+      "loss": 0.3173,
+      "step": 10371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.917023721222203e-07,
+      "loss": 0.3072,
+      "step": 10372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.908296386616573e-07,
+      "loss": 0.34,
+      "step": 10373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.899576609197854e-07,
+      "loss": 0.3573,
+      "step": 10374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.890864389659034e-07,
+      "loss": 0.344,
+      "step": 10375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.882159728692481e-07,
+      "loss": 0.3609,
+      "step": 10376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.873462626989999e-07,
+      "loss": 0.3147,
+      "step": 10377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.864773085242758e-07,
+      "loss": 0.2902,
+      "step": 10378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.856091104141336e-07,
+      "loss": 0.3331,
+      "step": 10379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.847416684375717e-07,
+      "loss": 0.3258,
+      "step": 10380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.83874982663528e-07,
+      "loss": 0.3549,
+      "step": 10381
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8300905316088e-07,
+      "loss": 0.3349,
+      "step": 10382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.821438799984457e-07,
+      "loss": 0.3186,
+      "step": 10383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.812794632449825e-07,
+      "loss": 0.3147,
+      "step": 10384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.804158029691886e-07,
+      "loss": 0.3301,
+      "step": 10385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.795528992397002e-07,
+      "loss": 0.3326,
+      "step": 10386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.786907521250961e-07,
+      "loss": 0.3003,
+      "step": 10387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7782936169389e-07,
+      "loss": 0.3205,
+      "step": 10388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.769687280145419e-07,
+      "loss": 0.3322,
+      "step": 10389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7610885115545014e-07,
+      "loss": 0.3073,
+      "step": 10390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.752497311849458e-07,
+      "loss": 0.331,
+      "step": 10391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.743913681713108e-07,
+      "loss": 0.3825,
+      "step": 10392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.735337621827574e-07,
+      "loss": 0.3583,
+      "step": 10393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7267691328744534e-07,
+      "loss": 0.3324,
+      "step": 10394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.718208215534703e-07,
+      "loss": 0.3997,
+      "step": 10395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7096548704886446e-07,
+      "loss": 0.3419,
+      "step": 10396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.701109098416079e-07,
+      "loss": 0.3241,
+      "step": 10397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6925708999961406e-07,
+      "loss": 0.3195,
+      "step": 10398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6840402759073754e-07,
+      "loss": 0.3282,
+      "step": 10399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.675517226827742e-07,
+      "loss": 0.3575,
+      "step": 10400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.667001753434575e-07,
+      "loss": 0.3892,
+      "step": 10401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6584938564046665e-07,
+      "loss": 0.3658,
+      "step": 10402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6499935364141213e-07,
+      "loss": 0.3377,
+      "step": 10403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.641500794138487e-07,
+      "loss": 0.3394,
+      "step": 10404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.633015630252713e-07,
+      "loss": 0.3137,
+      "step": 10405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.624538045431126e-07,
+      "loss": 0.334,
+      "step": 10406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.616068040347488e-07,
+      "loss": 0.3122,
+      "step": 10407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.607605615674904e-07,
+      "loss": 0.3809,
+      "step": 10408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5991507720859143e-07,
+      "loss": 0.3309,
+      "step": 10409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5907035102524475e-07,
+      "loss": 0.3512,
+      "step": 10410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5822638308458344e-07,
+      "loss": 0.3207,
+      "step": 10411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.573831734536782e-07,
+      "loss": 0.3458,
+      "step": 10412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5654072219954435e-07,
+      "loss": 0.3767,
+      "step": 10413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.556990293891306e-07,
+      "loss": 0.3152,
+      "step": 10414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.548580950893289e-07,
+      "loss": 0.4156,
+      "step": 10415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.540179193669725e-07,
+      "loss": 0.344,
+      "step": 10416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.531785022888302e-07,
+      "loss": 0.2872,
+      "step": 10417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5233984392161425e-07,
+      "loss": 0.3758,
+      "step": 10418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.515019443319746e-07,
+      "loss": 0.375,
+      "step": 10419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.506648035865002e-07,
+      "loss": 0.3259,
+      "step": 10420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4982842175172124e-07,
+      "loss": 0.2938,
+      "step": 10421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4899279889410896e-07,
+      "loss": 0.3407,
+      "step": 10422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.481579350800702e-07,
+      "loss": 0.3559,
+      "step": 10423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.473238303759553e-07,
+      "loss": 0.341,
+      "step": 10424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.464904848480522e-07,
+      "loss": 0.347,
+      "step": 10425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.456578985625903e-07,
+      "loss": 0.3391,
+      "step": 10426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.448260715857333e-07,
+      "loss": 0.3656,
+      "step": 10427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4399500398359386e-07,
+      "loss": 0.3565,
+      "step": 10428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4316469582221466e-07,
+      "loss": 0.3223,
+      "step": 10429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.423351471675852e-07,
+      "loss": 0.319,
+      "step": 10430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.415063580856327e-07,
+      "loss": 0.3535,
+      "step": 10431
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4067832864222003e-07,
+      "loss": 0.3147,
+      "step": 10432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.398510589031557e-07,
+      "loss": 0.3378,
+      "step": 10433
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.390245489341827e-07,
+      "loss": 0.3242,
+      "step": 10434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3819879880098726e-07,
+      "loss": 0.3684,
+      "step": 10435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3737380856919587e-07,
+      "loss": 0.3588,
+      "step": 10436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.365495783043683e-07,
+      "loss": 0.334,
+      "step": 10437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.357261080720121e-07,
+      "loss": 0.3343,
+      "step": 10438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.349033979375683e-07,
+      "loss": 0.3389,
+      "step": 10439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3408144796641905e-07,
+      "loss": 0.3134,
+      "step": 10440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.332602582238921e-07,
+      "loss": 0.3203,
+      "step": 10441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.324398287752418e-07,
+      "loss": 0.3015,
+      "step": 10442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.316201596856773e-07,
+      "loss": 0.33,
+      "step": 10443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.308012510203352e-07,
+      "loss": 0.3202,
+      "step": 10444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.299831028442969e-07,
+      "loss": 0.3302,
+      "step": 10445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.291657152225837e-07,
+      "loss": 0.3613,
+      "step": 10446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2834908822015574e-07,
+      "loss": 0.3381,
+      "step": 10447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2753322190191126e-07,
+      "loss": 0.3169,
+      "step": 10448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2671811633269055e-07,
+      "loss": 0.3838,
+      "step": 10449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.259037715772718e-07,
+      "loss": 0.3352,
+      "step": 10450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2509018770037326e-07,
+      "loss": 0.3767,
+      "step": 10451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2427736476665207e-07,
+      "loss": 0.3738,
+      "step": 10452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.234653028407054e-07,
+      "loss": 0.3448,
+      "step": 10453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.226540019870706e-07,
+      "loss": 0.3355,
+      "step": 10454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2184346227022276e-07,
+      "loss": 0.3214,
+      "step": 10455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.210336837545792e-07,
+      "loss": 0.3647,
+      "step": 10456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.20224666504494e-07,
+      "loss": 0.2919,
+      "step": 10457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.194164105842613e-07,
+      "loss": 0.3361,
+      "step": 10458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1860891605811747e-07,
+      "loss": 0.3392,
+      "step": 10459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.178021829902346e-07,
+      "loss": 0.3528,
+      "step": 10460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.169962114447257e-07,
+      "loss": 0.3496,
+      "step": 10461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.161910014856452e-07,
+      "loss": 0.3135,
+      "step": 10462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.153865531769818e-07,
+      "loss": 0.3389,
+      "step": 10463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1458286658267123e-07,
+      "loss": 0.3246,
+      "step": 10464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.137799417665822e-07,
+      "loss": 0.3922,
+      "step": 10465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1297777879252486e-07,
+      "loss": 0.3401,
+      "step": 10466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1217637772425157e-07,
+      "loss": 0.3534,
+      "step": 10467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.11375738625448e-07,
+      "loss": 0.3296,
+      "step": 10468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1057586155974773e-07,
+      "loss": 0.3105,
+      "step": 10469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0977674659071433e-07,
+      "loss": 0.377,
+      "step": 10470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.089783937818592e-07,
+      "loss": 0.3382,
+      "step": 10471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0818080319662943e-07,
+      "loss": 0.3513,
+      "step": 10472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0738397489840767e-07,
+      "loss": 0.3174,
+      "step": 10473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0658790895052536e-07,
+      "loss": 0.3206,
+      "step": 10474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.057926054162431e-07,
+      "loss": 0.3073,
+      "step": 10475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.049980643587681e-07,
+      "loss": 0.3043,
+      "step": 10476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.042042858412454e-07,
+      "loss": 0.3538,
+      "step": 10477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0341126992675673e-07,
+      "loss": 0.3474,
+      "step": 10478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0261901667832724e-07,
+      "loss": 0.3351,
+      "step": 10479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0182752615891775e-07,
+      "loss": 0.2974,
+      "step": 10480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0103679843142895e-07,
+      "loss": 0.3819,
+      "step": 10481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.002468335587073e-07,
+      "loss": 0.3228,
+      "step": 10482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.994576316035259e-07,
+      "loss": 0.3565,
+      "step": 10483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9866919262861125e-07,
+      "loss": 0.3384,
+      "step": 10484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.978815166966199e-07,
+      "loss": 0.3205,
+      "step": 10485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.970946038701495e-07,
+      "loss": 0.3334,
+      "step": 10486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.96308454211739e-07,
+      "loss": 0.344,
+      "step": 10487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.955230677838662e-07,
+      "loss": 0.3506,
+      "step": 10488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9473844464894775e-07,
+      "loss": 0.3155,
+      "step": 10489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9395458486933936e-07,
+      "loss": 0.3309,
+      "step": 10490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9317148850733564e-07,
+      "loss": 0.3592,
+      "step": 10491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9238915562517357e-07,
+      "loss": 0.365,
+      "step": 10492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.916075862850255e-07,
+      "loss": 0.3386,
+      "step": 10493
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.908267805490051e-07,
+      "loss": 0.3385,
+      "step": 10494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9004673847916506e-07,
+      "loss": 0.3323,
+      "step": 10495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8926746013749684e-07,
+      "loss": 0.3542,
+      "step": 10496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.884889455859331e-07,
+      "loss": 0.3145,
+      "step": 10497
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8771119488634323e-07,
+      "loss": 0.3465,
+      "step": 10498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8693420810053784e-07,
+      "loss": 0.4193,
+      "step": 10499
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.861579852902653e-07,
+      "loss": 0.3596,
+      "step": 10500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.853825265172151e-07,
+      "loss": 0.3489,
+      "step": 10501
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8460783184301464e-07,
+      "loss": 0.3152,
+      "step": 10502
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8383390132923136e-07,
+      "loss": 0.3584,
+      "step": 10503
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.830607350373683e-07,
+      "loss": 0.329,
+      "step": 10504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8228833302887515e-07,
+      "loss": 0.3417,
+      "step": 10505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.81516695365135e-07,
+      "loss": 0.3409,
+      "step": 10506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8074582210747223e-07,
+      "loss": 0.3497,
+      "step": 10507
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.799757133171511e-07,
+      "loss": 0.3113,
+      "step": 10508
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.792063690553716e-07,
+      "loss": 0.3176,
+      "step": 10509
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.784377893832769e-07,
+      "loss": 0.3569,
+      "step": 10510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.776699743619505e-07,
+      "loss": 0.3293,
+      "step": 10511
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.76902924052408e-07,
+      "loss": 0.3321,
+      "step": 10512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.761366385156129e-07,
+      "loss": 0.3181,
+      "step": 10513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7537111781245974e-07,
+      "loss": 0.3196,
+      "step": 10514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.746063620037921e-07,
+      "loss": 0.3254,
+      "step": 10515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.738423711503825e-07,
+      "loss": 0.4042,
+      "step": 10516
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.730791453129479e-07,
+      "loss": 0.3647,
+      "step": 10517
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.723166845521475e-07,
+      "loss": 0.3093,
+      "step": 10518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7155498892856967e-07,
+      "loss": 0.3192,
+      "step": 10519
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7079405850275583e-07,
+      "loss": 0.3537,
+      "step": 10520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7003389333517324e-07,
+      "loss": 0.3708,
+      "step": 10521
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.692744934862358e-07,
+      "loss": 0.3701,
+      "step": 10522
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6851585901629737e-07,
+      "loss": 0.3727,
+      "step": 10523
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.677579899856454e-07,
+      "loss": 0.3417,
+      "step": 10524
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.670008864545105e-07,
+      "loss": 0.3271,
+      "step": 10525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.662445484830635e-07,
+      "loss": 0.3637,
+      "step": 10526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6548897613141067e-07,
+      "loss": 0.3406,
+      "step": 10527
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.647341694596007e-07,
+      "loss": 0.3277,
+      "step": 10528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.639801285276179e-07,
+      "loss": 0.3338,
+      "step": 10529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.632268533953909e-07,
+      "loss": 0.3142,
+      "step": 10530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6247434412278184e-07,
+      "loss": 0.3519,
+      "step": 10531
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.617226007695962e-07,
+      "loss": 0.3323,
+      "step": 10532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6097162339557513e-07,
+      "loss": 0.362,
+      "step": 10533
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6022141206040305e-07,
+      "loss": 0.3053,
+      "step": 10534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.594719668237001e-07,
+      "loss": 0.3315,
+      "step": 10535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.587232877450275e-07,
+      "loss": 0.2949,
+      "step": 10536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5797537488388326e-07,
+      "loss": 0.3565,
+      "step": 10537
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5722822829970636e-07,
+      "loss": 0.3387,
+      "step": 10538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.56481848051875e-07,
+      "loss": 0.3643,
+      "step": 10539
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5573623419970615e-07,
+      "loss": 0.366,
+      "step": 10540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.549913868024557e-07,
+      "loss": 0.3445,
+      "step": 10541
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5424730591931853e-07,
+      "loss": 0.3551,
+      "step": 10542
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.535039916094274e-07,
+      "loss": 0.3169,
+      "step": 10543
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.527614439318583e-07,
+      "loss": 0.334,
+      "step": 10544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5201966294561963e-07,
+      "loss": 0.3875,
+      "step": 10545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5127864870966535e-07,
+      "loss": 0.3438,
+      "step": 10546
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5053840128288495e-07,
+      "loss": 0.3287,
+      "step": 10547
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4979892072410816e-07,
+      "loss": 0.3092,
+      "step": 10548
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4906020709210455e-07,
+      "loss": 0.3057,
+      "step": 10549
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4832226044557827e-07,
+      "loss": 0.3503,
+      "step": 10550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475850808431791e-07,
+      "loss": 0.2983,
+      "step": 10551
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468486683434935e-07,
+      "loss": 0.3512,
+      "step": 10552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4611302300504024e-07,
+      "loss": 0.3595,
+      "step": 10553
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.453781448862903e-07,
+      "loss": 0.3181,
+      "step": 10554
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4464403404564027e-07,
+      "loss": 0.3446,
+      "step": 10555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.439106905414369e-07,
+      "loss": 0.3394,
+      "step": 10556
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.431781144319568e-07,
+      "loss": 0.3527,
+      "step": 10557
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.424463057754213e-07,
+      "loss": 0.3608,
+      "step": 10558
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.417152646299915e-07,
+      "loss": 0.322,
+      "step": 10559
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4098499105376105e-07,
+      "loss": 0.3511,
+      "step": 10560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.40255485104769e-07,
+      "loss": 0.3019,
+      "step": 10561
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.395267468409913e-07,
+      "loss": 0.3352,
+      "step": 10562
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.387987763203404e-07,
+      "loss": 0.3285,
+      "step": 10563
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.380715736006746e-07,
+      "loss": 0.3297,
+      "step": 10564
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3734513873978193e-07,
+      "loss": 0.389,
+      "step": 10565
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3661947179539633e-07,
+      "loss": 0.365,
+      "step": 10566
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.358945728251883e-07,
+      "loss": 0.3439,
+      "step": 10567
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3517044188676627e-07,
+      "loss": 0.3868,
+      "step": 10568
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.344470790376808e-07,
+      "loss": 0.3142,
+      "step": 10569
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3372448433541707e-07,
+      "loss": 0.3268,
+      "step": 10570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.330026578374046e-07,
+      "loss": 0.3321,
+      "step": 10571
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.322815996010054e-07,
+      "loss": 0.3645,
+      "step": 10572
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3156130968352686e-07,
+      "loss": 0.3309,
+      "step": 10573
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.308417881422099e-07,
+      "loss": 0.3783,
+      "step": 10574
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.301230350342388e-07,
+      "loss": 0.3083,
+      "step": 10575
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.294050504167334e-07,
+      "loss": 0.3298,
+      "step": 10576
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.286878343467548e-07,
+      "loss": 0.3094,
+      "step": 10577
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.279713868813006e-07,
+      "loss": 0.3821,
+      "step": 10578
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.272557080773098e-07,
+      "loss": 0.3449,
+      "step": 10579
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2654079799166016e-07,
+      "loss": 0.3236,
+      "step": 10580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.258266566811652e-07,
+      "loss": 0.3297,
+      "step": 10581
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.251132842025817e-07,
+      "loss": 0.3424,
+      "step": 10582
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.244006806126021e-07,
+      "loss": 0.3491,
+      "step": 10583
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2368884596785886e-07,
+      "loss": 0.3315,
+      "step": 10584
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2297778032492564e-07,
+      "loss": 0.352,
+      "step": 10585
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2226748374030723e-07,
+      "loss": 0.3443,
+      "step": 10586
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.215579562704585e-07,
+      "loss": 0.3518,
+      "step": 10587
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.208491979717654e-07,
+      "loss": 0.3268,
+      "step": 10588
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.201412089005529e-07,
+      "loss": 0.3342,
+      "step": 10589
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.194339891130904e-07,
+      "loss": 0.3443,
+      "step": 10590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.187275386655786e-07,
+      "loss": 0.3238,
+      "step": 10591
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.180218576141636e-07,
+      "loss": 0.3217,
+      "step": 10592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1731694601492834e-07,
+      "loss": 0.3272,
+      "step": 10593
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1661280392389137e-07,
+      "loss": 0.3651,
+      "step": 10594
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.159094313970146e-07,
+      "loss": 0.3445,
+      "step": 10595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1520682849019545e-07,
+      "loss": 0.3537,
+      "step": 10596
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.145049952592738e-07,
+      "loss": 0.3428,
+      "step": 10597
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.138039317600239e-07,
+      "loss": 0.3611,
+      "step": 10598
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.131036380481611e-07,
+      "loss": 0.3605,
+      "step": 10599
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1240411417934325e-07,
+      "loss": 0.3504,
+      "step": 10600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.117053602091591e-07,
+      "loss": 0.3264,
+      "step": 10601
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1100737619314203e-07,
+      "loss": 0.3375,
+      "step": 10602
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.103101621867621e-07,
+      "loss": 0.328,
+      "step": 10603
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.096137182454284e-07,
+      "loss": 0.3417,
+      "step": 10604
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0891804442449325e-07,
+      "loss": 0.3546,
+      "step": 10605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0822314077923796e-07,
+      "loss": 0.3244,
+      "step": 10606
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.075290073648918e-07,
+      "loss": 0.3505,
+      "step": 10607
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0683564423661716e-07,
+      "loss": 0.3224,
+      "step": 10608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0614305144952003e-07,
+      "loss": 0.3356,
+      "step": 10609
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0545122905863976e-07,
+      "loss": 0.306,
+      "step": 10610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.047601771189601e-07,
+      "loss": 0.3514,
+      "step": 10611
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0406989568539935e-07,
+      "loss": 0.304,
+      "step": 10612
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0338038481281475e-07,
+      "loss": 0.3571,
+      "step": 10613
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0269164455600577e-07,
+      "loss": 0.3423,
+      "step": 10614
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0200367496970753e-07,
+      "loss": 0.3293,
+      "step": 10615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0131647610859405e-07,
+      "loss": 0.3308,
+      "step": 10616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0063004802727947e-07,
+      "loss": 0.3174,
+      "step": 10617
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.999443907803168e-07,
+      "loss": 0.3403,
+      "step": 10618
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9925950442219573e-07,
+      "loss": 0.3362,
+      "step": 10619
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.985753890073462e-07,
+      "loss": 0.3279,
+      "step": 10620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.978920445901379e-07,
+      "loss": 0.3284,
+      "step": 10621
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9720947122487633e-07,
+      "loss": 0.3242,
+      "step": 10622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9652766896580807e-07,
+      "loss": 0.342,
+      "step": 10623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9584663786711766e-07,
+      "loss": 0.3279,
+      "step": 10624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.951663779829284e-07,
+      "loss": 0.3375,
+      "step": 10625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9448688936730364e-07,
+      "loss": 0.3945,
+      "step": 10626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.938081720742414e-07,
+      "loss": 0.3243,
+      "step": 10627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.931302261576829e-07,
+      "loss": 0.3494,
+      "step": 10628
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.924530516715074e-07,
+      "loss": 0.3581,
+      "step": 10629
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9177664866952835e-07,
+      "loss": 0.3357,
+      "step": 10630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.911010172055051e-07,
+      "loss": 0.3574,
+      "step": 10631
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9042615733312797e-07,
+      "loss": 0.3532,
+      "step": 10632
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.89752069106034e-07,
+      "loss": 0.3359,
+      "step": 10633
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8907875257779274e-07,
+      "loss": 0.3501,
+      "step": 10634
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8840620780191233e-07,
+      "loss": 0.3304,
+      "step": 10635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.877344348318456e-07,
+      "loss": 0.3473,
+      "step": 10636
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8706343372097765e-07,
+      "loss": 0.3217,
+      "step": 10637
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.863932045226348e-07,
+      "loss": 0.3069,
+      "step": 10638
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8572374729008203e-07,
+      "loss": 0.3349,
+      "step": 10639
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8505506207652246e-07,
+      "loss": 0.3303,
+      "step": 10640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8438714893510134e-07,
+      "loss": 0.3227,
+      "step": 10641
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.837200079188962e-07,
+      "loss": 0.3351,
+      "step": 10642
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8305363908092685e-07,
+      "loss": 0.3891,
+      "step": 10643
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8238804247415207e-07,
+      "loss": 0.3283,
+      "step": 10644
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.817232181514673e-07,
+      "loss": 0.3439,
+      "step": 10645
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.810591661657114e-07,
+      "loss": 0.3231,
+      "step": 10646
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8039588656965433e-07,
+      "loss": 0.3655,
+      "step": 10647
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7973337941601066e-07,
+      "loss": 0.3473,
+      "step": 10648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.790716447574304e-07,
+      "loss": 0.3558,
+      "step": 10649
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7841068264650494e-07,
+      "loss": 0.3752,
+      "step": 10650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7775049313576106e-07,
+      "loss": 0.3186,
+      "step": 10651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7709107627766686e-07,
+      "loss": 0.354,
+      "step": 10652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.764324321246281e-07,
+      "loss": 0.354,
+      "step": 10653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7577456072898854e-07,
+      "loss": 0.3318,
+      "step": 10654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.751174621430297e-07,
+      "loss": 0.357,
+      "step": 10655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.744611364189742e-07,
+      "loss": 0.3492,
+      "step": 10656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7380558360898255e-07,
+      "loss": 0.3599,
+      "step": 10657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.73150803765152e-07,
+      "loss": 0.3228,
+      "step": 10658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.724967969395209e-07,
+      "loss": 0.3336,
+      "step": 10659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7184356318406326e-07,
+      "loss": 0.3206,
+      "step": 10660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7119110255069414e-07,
+      "loss": 0.3386,
+      "step": 10661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.705394150912677e-07,
+      "loss": 0.3533,
+      "step": 10662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.698885008575725e-07,
+      "loss": 0.3352,
+      "step": 10663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6923835990133927e-07,
+      "loss": 0.3609,
+      "step": 10664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.685889922742391e-07,
+      "loss": 0.33,
+      "step": 10665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6794039802787386e-07,
+      "loss": 0.3398,
+      "step": 10666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6729257721379465e-07,
+      "loss": 0.3195,
+      "step": 10667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6664552988348025e-07,
+      "loss": 0.315,
+      "step": 10668
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.659992560883573e-07,
+      "loss": 0.3337,
+      "step": 10669
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6535375587978585e-07,
+      "loss": 0.3541,
+      "step": 10670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.647090293090626e-07,
+      "loss": 0.3871,
+      "step": 10671
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.640650764274311e-07,
+      "loss": 0.3423,
+      "step": 10672
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6342189728606135e-07,
+      "loss": 0.3383,
+      "step": 10673
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6277949193607377e-07,
+      "loss": 0.3287,
+      "step": 10674
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6213786042852075e-07,
+      "loss": 0.362,
+      "step": 10675
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.614970028143926e-07,
+      "loss": 0.3074,
+      "step": 10676
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6085691914462306e-07,
+      "loss": 0.3179,
+      "step": 10677
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6021760947007813e-07,
+      "loss": 0.3414,
+      "step": 10678
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.59579073841566e-07,
+      "loss": 0.3369,
+      "step": 10679
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.589413123098361e-07,
+      "loss": 0.3558,
+      "step": 10680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5830432492556676e-07,
+      "loss": 0.3272,
+      "step": 10681
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5766811173938753e-07,
+      "loss": 0.3511,
+      "step": 10682
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5703267280185684e-07,
+      "loss": 0.3683,
+      "step": 10683
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5639800816347426e-07,
+      "loss": 0.3544,
+      "step": 10684
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.557641178746795e-07,
+      "loss": 0.3489,
+      "step": 10685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5513100198584774e-07,
+      "loss": 0.3375,
+      "step": 10686
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5449866054729654e-07,
+      "loss": 0.3582,
+      "step": 10687
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5386709360927797e-07,
+      "loss": 0.3186,
+      "step": 10688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.532363012219863e-07,
+      "loss": 0.2948,
+      "step": 10689
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5260628343555027e-07,
+      "loss": 0.3489,
+      "step": 10690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.519770403000399e-07,
+      "loss": 0.3278,
+      "step": 10691
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.513485718654618e-07,
+      "loss": 0.3143,
+      "step": 10692
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.507208781817638e-07,
+      "loss": 0.3871,
+      "step": 10693
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.500939592988294e-07,
+      "loss": 0.3108,
+      "step": 10694
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.49467815266482e-07,
+      "loss": 0.3258,
+      "step": 10695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.488424461344818e-07,
+      "loss": 0.338,
+      "step": 10696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.482178519525291e-07,
+      "loss": 0.3143,
+      "step": 10697
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4759403277026304e-07,
+      "loss": 0.3486,
+      "step": 10698
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4697098863725844e-07,
+      "loss": 0.3588,
+      "step": 10699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.463487196030312e-07,
+      "loss": 0.3578,
+      "step": 10700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.457272257170351e-07,
+      "loss": 0.3635,
+      "step": 10701
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4510650702865956e-07,
+      "loss": 0.3492,
+      "step": 10702
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4448656358723843e-07,
+      "loss": 0.3382,
+      "step": 10703
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4386739544203674e-07,
+      "loss": 0.299,
+      "step": 10704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4324900264226405e-07,
+      "loss": 0.3476,
+      "step": 10705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.426313852370654e-07,
+      "loss": 0.3186,
+      "step": 10706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4201454327552166e-07,
+      "loss": 0.3086,
+      "step": 10707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4139847680665685e-07,
+      "loss": 0.3394,
+      "step": 10708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4078318587943183e-07,
+      "loss": 0.3101,
+      "step": 10709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4016867054274526e-07,
+      "loss": 0.3729,
+      "step": 10710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.395549308454348e-07,
+      "loss": 0.3244,
+      "step": 10711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.389419668362736e-07,
+      "loss": 0.348,
+      "step": 10712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.383297785639782e-07,
+      "loss": 0.344,
+      "step": 10713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.377183660771998e-07,
+      "loss": 0.3327,
+      "step": 10714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.371077294245272e-07,
+      "loss": 0.3067,
+      "step": 10715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3649786865449164e-07,
+      "loss": 0.315,
+      "step": 10716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.358887838155588e-07,
+      "loss": 0.3567,
+      "step": 10717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.352804749561377e-07,
+      "loss": 0.3287,
+      "step": 10718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3467294212456747e-07,
+      "loss": 0.3468,
+      "step": 10719
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3406618536913173e-07,
+      "loss": 0.3123,
+      "step": 10720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3346020473805297e-07,
+      "loss": 0.3017,
+      "step": 10721
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3285500027948826e-07,
+      "loss": 0.3409,
+      "step": 10722
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.322505720415369e-07,
+      "loss": 0.3271,
+      "step": 10723
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3164692007223154e-07,
+      "loss": 0.3146,
+      "step": 10724
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3104404441954719e-07,
+      "loss": 0.3585,
+      "step": 10725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3044194513139551e-07,
+      "loss": 0.2979,
+      "step": 10726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.298406222556282e-07,
+      "loss": 0.3592,
+      "step": 10727
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2924007584003372e-07,
+      "loss": 0.337,
+      "step": 10728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2864030593233722e-07,
+      "loss": 0.3502,
+      "step": 10729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2804131258020612e-07,
+      "loss": 0.3312,
+      "step": 10730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2744309583124347e-07,
+      "loss": 0.3239,
+      "step": 10731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2684565573299122e-07,
+      "loss": 0.3332,
+      "step": 10732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2624899233292807e-07,
+      "loss": 0.3124,
+      "step": 10733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2565310567847387e-07,
+      "loss": 0.3135,
+      "step": 10734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2505799581698518e-07,
+      "loss": 0.3373,
+      "step": 10735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.244636627957564e-07,
+      "loss": 0.3141,
+      "step": 10736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.238701066620208e-07,
+      "loss": 0.3658,
+      "step": 10737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2327732746295072e-07,
+      "loss": 0.2917,
+      "step": 10738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.22685325245654e-07,
+      "loss": 0.3456,
+      "step": 10739
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.220941000571808e-07,
+      "loss": 0.3365,
+      "step": 10740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2150365194451572e-07,
+      "loss": 0.335,
+      "step": 10741
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2091398095458572e-07,
+      "loss": 0.3158,
+      "step": 10742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2032508713424883e-07,
+      "loss": 0.3497,
+      "step": 10743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1973697053031094e-07,
+      "loss": 0.3407,
+      "step": 10744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1914963118950804e-07,
+      "loss": 0.3698,
+      "step": 10745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1856306915851833e-07,
+      "loss": 0.3434,
+      "step": 10746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1797728448395894e-07,
+      "loss": 0.313,
+      "step": 10747
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1739227721238043e-07,
+      "loss": 0.3392,
+      "step": 10748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1680804739027782e-07,
+      "loss": 0.3164,
+      "step": 10749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1622459506408178e-07,
+      "loss": 0.3663,
+      "step": 10750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.156419202801574e-07,
+      "loss": 0.3379,
+      "step": 10751
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1506002308481545e-07,
+      "loss": 0.3463,
+      "step": 10752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1447890352429668e-07,
+      "loss": 0.3547,
+      "step": 10753
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.138985616447875e-07,
+      "loss": 0.299,
+      "step": 10754
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.133189974924077e-07,
+      "loss": 0.3222,
+      "step": 10755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1274021111321597e-07,
+      "loss": 0.3424,
+      "step": 10756
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.121622025532133e-07,
+      "loss": 0.3852,
+      "step": 10757
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1158497185833072e-07,
+      "loss": 0.3829,
+      "step": 10758
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1100851907444708e-07,
+      "loss": 0.3418,
+      "step": 10759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1043284424737132e-07,
+      "loss": 0.335,
+      "step": 10760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.098579474228546e-07,
+      "loss": 0.3233,
+      "step": 10761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0928382864658702e-07,
+      "loss": 0.3401,
+      "step": 10762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0871048796419323e-07,
+      "loss": 0.32,
+      "step": 10763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0813792542123901e-07,
+      "loss": 0.3693,
+      "step": 10764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.075661410632268e-07,
+      "loss": 0.3382,
+      "step": 10765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.069951349355981e-07,
+      "loss": 0.3575,
+      "step": 10766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.064249070837332e-07,
+      "loss": 0.3206,
+      "step": 10767
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0585545755294923e-07,
+      "loss": 0.388,
+      "step": 10768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0528678638850108e-07,
+      "loss": 0.3398,
+      "step": 10769
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.047188936355826e-07,
+      "loss": 0.352,
+      "step": 10770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0415177933932662e-07,
+      "loss": 0.3195,
+      "step": 10771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0358544354480148e-07,
+      "loss": 0.3816,
+      "step": 10772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0301988629701786e-07,
+      "loss": 0.3309,
+      "step": 10773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.024551076409198e-07,
+      "loss": 0.3334,
+      "step": 10774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0189110762139362e-07,
+      "loss": 0.3486,
+      "step": 10775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0132788628326018e-07,
+      "loss": 0.3505,
+      "step": 10776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0076544367128137e-07,
+      "loss": 0.2916,
+      "step": 10777
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0020377983015592e-07,
+      "loss": 0.3149,
+      "step": 10778
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9964289480451925e-07,
+      "loss": 0.3112,
+      "step": 10779
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9908278863894793e-07,
+      "loss": 0.3335,
+      "step": 10780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.985234613779552e-07,
+      "loss": 0.344,
+      "step": 10781
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9796491306599108e-07,
+      "loss": 0.3725,
+      "step": 10782
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9740714374744562e-07,
+      "loss": 0.3299,
+      "step": 10783
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968501534666456e-07,
+      "loss": 0.3538,
+      "step": 10784
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.962939422678567e-07,
+      "loss": 0.3618,
+      "step": 10785
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9573851019528355e-07,
+      "loss": 0.351,
+      "step": 10786
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9518385729306643e-07,
+      "loss": 0.3422,
+      "step": 10787
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9462998360528673e-07,
+      "loss": 0.349,
+      "step": 10788
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9407688917595923e-07,
+      "loss": 0.4068,
+      "step": 10789
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9352457404904212e-07,
+      "loss": 0.3103,
+      "step": 10790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9297303826843027e-07,
+      "loss": 0.3312,
+      "step": 10791
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9242228187795197e-07,
+      "loss": 0.3457,
+      "step": 10792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.918723049213822e-07,
+      "loss": 0.3494,
+      "step": 10793
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9132310744242377e-07,
+      "loss": 0.3378,
+      "step": 10794
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9077468948472844e-07,
+      "loss": 0.3714,
+      "step": 10795
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9022705109187578e-07,
+      "loss": 0.381,
+      "step": 10796
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8968019230738876e-07,
+      "loss": 0.379,
+      "step": 10797
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.891341131747304e-07,
+      "loss": 0.35,
+      "step": 10798
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8858881373729708e-07,
+      "loss": 0.3688,
+      "step": 10799
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8804429403842527e-07,
+      "loss": 0.3084,
+      "step": 10800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8750055412138922e-07,
+      "loss": 0.3641,
+      "step": 10801
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8695759402940105e-07,
+      "loss": 0.3493,
+      "step": 10802
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.864154138056129e-07,
+      "loss": 0.3264,
+      "step": 10803
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8587401349311252e-07,
+      "loss": 0.3197,
+      "step": 10804
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.853333931349255e-07,
+      "loss": 0.323,
+      "step": 10805
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8479355277401635e-07,
+      "loss": 0.3184,
+      "step": 10806
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8425449245328964e-07,
+      "loss": 0.3492,
+      "step": 10807
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8371621221558334e-07,
+      "loss": 0.3035,
+      "step": 10808
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8317871210367767e-07,
+      "loss": 0.3557,
+      "step": 10809
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8264199216028844e-07,
+      "loss": 0.3248,
+      "step": 10810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8210605242806933e-07,
+      "loss": 0.3462,
+      "step": 10811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8157089294961516e-07,
+      "loss": 0.3114,
+      "step": 10812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8103651376745522e-07,
+      "loss": 0.3288,
+      "step": 10813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.805029149240578e-07,
+      "loss": 0.3303,
+      "step": 10814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7997009646182896e-07,
+      "loss": 0.3479,
+      "step": 10815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.794380584231148e-07,
+      "loss": 0.3185,
+      "step": 10816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7890680085019597e-07,
+      "loss": 0.3074,
+      "step": 10817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7837632378529424e-07,
+      "loss": 0.3923,
+      "step": 10818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7784662727056812e-07,
+      "loss": 0.3238,
+      "step": 10819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7731771134811282e-07,
+      "loss": 0.315,
+      "step": 10820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.767895760599625e-07,
+      "loss": 0.3065,
+      "step": 10821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7626222144809136e-07,
+      "loss": 0.3418,
+      "step": 10822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7573564755440807e-07,
+      "loss": 0.2975,
+      "step": 10823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7520985442076144e-07,
+      "loss": 0.3404,
+      "step": 10824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7468484208893689e-07,
+      "loss": 0.3383,
+      "step": 10825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.741606106006588e-07,
+      "loss": 0.3169,
+      "step": 10826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7363715999759057e-07,
+      "loss": 0.3167,
+      "step": 10827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7311449032133e-07,
+      "loss": 0.3308,
+      "step": 10828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7259260161341718e-07,
+      "loss": 0.3229,
+      "step": 10829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7207149391532563e-07,
+      "loss": 0.3286,
+      "step": 10830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7155116726847109e-07,
+      "loss": 0.3488,
+      "step": 10831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7103162171420605e-07,
+      "loss": 0.3364,
+      "step": 10832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.705128572938164e-07,
+      "loss": 0.3172,
+      "step": 10833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6999487404853353e-07,
+      "loss": 0.3294,
+      "step": 10834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6947767201952126e-07,
+      "loss": 0.375,
+      "step": 10835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6896125124788333e-07,
+      "loss": 0.3457,
+      "step": 10836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6844561177466025e-07,
+      "loss": 0.3175,
+      "step": 10837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6793075364083143e-07,
+      "loss": 0.3229,
+      "step": 10838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6741667688731644e-07,
+      "loss": 0.3208,
+      "step": 10839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.66903381554967e-07,
+      "loss": 0.3379,
+      "step": 10840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6639086768457713e-07,
+      "loss": 0.318,
+      "step": 10841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.658791353168776e-07,
+      "loss": 0.3376,
+      "step": 10842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.653681844925381e-07,
+      "loss": 0.3115,
+      "step": 10843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6485801525216505e-07,
+      "loss": 0.3214,
+      "step": 10844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6434862763630156e-07,
+      "loss": 0.3003,
+      "step": 10845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6384002168543077e-07,
+      "loss": 0.3472,
+      "step": 10846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6333219743997374e-07,
+      "loss": 0.3127,
+      "step": 10847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6282515494028705e-07,
+      "loss": 0.3625,
+      "step": 10848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6231889422666737e-07,
+      "loss": 0.3311,
+      "step": 10849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6181341533934913e-07,
+      "loss": 0.3508,
+      "step": 10850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.613087183185047e-07,
+      "loss": 0.3777,
+      "step": 10851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6080480320424086e-07,
+      "loss": 0.3354,
+      "step": 10852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6030167003660778e-07,
+      "loss": 0.3829,
+      "step": 10853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5979931885559018e-07,
+      "loss": 0.3468,
+      "step": 10854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5929774970110944e-07,
+      "loss": 0.3479,
+      "step": 10855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5879696261302923e-07,
+      "loss": 0.3337,
+      "step": 10856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.582969576311466e-07,
+      "loss": 0.3313,
+      "step": 10857
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5779773479519867e-07,
+      "loss": 0.3381,
+      "step": 10858
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5729929414486145e-07,
+      "loss": 0.295,
+      "step": 10859
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5680163571974439e-07,
+      "loss": 0.3095,
+      "step": 10860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5630475955940027e-07,
+      "loss": 0.3029,
+      "step": 10861
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5580866570331644e-07,
+      "loss": 0.3296,
+      "step": 10862
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5531335419091797e-07,
+      "loss": 0.3331,
+      "step": 10863
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5481882506156898e-07,
+      "loss": 0.3356,
+      "step": 10864
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5432507835457243e-07,
+      "loss": 0.315,
+      "step": 10865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5383211410916477e-07,
+      "loss": 0.3488,
+      "step": 10866
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.533399323645246e-07,
+      "loss": 0.3235,
+      "step": 10867
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5284853315976955e-07,
+      "loss": 0.3815,
+      "step": 10868
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5235791653394726e-07,
+      "loss": 0.3054,
+      "step": 10869
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5186808252605211e-07,
+      "loss": 0.3269,
+      "step": 10870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5137903117501075e-07,
+      "loss": 0.3118,
+      "step": 10871
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5089076251969093e-07,
+      "loss": 0.328,
+      "step": 10872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.504032765988961e-07,
+      "loss": 0.3445,
+      "step": 10873
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4991657345136635e-07,
+      "loss": 0.3599,
+      "step": 10874
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4943065311578297e-07,
+      "loss": 0.3315,
+      "step": 10875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4894551563076286e-07,
+      "loss": 0.3576,
+      "step": 10876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4846116103486186e-07,
+      "loss": 0.3758,
+      "step": 10877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4797758936657137e-07,
+      "loss": 0.3252,
+      "step": 10878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4749480066432286e-07,
+      "loss": 0.3323,
+      "step": 10879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.470127949664868e-07,
+      "loss": 0.382,
+      "step": 10880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4653157231136583e-07,
+      "loss": 0.3513,
+      "step": 10881
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4605113273720606e-07,
+      "loss": 0.3076,
+      "step": 10882
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4557147628218914e-07,
+      "loss": 0.3485,
+      "step": 10883
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.450926029844335e-07,
+      "loss": 0.3097,
+      "step": 10884
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4461451288199868e-07,
+      "loss": 0.3387,
+      "step": 10885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4413720601287762e-07,
+      "loss": 0.3039,
+      "step": 10886
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4366068241500442e-07,
+      "loss": 0.3331,
+      "step": 10887
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4318494212624989e-07,
+      "loss": 0.3199,
+      "step": 10888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4270998518442047e-07,
+      "loss": 0.3587,
+      "step": 10889
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4223581162726374e-07,
+      "loss": 0.3314,
+      "step": 10890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.41762421492464e-07,
+      "loss": 0.293,
+      "step": 10891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4128981481764115e-07,
+      "loss": 0.3707,
+      "step": 10892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4081799164035625e-07,
+      "loss": 0.3382,
+      "step": 10893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.403469519981049e-07,
+      "loss": 0.357,
+      "step": 10894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3987669592832265e-07,
+      "loss": 0.3648,
+      "step": 10895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3940722346838075e-07,
+      "loss": 0.3383,
+      "step": 10896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.389385346555916e-07,
+      "loss": 0.3627,
+      "step": 10897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3847062952720093e-07,
+      "loss": 0.3333,
+      "step": 10898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3800350812039675e-07,
+      "loss": 0.2982,
+      "step": 10899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.375371704722994e-07,
+      "loss": 0.3397,
+      "step": 10900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.370716166199726e-07,
+      "loss": 0.3497,
+      "step": 10901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3660684660041446e-07,
+      "loss": 0.3086,
+      "step": 10902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3614286045056102e-07,
+      "loss": 0.3396,
+      "step": 10903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3567965820728613e-07,
+      "loss": 0.349,
+      "step": 10904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3521723990740255e-07,
+      "loss": 0.3356,
+      "step": 10905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3475560558765977e-07,
+      "loss": 0.3192,
+      "step": 10906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.342947552847429e-07,
+      "loss": 0.3188,
+      "step": 10907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.338346890352804e-07,
+      "loss": 0.3414,
+      "step": 10908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3337540687583416e-07,
+      "loss": 0.3248,
+      "step": 10909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3291690884290166e-07,
+      "loss": 0.3236,
+      "step": 10910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.324591949729248e-07,
+      "loss": 0.3362,
+      "step": 10911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3200226530227678e-07,
+      "loss": 0.3852,
+      "step": 10912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3154611986727072e-07,
+      "loss": 0.316,
+      "step": 10913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.31090758704161e-07,
+      "loss": 0.3251,
+      "step": 10914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3063618184913195e-07,
+      "loss": 0.3603,
+      "step": 10915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3018238933831362e-07,
+      "loss": 0.3477,
+      "step": 10916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2972938120776822e-07,
+      "loss": 0.3174,
+      "step": 10917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.29277157493497e-07,
+      "loss": 0.2766,
+      "step": 10918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.288257182314412e-07,
+      "loss": 0.3734,
+      "step": 10919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.283750634574754e-07,
+      "loss": 0.3844,
+      "step": 10920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.279251932074177e-07,
+      "loss": 0.3289,
+      "step": 10921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2747610751701834e-07,
+      "loss": 0.3718,
+      "step": 10922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2702780642196654e-07,
+      "loss": 0.3413,
+      "step": 10923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2658028995789162e-07,
+      "loss": 0.3202,
+      "step": 10924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2613355816035845e-07,
+      "loss": 0.3311,
+      "step": 10925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.256876110648686e-07,
+      "loss": 0.3385,
+      "step": 10926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2524244870686486e-07,
+      "loss": 0.3278,
+      "step": 10927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2479807112172448e-07,
+      "loss": 0.4009,
+      "step": 10928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2435447834476254e-07,
+      "loss": 0.3444,
+      "step": 10929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.239116704112342e-07,
+      "loss": 0.3034,
+      "step": 10930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2346964735632905e-07,
+      "loss": 0.3769,
+      "step": 10931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.230284092151768e-07,
+      "loss": 0.3215,
+      "step": 10932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.225879560228427e-07,
+      "loss": 0.352,
+      "step": 10933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2214828781433207e-07,
+      "loss": 0.3317,
+      "step": 10934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2170940462458593e-07,
+      "loss": 0.332,
+      "step": 10935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2127130648848295e-07,
+      "loss": 0.326,
+      "step": 10936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.208339934408398e-07,
+      "loss": 0.2997,
+      "step": 10937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2039746551641307e-07,
+      "loss": 0.3768,
+      "step": 10938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.199617227498917e-07,
+      "loss": 0.3468,
+      "step": 10939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.19526765175908e-07,
+      "loss": 0.3752,
+      "step": 10940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1909259282902652e-07,
+      "loss": 0.3688,
+      "step": 10941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1865920574375411e-07,
+      "loss": 0.3375,
+      "step": 10942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1822660395453323e-07,
+      "loss": 0.3303,
+      "step": 10943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1779478749574302e-07,
+      "loss": 0.3347,
+      "step": 10944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1736375640170161e-07,
+      "loss": 0.3527,
+      "step": 10945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1693351070666159e-07,
+      "loss": 0.341,
+      "step": 10946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1650405044482005e-07,
+      "loss": 0.332,
+      "step": 10947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1607537565030524e-07,
+      "loss": 0.3068,
+      "step": 10948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1564748635718436e-07,
+      "loss": 0.3933,
+      "step": 10949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1522038259946466e-07,
+      "loss": 0.3064,
+      "step": 10950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1479406441108676e-07,
+      "loss": 0.3196,
+      "step": 10951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1436853182593354e-07,
+      "loss": 0.3471,
+      "step": 10952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1394378487782242e-07,
+      "loss": 0.3101,
+      "step": 10953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1351982360050751e-07,
+      "loss": 0.3411,
+      "step": 10954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1309664802768627e-07,
+      "loss": 0.3066,
+      "step": 10955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1267425819298406e-07,
+      "loss": 0.3053,
+      "step": 10956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.122526541299751e-07,
+      "loss": 0.3258,
+      "step": 10957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.118318358721604e-07,
+      "loss": 0.3438,
+      "step": 10958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1141180345298542e-07,
+      "loss": 0.3968,
+      "step": 10959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1099255690583344e-07,
+      "loss": 0.3561,
+      "step": 10960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1057409626401894e-07,
+      "loss": 0.3254,
+      "step": 10961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1015642156080086e-07,
+      "loss": 0.3175,
+      "step": 10962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0973953282937267e-07,
+      "loss": 0.3366,
+      "step": 10963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.093234301028645e-07,
+      "loss": 0.3014,
+      "step": 10964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.089081134143466e-07,
+      "loss": 0.3459,
+      "step": 10965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0849358279682365e-07,
+      "loss": 0.3699,
+      "step": 10966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0807983828324043e-07,
+      "loss": 0.3258,
+      "step": 10967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.076668799064784e-07,
+      "loss": 0.376,
+      "step": 10968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0725470769935687e-07,
+      "loss": 0.3781,
+      "step": 10969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0684332169463074e-07,
+      "loss": 0.3347,
+      "step": 10970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0643272192499499e-07,
+      "loss": 0.3466,
+      "step": 10971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0602290842308017e-07,
+      "loss": 0.3127,
+      "step": 10972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0561388122145688e-07,
+      "loss": 0.3189,
+      "step": 10973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0520564035263026e-07,
+      "loss": 0.3041,
+      "step": 10974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0479818584904433e-07,
+      "loss": 0.3511,
+      "step": 10975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0439151774308098e-07,
+      "loss": 0.3222,
+      "step": 10976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0398563606705881e-07,
+      "loss": 0.32,
+      "step": 10977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0358054085323532e-07,
+      "loss": 0.3392,
+      "step": 10978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0317623213380368e-07,
+      "loss": 0.3437,
+      "step": 10979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0277270994089483e-07,
+      "loss": 0.3149,
+      "step": 10980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0236997430657869e-07,
+      "loss": 0.3413,
+      "step": 10981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0196802526286076e-07,
+      "loss": 0.315,
+      "step": 10982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.015668628416866e-07,
+      "loss": 0.4185,
+      "step": 10983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0116648707493626e-07,
+      "loss": 0.3416,
+      "step": 10984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0076689799442874e-07,
+      "loss": 0.3239,
+      "step": 10985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0036809563192085e-07,
+      "loss": 0.3101,
+      "step": 10986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.997008001910613e-08,
+      "loss": 0.3345,
+      "step": 10987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.957285118761595e-08,
+      "loss": 0.3234,
+      "step": 10988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.917640916902061e-08,
+      "loss": 0.3646,
+      "step": 10989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.878075399482268e-08,
+      "loss": 0.3339,
+      "step": 10990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.838588569647034e-08,
+      "loss": 0.2763,
+      "step": 10991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.799180430534184e-08,
+      "loss": 0.3694,
+      "step": 10992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.759850985275765e-08,
+      "loss": 0.3464,
+      "step": 10993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.720600236997279e-08,
+      "loss": 0.3415,
+      "step": 10994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.68142818881801e-08,
+      "loss": 0.3237,
+      "step": 10995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.642334843851242e-08,
+      "loss": 0.3078,
+      "step": 10996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.603320205203714e-08,
+      "loss": 0.3612,
+      "step": 10997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.564384275976169e-08,
+      "loss": 0.3472,
+      "step": 10998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.525527059262684e-08,
+      "loss": 0.3349,
+      "step": 10999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.48674855815146e-08,
+      "loss": 0.3346,
+      "step": 11000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.448048775724583e-08,
+      "loss": 0.3507,
+      "step": 11001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.409427715057262e-08,
+      "loss": 0.3134,
+      "step": 11002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.37088537921893e-08,
+      "loss": 0.3407,
+      "step": 11003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.332421771272804e-08,
+      "loss": 0.3609,
+      "step": 11004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.294036894275438e-08,
+      "loss": 0.34,
+      "step": 11005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.255730751277503e-08,
+      "loss": 0.3496,
+      "step": 11006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.217503345323342e-08,
+      "loss": 0.334,
+      "step": 11007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.179354679450858e-08,
+      "loss": 0.317,
+      "step": 11008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.141284756691959e-08,
+      "loss": 0.3613,
+      "step": 11009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.103293580072004e-08,
+      "loss": 0.3704,
+      "step": 11010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.065381152610465e-08,
+      "loss": 0.3635,
+      "step": 11011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.027547477320154e-08,
+      "loss": 0.3497,
+      "step": 11012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.989792557207889e-08,
+      "loss": 0.3455,
+      "step": 11013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.952116395274158e-08,
+      "loss": 0.3097,
+      "step": 11014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.914518994513121e-08,
+      "loss": 0.319,
+      "step": 11015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.877000357912836e-08,
+      "loss": 0.362,
+      "step": 11016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.839560488455023e-08,
+      "loss": 0.3198,
+      "step": 11017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.802199389114974e-08,
+      "loss": 0.3334,
+      "step": 11018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.764917062861977e-08,
+      "loss": 0.3398,
+      "step": 11019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.727713512658997e-08,
+      "loss": 0.3379,
+      "step": 11020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.690588741462669e-08,
+      "loss": 0.3345,
+      "step": 11021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.653542752223299e-08,
+      "loss": 0.3516,
+      "step": 11022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.616575547885086e-08,
+      "loss": 0.3662,
+      "step": 11023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.579687131386016e-08,
+      "loss": 0.3694,
+      "step": 11024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.54287750565752e-08,
+      "loss": 0.3067,
+      "step": 11025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.506146673625149e-08,
+      "loss": 0.3185,
+      "step": 11026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.469494638207898e-08,
+      "loss": 0.3627,
+      "step": 11027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.43292140231855e-08,
+      "loss": 0.3059,
+      "step": 11028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.39642696886367e-08,
+      "loss": 0.3852,
+      "step": 11029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.360011340743823e-08,
+      "loss": 0.3613,
+      "step": 11030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.323674520852699e-08,
+      "loss": 0.36,
+      "step": 11031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.287416512078317e-08,
+      "loss": 0.3233,
+      "step": 11032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.251237317302041e-08,
+      "loss": 0.3457,
+      "step": 11033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.215136939399349e-08,
+      "loss": 0.3164,
+      "step": 11034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.179115381238945e-08,
+      "loss": 0.3479,
+      "step": 11035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.143172645683761e-08,
+      "loss": 0.3444,
+      "step": 11036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.10730873559018e-08,
+      "loss": 0.3855,
+      "step": 11037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.071523653808477e-08,
+      "loss": 0.3434,
+      "step": 11038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.035817403182489e-08,
+      "loss": 0.3506,
+      "step": 11039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.000189986549944e-08,
+      "loss": 0.3184,
+      "step": 11040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.964641406742135e-08,
+      "loss": 0.3422,
+      "step": 11041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.929171666584468e-08,
+      "loss": 0.3322,
+      "step": 11042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.893780768895576e-08,
+      "loss": 0.3496,
+      "step": 11043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.858468716488099e-08,
+      "loss": 0.3653,
+      "step": 11044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.82323551216846e-08,
+      "loss": 0.3784,
+      "step": 11045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.788081158736638e-08,
+      "loss": 0.3486,
+      "step": 11046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.753005658986513e-08,
+      "loss": 0.366,
+      "step": 11047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.718009015705518e-08,
+      "loss": 0.3068,
+      "step": 11048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.6830912316751e-08,
+      "loss": 0.3284,
+      "step": 11049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.648252309670145e-08,
+      "loss": 0.3483,
+      "step": 11050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.613492252459553e-08,
+      "loss": 0.3157,
+      "step": 11051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.578811062805557e-08,
+      "loss": 0.3659,
+      "step": 11052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.544208743464509e-08,
+      "loss": 0.3429,
+      "step": 11053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.50968529718632e-08,
+      "loss": 0.3426,
+      "step": 11054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.475240726714572e-08,
+      "loss": 0.4067,
+      "step": 11055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.440875034786854e-08,
+      "loss": 0.3213,
+      "step": 11056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.406588224134203e-08,
+      "loss": 0.3607,
+      "step": 11057
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.37238029748133e-08,
+      "loss": 0.3199,
+      "step": 11058
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.338251257547169e-08,
+      "loss": 0.3261,
+      "step": 11059
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.304201107043663e-08,
+      "loss": 0.3561,
+      "step": 11060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.270229848677201e-08,
+      "loss": 0.3011,
+      "step": 11061
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.236337485147294e-08,
+      "loss": 0.3385,
+      "step": 11062
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.202524019147672e-08,
+      "loss": 0.3909,
+      "step": 11063
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.16878945336541e-08,
+      "loss": 0.3274,
+      "step": 11064
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.135133790481586e-08,
+      "loss": 0.3577,
+      "step": 11065
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.101557033170837e-08,
+      "loss": 0.3121,
+      "step": 11066
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.068059184101694e-08,
+      "loss": 0.3096,
+      "step": 11067
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.034640245936253e-08,
+      "loss": 0.361,
+      "step": 11068
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.001300221330387e-08,
+      "loss": 0.349,
+      "step": 11069
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.968039112933755e-08,
+      "loss": 0.3618,
+      "step": 11070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.934856923389688e-08,
+      "loss": 0.3365,
+      "step": 11071
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.901753655335185e-08,
+      "loss": 0.3331,
+      "step": 11072
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.868729311401257e-08,
+      "loss": 0.3329,
+      "step": 11073
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.835783894212133e-08,
+      "loss": 0.3293,
+      "step": 11074
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.8029174063865e-08,
+      "loss": 0.3275,
+      "step": 11075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.770129850535934e-08,
+      "loss": 0.3427,
+      "step": 11076
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.737421229266238e-08,
+      "loss": 0.3226,
+      "step": 11077
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.704791545177003e-08,
+      "loss": 0.3399,
+      "step": 11078
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.672240800861262e-08,
+      "loss": 0.3743,
+      "step": 11079
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.639768998906059e-08,
+      "loss": 0.3309,
+      "step": 11080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.607376141891775e-08,
+      "loss": 0.3543,
+      "step": 11081
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.575062232392903e-08,
+      "loss": 0.3659,
+      "step": 11082
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.542827272977614e-08,
+      "loss": 0.3561,
+      "step": 11083
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.510671266207525e-08,
+      "loss": 0.3438,
+      "step": 11084
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.478594214638145e-08,
+      "loss": 0.3306,
+      "step": 11085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.446596120818882e-08,
+      "loss": 0.3463,
+      "step": 11086
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.414676987292589e-08,
+      "loss": 0.3112,
+      "step": 11087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.382836816596016e-08,
+      "loss": 0.3135,
+      "step": 11088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.351075611259583e-08,
+      "loss": 0.3495,
+      "step": 11089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.319393373807492e-08,
+      "loss": 0.3177,
+      "step": 11090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.287790106757396e-08,
+      "loss": 0.3587,
+      "step": 11091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.256265812621177e-08,
+      "loss": 0.3197,
+      "step": 11092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.224820493903937e-08,
+      "loss": 0.3196,
+      "step": 11093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.193454153104795e-08,
+      "loss": 0.3827,
+      "step": 11094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.162166792716529e-08,
+      "loss": 0.3735,
+      "step": 11095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.130958415225597e-08,
+      "loss": 0.3579,
+      "step": 11096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.099829023112236e-08,
+      "loss": 0.3008,
+      "step": 11097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.068778618850357e-08,
+      "loss": 0.3134,
+      "step": 11098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.037807204907653e-08,
+      "loss": 0.3467,
+      "step": 11099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.006914783745488e-08,
+      "loss": 0.345,
+      "step": 11100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.976101357818897e-08,
+      "loss": 0.35,
+      "step": 11101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9453669295769236e-08,
+      "loss": 0.3472,
+      "step": 11102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.914711501461723e-08,
+      "loss": 0.3203,
+      "step": 11103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.884135075910014e-08,
+      "loss": 0.3559,
+      "step": 11104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.853637655351407e-08,
+      "loss": 0.3642,
+      "step": 11105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8232192422098544e-08,
+      "loss": 0.3289,
+      "step": 11106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.792879838902643e-08,
+      "loss": 0.3724,
+      "step": 11107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7626194478410666e-08,
+      "loss": 0.3514,
+      "step": 11108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.732438071429869e-08,
+      "loss": 0.3726,
+      "step": 11109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.702335712067686e-08,
+      "loss": 0.3235,
+      "step": 11110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.672312372146827e-08,
+      "loss": 0.3113,
+      "step": 11111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.642368054053382e-08,
+      "loss": 0.3396,
+      "step": 11112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6125027601668936e-08,
+      "loss": 0.3848,
+      "step": 11113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.58271649286124e-08,
+      "loss": 0.3413,
+      "step": 11114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5530092545031944e-08,
+      "loss": 0.3622,
+      "step": 11115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5233810474538684e-08,
+      "loss": 0.3466,
+      "step": 11116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.493831874067934e-08,
+      "loss": 0.3268,
+      "step": 11117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4643617366936244e-08,
+      "loss": 0.3284,
+      "step": 11118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.434970637673176e-08,
+      "loss": 0.3326,
+      "step": 11119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4056585793421654e-08,
+      "loss": 0.3164,
+      "step": 11120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.376425564030174e-08,
+      "loss": 0.3115,
+      "step": 11121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.347271594060566e-08,
+      "loss": 0.3474,
+      "step": 11122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.318196671750042e-08,
+      "loss": 0.395,
+      "step": 11123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.289200799409422e-08,
+      "loss": 0.3757,
+      "step": 11124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.260283979343084e-08,
+      "loss": 0.3344,
+      "step": 11125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.231446213849078e-08,
+      "loss": 0.3388,
+      "step": 11126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.202687505219128e-08,
+      "loss": 0.3499,
+      "step": 11127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1740078557389605e-08,
+      "loss": 0.3414,
+      "step": 11128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.145407267687752e-08,
+      "loss": 0.3477,
+      "step": 11129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.116885743338351e-08,
+      "loss": 0.3388,
+      "step": 11130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.088443284957612e-08,
+      "loss": 0.3291,
+      "step": 11131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.060079894805725e-08,
+      "loss": 0.3269,
+      "step": 11132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0317955751369996e-08,
+      "loss": 0.3424,
+      "step": 11133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0035903281991923e-08,
+      "loss": 0.3376,
+      "step": 11134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.975464156233844e-08,
+      "loss": 0.3757,
+      "step": 11135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9474170614762787e-08,
+      "loss": 0.3314,
+      "step": 11136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.919449046155267e-08,
+      "loss": 0.3287,
+      "step": 11137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8915601124936984e-08,
+      "loss": 0.3037,
+      "step": 11138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8637502627080224e-08,
+      "loss": 0.3309,
+      "step": 11139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.836019499008138e-08,
+      "loss": 0.3345,
+      "step": 11140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.808367823598059e-08,
+      "loss": 0.3227,
+      "step": 11141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.780795238675251e-08,
+      "loss": 0.3501,
+      "step": 11142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7533017464310714e-08,
+      "loss": 0.3233,
+      "step": 11143
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.725887349050329e-08,
+      "loss": 0.3637,
+      "step": 11144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6985520487119463e-08,
+      "loss": 0.3626,
+      "step": 11145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6712958475881866e-08,
+      "loss": 0.3469,
+      "step": 11146
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6441187478452055e-08,
+      "loss": 0.3482,
+      "step": 11147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.617020751642831e-08,
+      "loss": 0.3528,
+      "step": 11148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.590001861134563e-08,
+      "loss": 0.3617,
+      "step": 11149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5630620784677947e-08,
+      "loss": 0.3207,
+      "step": 11150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.536201405783369e-08,
+      "loss": 0.3199,
+      "step": 11151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.509419845215912e-08,
+      "loss": 0.3833,
+      "step": 11152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.482717398894165e-08,
+      "loss": 0.3393,
+      "step": 11153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4560940689397646e-08,
+      "loss": 0.3525,
+      "step": 11154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4295498574690176e-08,
+      "loss": 0.3497,
+      "step": 11155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.403084766591126e-08,
+      "loss": 0.3558,
+      "step": 11156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3766987984094065e-08,
+      "loss": 0.3415,
+      "step": 11157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.350391955020961e-08,
+      "loss": 0.3242,
+      "step": 11158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.324164238516226e-08,
+      "loss": 0.3208,
+      "step": 11159
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.29801565097987e-08,
+      "loss": 0.3529,
+      "step": 11160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.271946194489784e-08,
+      "loss": 0.3531,
+      "step": 11161
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2459558711177573e-08,
+      "loss": 0.3029,
+      "step": 11162
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.220044682929469e-08,
+      "loss": 0.3384,
+      "step": 11163
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1942126319840514e-08,
+      "loss": 0.349,
+      "step": 11164
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.168459720334417e-08,
+      "loss": 0.352,
+      "step": 11165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.142785950027373e-08,
+      "loss": 0.3595,
+      "step": 11166
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.117191323103065e-08,
+      "loss": 0.3475,
+      "step": 11167
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.091675841595644e-08,
+      "loss": 0.3077,
+      "step": 11168
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.066239507533043e-08,
+      "loss": 0.3164,
+      "step": 11169
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0408823229365346e-08,
+      "loss": 0.3431,
+      "step": 11170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.015604289821395e-08,
+      "loss": 0.3341,
+      "step": 11171
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.990405410196574e-08,
+      "loss": 0.3459,
+      "step": 11172
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.96528568606469e-08,
+      "loss": 0.3278,
+      "step": 11173
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.940245119422148e-08,
+      "loss": 0.3098,
+      "step": 11174
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.915283712258799e-08,
+      "loss": 0.3579,
+      "step": 11175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.890401466558391e-08,
+      "loss": 0.3399,
+      "step": 11176
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.865598384298675e-08,
+      "loss": 0.3573,
+      "step": 11177
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8408744674505175e-08,
+      "loss": 0.3612,
+      "step": 11178
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.816229717978903e-08,
+      "loss": 0.316,
+      "step": 11179
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.791664137842377e-08,
+      "loss": 0.3885,
+      "step": 11180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7671777289932654e-08,
+      "loss": 0.3393,
+      "step": 11181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.742770493377568e-08,
+      "loss": 0.3412,
+      "step": 11182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7184424329350656e-08,
+      "loss": 0.3229,
+      "step": 11183
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.694193549599101e-08,
+      "loss": 0.3562,
+      "step": 11184
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6700238452966884e-08,
+      "loss": 0.3551,
+      "step": 11185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.645933321948736e-08,
+      "loss": 0.3384,
+      "step": 11186
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.621921981469933e-08,
+      "loss": 0.3562,
+      "step": 11187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.597989825768311e-08,
+      "loss": 0.3335,
+      "step": 11188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.574136856745902e-08,
+      "loss": 0.3235,
+      "step": 11189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.550363076298302e-08,
+      "loss": 0.3361,
+      "step": 11190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5266684863151103e-08,
+      "loss": 0.3071,
+      "step": 11191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5030530886791536e-08,
+      "loss": 0.3421,
+      "step": 11192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4795168852671535e-08,
+      "loss": 0.3314,
+      "step": 11193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4560598779499465e-08,
+      "loss": 0.3343,
+      "step": 11194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.432682068591375e-08,
+      "loss": 0.3015,
+      "step": 11195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.409383459049509e-08,
+      "loss": 0.3314,
+      "step": 11196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3861640511757554e-08,
+      "loss": 0.3623,
+      "step": 11197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.363023846815638e-08,
+      "loss": 0.3162,
+      "step": 11198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3399628478082424e-08,
+      "loss": 0.3402,
+      "step": 11199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3169810559859904e-08,
+      "loss": 0.3155,
+      "step": 11200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2940784731755324e-08,
+      "loss": 0.3107,
+      "step": 11201
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.271255101196857e-08,
+      "loss": 0.3728,
+      "step": 11202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.248510941863958e-08,
+      "loss": 0.3462,
+      "step": 11203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2258459969841674e-08,
+      "loss": 0.3023,
+      "step": 11204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.203260268358821e-08,
+      "loss": 0.3409,
+      "step": 11205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1807537577830396e-08,
+      "loss": 0.348,
+      "step": 11206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1583264670451695e-08,
+      "loss": 0.3723,
+      "step": 11207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1359783979277857e-08,
+      "loss": 0.3333,
+      "step": 11208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1137095522068006e-08,
+      "loss": 0.3375,
+      "step": 11209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.091519931652131e-08,
+      "loss": 0.3494,
+      "step": 11210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.069409538027146e-08,
+      "loss": 0.3071,
+      "step": 11211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.047378373088994e-08,
+      "loss": 0.3107,
+      "step": 11212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0254264385887186e-08,
+      "loss": 0.3803,
+      "step": 11213
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0035537362705927e-08,
+      "loss": 0.3527,
+      "step": 11214
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.981760267873224e-08,
+      "loss": 0.3116,
+      "step": 11215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9600460351283388e-08,
+      "loss": 0.3618,
+      "step": 11216
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9384110397617792e-08,
+      "loss": 0.3327,
+      "step": 11217
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9168552834929476e-08,
+      "loss": 0.3244,
+      "step": 11218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8953787680348067e-08,
+      "loss": 0.359,
+      "step": 11219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8739814950943245e-08,
+      "loss": 0.3561,
+      "step": 11220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.852663466371697e-08,
+      "loss": 0.3439,
+      "step": 11221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8314246835614566e-08,
+      "loss": 0.3387,
+      "step": 11222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8102651483513655e-08,
+      "loss": 0.357,
+      "step": 11223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7891848624230778e-08,
+      "loss": 0.349,
+      "step": 11224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.768183827451809e-08,
+      "loss": 0.3231,
+      "step": 11225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.747262045106558e-08,
+      "loss": 0.3357,
+      "step": 11226
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.726419517050105e-08,
+      "loss": 0.3116,
+      "step": 11227
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7056562449389033e-08,
+      "loss": 0.3266,
+      "step": 11228
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6849722304229664e-08,
+      "loss": 0.301,
+      "step": 11229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6643674751462014e-08,
+      "loss": 0.3506,
+      "step": 11230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6438419807459647e-08,
+      "loss": 0.3611,
+      "step": 11231
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6233957488537298e-08,
+      "loss": 0.3458,
+      "step": 11232
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6030287810940858e-08,
+      "loss": 0.3522,
+      "step": 11233
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.582741079085849e-08,
+      "loss": 0.3123,
+      "step": 11234
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5625326444413955e-08,
+      "loss": 0.2981,
+      "step": 11235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.542403478766553e-08,
+      "loss": 0.3375,
+      "step": 11236
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5223535836612634e-08,
+      "loss": 0.3489,
+      "step": 11237
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5023829607186967e-08,
+      "loss": 0.3066,
+      "step": 11238
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4824916115261387e-08,
+      "loss": 0.339,
+      "step": 11239
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4626795376644364e-08,
+      "loss": 0.3089,
+      "step": 11240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4429467407078856e-08,
+      "loss": 0.3519,
+      "step": 11241
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4232932222248982e-08,
+      "loss": 0.35,
+      "step": 11242
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4037189837774476e-08,
+      "loss": 0.3266,
+      "step": 11243
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.384224026920956e-08,
+      "loss": 0.3488,
+      "step": 11244
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3648083532048504e-08,
+      "loss": 0.3882,
+      "step": 11245
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.345471964172119e-08,
+      "loss": 0.3202,
+      "step": 11246
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3262148613594215e-08,
+      "loss": 0.3339,
+      "step": 11247
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3070370462973114e-08,
+      "loss": 0.3041,
+      "step": 11248
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2879385205097914e-08,
+      "loss": 0.3294,
+      "step": 11249
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.268919285514759e-08,
+      "loss": 0.3164,
+      "step": 11250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2499793428235605e-08,
+      "loss": 0.3172,
+      "step": 11251
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2311186939415476e-08,
+      "loss": 0.3311,
+      "step": 11252
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.212337340367521e-08,
+      "loss": 0.3593,
+      "step": 11253
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1936352835941754e-08,
+      "loss": 0.3389,
+      "step": 11254
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.175012525107878e-08,
+      "loss": 0.3441,
+      "step": 11255
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1564690663884448e-08,
+      "loss": 0.3188,
+      "step": 11256
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1380049089096966e-08,
+      "loss": 0.3209,
+      "step": 11257
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1196200541389043e-08,
+      "loss": 0.3471,
+      "step": 11258
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.101314503537344e-08,
+      "loss": 0.3848,
+      "step": 11259
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.083088258559629e-08,
+      "loss": 0.3292,
+      "step": 11260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.064941320654379e-08,
+      "loss": 0.3406,
+      "step": 11261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.046873691263773e-08,
+      "loss": 0.3135,
+      "step": 11262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0288853718235523e-08,
+      "loss": 0.2999,
+      "step": 11263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0109763637634617e-08,
+      "loss": 0.3509,
+      "step": 11264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.993146668506585e-08,
+      "loss": 0.3443,
+      "step": 11265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9753962874702325e-08,
+      "loss": 0.3019,
+      "step": 11266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9577252220646102e-08,
+      "loss": 0.3443,
+      "step": 11267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9401334736944833e-08,
+      "loss": 0.3323,
+      "step": 11268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.922621043757733e-08,
+      "loss": 0.3232,
+      "step": 11269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9051879336461354e-08,
+      "loss": 0.3327,
+      "step": 11270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8878341447451376e-08,
+      "loss": 0.3194,
+      "step": 11271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8705596784339697e-08,
+      "loss": 0.3189,
+      "step": 11272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8533645360853114e-08,
+      "loss": 0.3475,
+      "step": 11273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8362487190659585e-08,
+      "loss": 0.3397,
+      "step": 11274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8192122287358228e-08,
+      "loss": 0.3581,
+      "step": 11275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8022550664491544e-08,
+      "loss": 0.323,
+      "step": 11276
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7853772335534313e-08,
+      "loss": 0.302,
+      "step": 11277
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7685787313900248e-08,
+      "loss": 0.3626,
+      "step": 11278
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7518595612938673e-08,
+      "loss": 0.3545,
+      "step": 11279
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.735219724593673e-08,
+      "loss": 0.3454,
+      "step": 11280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7186592226120514e-08,
+      "loss": 0.3003,
+      "step": 11281
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7021780566649492e-08,
+      "loss": 0.3565,
+      "step": 11282
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.685776228062097e-08,
+      "loss": 0.3153,
+      "step": 11283
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6694537381071186e-08,
+      "loss": 0.3421,
+      "step": 11284
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.653210588097198e-08,
+      "loss": 0.353,
+      "step": 11285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6370467793233034e-08,
+      "loss": 0.3294,
+      "step": 11286
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6209623130697405e-08,
+      "loss": 0.3261,
+      "step": 11287
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6049571906150418e-08,
+      "loss": 0.3446,
+      "step": 11288
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.589031413231079e-08,
+      "loss": 0.3219,
+      "step": 11289
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5731849821833955e-08,
+      "loss": 0.3766,
+      "step": 11290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.55741789873165e-08,
+      "loss": 0.3465,
+      "step": 11291
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5417301641286188e-08,
+      "loss": 0.3394,
+      "step": 11292
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5261217796211923e-08,
+      "loss": 0.3683,
+      "step": 11293
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5105927464497128e-08,
+      "loss": 0.3184,
+      "step": 11294
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4951430658484145e-08,
+      "loss": 0.3628,
+      "step": 11295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4797727390450933e-08,
+      "loss": 0.3315,
+      "step": 11296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4644817672612165e-08,
+      "loss": 0.3489,
+      "step": 11297
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4492701517120344e-08,
+      "loss": 0.3354,
+      "step": 11298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4341378936063577e-08,
+      "loss": 0.3161,
+      "step": 11299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.419084994147002e-08,
+      "loss": 0.3359,
+      "step": 11300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4041114545301216e-08,
+      "loss": 0.3528,
+      "step": 11301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3892172759457645e-08,
+      "loss": 0.3738,
+      "step": 11302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3744024595774285e-08,
+      "loss": 0.3441,
+      "step": 11303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3596670066027274e-08,
+      "loss": 0.3229,
+      "step": 11304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.345010918192613e-08,
+      "loss": 0.3584,
+      "step": 11305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3304341955118205e-08,
+      "loss": 0.2974,
+      "step": 11306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3159368397188677e-08,
+      "loss": 0.326,
+      "step": 11307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3015188519658328e-08,
+      "loss": 0.3113,
+      "step": 11308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2871802333985772e-08,
+      "loss": 0.3578,
+      "step": 11309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2729209851566337e-08,
+      "loss": 0.3614,
+      "step": 11310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2587411083732071e-08,
+      "loss": 0.3241,
+      "step": 11311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2446406041752846e-08,
+      "loss": 0.3639,
+      "step": 11312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2306194736833033e-08,
+      "loss": 0.3359,
+      "step": 11313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.216677718011816e-08,
+      "loss": 0.3243,
+      "step": 11314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2028153382684926e-08,
+      "loss": 0.3311,
+      "step": 11315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1890323355552292e-08,
+      "loss": 0.3222,
+      "step": 11316
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1753287109673717e-08,
+      "loss": 0.3657,
+      "step": 11317
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1617044655940491e-08,
+      "loss": 0.355,
+      "step": 11318
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1481596005177287e-08,
+      "loss": 0.3639,
+      "step": 11319
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1346941168152158e-08,
+      "loss": 0.3691,
+      "step": 11320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1213080155564327e-08,
+      "loss": 0.3548,
+      "step": 11321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1080012978051947e-08,
+      "loss": 0.3164,
+      "step": 11322
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0947739646192112e-08,
+      "loss": 0.3416,
+      "step": 11323
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0816260170494197e-08,
+      "loss": 0.3441,
+      "step": 11324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.068557456140984e-08,
+      "loss": 0.3239,
+      "step": 11325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0555682829322956e-08,
+      "loss": 0.3715,
+      "step": 11326
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0426584984558619e-08,
+      "loss": 0.3217,
+      "step": 11327
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.029828103737418e-08,
+      "loss": 0.3596,
+      "step": 11328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0170770997967039e-08,
+      "loss": 0.3429,
+      "step": 11329
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.004405487647131e-08,
+      "loss": 0.3325,
+      "step": 11330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.918132682957827e-09,
+      "loss": 0.3374,
+      "step": 11331
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.793004427433028e-09,
+      "loss": 0.3372,
+      "step": 11332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.668670119841184e-09,
+      "loss": 0.3659,
+      "step": 11333
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.545129770063277e-09,
+      "loss": 0.3809,
+      "step": 11334
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.422383387919232e-09,
+      "loss": 0.3394,
+      "step": 11335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.300430983161245e-09,
+      "loss": 0.3317,
+      "step": 11336
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.179272565482678e-09,
+      "loss": 0.3431,
+      "step": 11337
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.0589081445136e-09,
+      "loss": 0.3618,
+      "step": 11338
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.939337729817477e-09,
+      "loss": 0.3368,
+      "step": 11339
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.820561330897814e-09,
+      "loss": 0.3503,
+      "step": 11340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.702578957193731e-09,
+      "loss": 0.3737,
+      "step": 11341
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.585390618081057e-09,
+      "loss": 0.3346,
+      "step": 11342
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.468996322875678e-09,
+      "loss": 0.3118,
+      "step": 11343
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.353396080824639e-09,
+      "loss": 0.3488,
+      "step": 11344
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.238589901116145e-09,
+      "loss": 0.3623,
+      "step": 11345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.12457779287512e-09,
+      "loss": 0.3152,
+      "step": 11346
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.011359765160986e-09,
+      "loss": 0.3402,
+      "step": 11347
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.898935826973208e-09,
+      "loss": 0.3712,
+      "step": 11348
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.787305987243532e-09,
+      "loss": 0.3544,
+      "step": 11349
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.676470254845969e-09,
+      "loss": 0.3857,
+      "step": 11350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.56642863858903e-09,
+      "loss": 0.341,
+      "step": 11351
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.45718114721572e-09,
+      "loss": 0.3701,
+      "step": 11352
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.348727789411314e-09,
+      "loss": 0.3557,
+      "step": 11353
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.2410685737911434e-09,
+      "loss": 0.3585,
+      "step": 11354
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.1342035089150275e-09,
+      "loss": 0.3676,
+      "step": 11355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.028132603272841e-09,
+      "loss": 0.3214,
+      "step": 11356
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.922855865295619e-09,
+      "loss": 0.3052,
+      "step": 11357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.8183733033511114e-09,
+      "loss": 0.3301,
+      "step": 11358
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.714684925740456e-09,
+      "loss": 0.3281,
+      "step": 11359
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.611790740704838e-09,
+      "loss": 0.3058,
+      "step": 11360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.50969075642216e-09,
+      "loss": 0.319,
+      "step": 11361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.408384981007043e-09,
+      "loss": 0.354,
+      "step": 11362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.307873422508604e-09,
+      "loss": 0.3102,
+      "step": 11363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.208156088917117e-09,
+      "loss": 0.2974,
+      "step": 11364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.109232988155134e-09,
+      "loss": 0.3321,
+      "step": 11365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.011104128085254e-09,
+      "loss": 0.3283,
+      "step": 11366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9137695165067955e-09,
+      "loss": 0.3351,
+      "step": 11367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.817229161154681e-09,
+      "loss": 0.354,
+      "step": 11368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.721483069700551e-09,
+      "loss": 0.3535,
+      "step": 11369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.626531249753875e-09,
+      "loss": 0.3614,
+      "step": 11370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5323737088619486e-09,
+      "loss": 0.3496,
+      "step": 11371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.439010454505456e-09,
+      "loss": 0.3439,
+      "step": 11372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.346441494106236e-09,
+      "loss": 0.3389,
+      "step": 11373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.254666835020628e-09,
+      "loss": 0.325,
+      "step": 11374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.163686484541686e-09,
+      "loss": 0.3434,
+      "step": 11375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0735004498991824e-09,
+      "loss": 0.3084,
+      "step": 11376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.984108738261828e-09,
+      "loss": 0.3653,
+      "step": 11377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8955113567339394e-09,
+      "loss": 0.3812,
+      "step": 11378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.807708312355442e-09,
+      "loss": 0.3457,
+      "step": 11379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.720699612104085e-09,
+      "loss": 0.3139,
+      "step": 11380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.634485262895449e-09,
+      "loss": 0.2929,
+      "step": 11381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.549065271581832e-09,
+      "loss": 0.3228,
+      "step": 11382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.464439644951135e-09,
+      "loss": 0.3399,
+      "step": 11383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.380608389727981e-09,
+      "loss": 0.326,
+      "step": 11384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.297571512574816e-09,
+      "loss": 0.3407,
+      "step": 11385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.215329020093028e-09,
+      "loss": 0.3331,
+      "step": 11386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1338809188162795e-09,
+      "loss": 0.315,
+      "step": 11387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.053227215217171e-09,
+      "loss": 0.2944,
+      "step": 11388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.973367915707238e-09,
+      "loss": 0.3228,
+      "step": 11389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.894303026632518e-09,
+      "loss": 0.3493,
+      "step": 11390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.816032554275762e-09,
+      "loss": 0.3762,
+      "step": 11391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.738556504857549e-09,
+      "loss": 0.3774,
+      "step": 11392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.661874884536287e-09,
+      "loss": 0.3219,
+      "step": 11393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5859876994048804e-09,
+      "loss": 0.3246,
+      "step": 11394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.51089495549517e-09,
+      "loss": 0.3538,
+      "step": 11395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4365966587734945e-09,
+      "loss": 0.3009,
+      "step": 11396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.36309281514513e-09,
+      "loss": 0.3362,
+      "step": 11397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2903834304531814e-09,
+      "loss": 0.3202,
+      "step": 11398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2184685104741376e-09,
+      "loss": 0.3599,
+      "step": 11399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1473480609234276e-09,
+      "loss": 0.3554,
+      "step": 11400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0770220874543068e-09,
+      "loss": 0.3588,
+      "step": 11401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0074905956556376e-09,
+      "loss": 0.3226,
+      "step": 11402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9387535910518905e-09,
+      "loss": 0.3106,
+      "step": 11403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8708110791075827e-09,
+      "loss": 0.3792,
+      "step": 11404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8036630652206187e-09,
+      "loss": 0.3378,
+      "step": 11405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7373095547278407e-09,
+      "loss": 0.3209,
+      "step": 11406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6717505529028077e-09,
+      "loss": 0.3239,
+      "step": 11407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.606986064956907e-09,
+      "loss": 0.342,
+      "step": 11408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5430160960349116e-09,
+      "loss": 0.3672,
+      "step": 11409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4798406512227535e-09,
+      "loss": 0.3588,
+      "step": 11410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4174597355386408e-09,
+      "loss": 0.3055,
+      "step": 11411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.35587335394305e-09,
+      "loss": 0.3562,
+      "step": 11412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.295081511327624e-09,
+      "loss": 0.2943,
+      "step": 11413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.235084212526273e-09,
+      "loss": 0.3213,
+      "step": 11414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.175881462305185e-09,
+      "loss": 0.3484,
+      "step": 11415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1174732653705952e-09,
+      "loss": 0.3447,
+      "step": 11416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0598596263632365e-09,
+      "loss": 0.3482,
+      "step": 11417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003040549862778e-09,
+      "loss": 0.3387,
+      "step": 11418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9470160403844974e-09,
+      "loss": 0.3137,
+      "step": 11419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8917861023803885e-09,
+      "loss": 0.3402,
+      "step": 11420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8373507402402735e-09,
+      "loss": 0.3153,
+      "step": 11421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7837099582895812e-09,
+      "loss": 0.3526,
+      "step": 11422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.730863760791568e-09,
+      "loss": 0.3315,
+      "step": 11423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6788121519473178e-09,
+      "loss": 0.3411,
+      "step": 11424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.627555135891301e-09,
+      "loss": 0.3422,
+      "step": 11425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5770927166991468e-09,
+      "loss": 0.3142,
+      "step": 11426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.52742489837987e-09,
+      "loss": 0.3202,
+      "step": 11427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.478551684881424e-09,
+      "loss": 0.3143,
+      "step": 11428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4304730800862588e-09,
+      "loss": 0.3352,
+      "step": 11429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.383189087817982e-09,
+      "loss": 0.3764,
+      "step": 11430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3366997118324787e-09,
+      "loss": 0.3067,
+      "step": 11431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.291004955825681e-09,
+      "loss": 0.3894,
+      "step": 11432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.246104823426908e-09,
+      "loss": 0.3408,
+      "step": 11433
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.201999318206637e-09,
+      "loss": 0.3212,
+      "step": 11434
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.158688443669842e-09,
+      "loss": 0.3788,
+      "step": 11435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1161722032571043e-09,
+      "loss": 0.3796,
+      "step": 11436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0744506003479427e-09,
+      "loss": 0.3442,
+      "step": 11437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0335236382597036e-09,
+      "loss": 0.3749,
+      "step": 11438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.9339132024312e-10,
+      "loss": 0.3287,
+      "step": 11439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.540536494867526e-10,
+      "loss": 0.3214,
+      "step": 11440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.155106291192095e-10,
+      "loss": 0.3659,
+      "step": 11441
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.777622622035963e-10,
+      "loss": 0.3489,
+      "step": 11442
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.408085517375153e-10,
+      "loss": 0.2979,
+      "step": 11443
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.046495006597265e-10,
+      "loss": 0.3229,
+      "step": 11444
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.692851118423772e-10,
+      "loss": 0.3196,
+      "step": 11445
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.347153880976621e-10,
+      "loss": 0.3208,
+      "step": 11446
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.009403321722729e-10,
+      "loss": 0.3401,
+      "step": 11447
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.679599467507292e-10,
+      "loss": 0.3264,
+      "step": 11448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.35774234453157e-10,
+      "loss": 0.3185,
+      "step": 11449
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.043831978375103e-10,
+      "loss": 0.3598,
+      "step": 11450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.737868393984602e-10,
+      "loss": 0.3482,
+      "step": 11451
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.439851615696157e-10,
+      "loss": 0.324,
+      "step": 11452
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.149781667168619e-10,
+      "loss": 0.3496,
+      "step": 11453
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.86765857146132e-10,
+      "loss": 0.3349,
+      "step": 11454
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5934823510007667e-10,
+      "loss": 0.3632,
+      "step": 11455
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3272530275695334e-10,
+      "loss": 0.3029,
+      "step": 11456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.068970622339574e-10,
+      "loss": 0.3817,
+      "step": 11457
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.818635155816708e-10,
+      "loss": 0.3813,
+      "step": 11458
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5762466479183354e-10,
+      "loss": 0.3314,
+      "step": 11459
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341805117884622e-10,
+      "loss": 0.3431,
+      "step": 11460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.115310584367315e-10,
+      "loss": 0.3571,
+      "step": 11461
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.896763065352026e-10,
+      "loss": 0.323,
+      "step": 11462
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.686162578213747e-10,
+      "loss": 0.3302,
+      "step": 11463
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.48350913969464e-10,
+      "loss": 0.3209,
+      "step": 11464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.288802765892939e-10,
+      "loss": 0.3599,
+      "step": 11465
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1020434722740512e-10,
+      "loss": 0.3355,
+      "step": 11466
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9232312737038627e-10,
+      "loss": 0.3237,
+      "step": 11467
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7523661843710238e-10,
+      "loss": 0.3445,
+      "step": 11468
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5894482178646643e-10,
+      "loss": 0.3513,
+      "step": 11469
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.434477387118882e-10,
+      "loss": 0.3439,
+      "step": 11470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2874537044793577e-10,
+      "loss": 0.3339,
+      "step": 11471
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1483771816034328e-10,
+      "loss": 0.3253,
+      "step": 11472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0172478295489286e-10,
+      "loss": 0.356,
+      "step": 11473
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.940656587519414e-11,
+      "loss": 0.3536,
+      "step": 11474
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.78830678982434e-11,
+      "loss": 0.3554,
+      "step": 11475
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.715428993997463e-11,
+      "loss": 0.4012,
+      "step": 11476
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.722023285525957e-11,
+      "loss": 0.3268,
+      "step": 11477
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.808089743124633e-11,
+      "loss": 0.324,
+      "step": 11478
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9736284395131e-11,
+      "loss": 0.3138,
+      "step": 11479
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2186394409716716e-11,
+      "loss": 0.365,
+      "step": 11480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.543122807563414e-11,
+      "loss": 0.3483,
+      "step": 11481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9470785930231218e-11,
+      "loss": 0.4063,
+      "step": 11482
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4305068445352733e-11,
+      "loss": 0.3629,
+      "step": 11483
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.934076032891427e-12,
+      "loss": 0.3295,
+      "step": 11484
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.357809040347107e-12,
+      "loss": 0.3493,
+      "step": 11485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5762677508266454e-12,
+      "loss": 0.3197,
+      "step": 11486
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5894523863746458e-12,
+      "loss": 0.3561,
+      "step": 11487
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.973631046427784e-13,
+      "loss": 0.3428,
+      "step": 11488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.3468,
+      "step": 11489
+    },
+    {
+      "epoch": 1.0,
+      "step": 11489,
+      "total_flos": 4.704993542275072e+16,
+      "train_loss": 0.37806004422256123,
+      "train_runtime": 218905.9407,
+      "train_samples_per_second": 6.718,
+      "train_steps_per_second": 0.052
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 11489,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 4.704993542275072e+16,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}